Model save
Browse files- README.md +58 -0
- all_results.json +8 -0
- generation_config.json +6 -0
- train_results.json +8 -0
- trainer_state.json +2539 -0
README.md
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: allenai/OLMo-2-1124-7B-Instruct
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: OLMo-2-1124-7B-Instruct_SFTv02.10_full
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- sft
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for OLMo-2-1124-7B-Instruct_SFTv02.10_full
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [allenai/OLMo-2-1124-7B-Instruct](https://huggingface.co/allenai/OLMo-2-1124-7B-Instruct).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="Neelectric/OLMo-2-1124-7B-Instruct_SFTv02.10_full", device="cuda")
|
| 24 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 25 |
+
print(output["generated_text"])
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Training procedure
|
| 29 |
+
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/neelectric/open-r1_SFT/runs/7qb8x24u)
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
This model was trained with SFT.
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.17.0
|
| 38 |
+
- Transformers: 4.51.3
|
| 39 |
+
- Pytorch: 2.6.0
|
| 40 |
+
- Datasets: 3.6.0
|
| 41 |
+
- Tokenizers: 0.21.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
Cite TRL as:
|
| 48 |
+
|
| 49 |
+
```bibtex
|
| 50 |
+
@misc{vonwerra2022trl,
|
| 51 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 52 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
|
| 53 |
+
year = 2020,
|
| 54 |
+
journal = {GitHub repository},
|
| 55 |
+
publisher = {GitHub},
|
| 56 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 57 |
+
}
|
| 58 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 3.993677328755458e+19,
|
| 3 |
+
"train_loss": 0.4935996499963296,
|
| 4 |
+
"train_runtime": 22258.2821,
|
| 5 |
+
"train_samples": 93733,
|
| 6 |
+
"train_samples_per_second": 12.633,
|
| 7 |
+
"train_steps_per_second": 0.014
|
| 8 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"eos_token_id": 100257,
|
| 4 |
+
"pad_token_id": 100277,
|
| 5 |
+
"transformers_version": "4.51.3"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 3.993677328755458e+19,
|
| 3 |
+
"train_loss": 0.4935996499963296,
|
| 4 |
+
"train_runtime": 22258.2821,
|
| 5 |
+
"train_samples": 93733,
|
| 6 |
+
"train_samples_per_second": 12.633,
|
| 7 |
+
"train_steps_per_second": 0.014
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2539 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.974910394265233,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 312,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.009557945041816009,
|
| 14 |
+
"grad_norm": 4.787891864776611,
|
| 15 |
+
"learning_rate": 0.0,
|
| 16 |
+
"loss": 1.2182,
|
| 17 |
+
"num_tokens": 3087662.0,
|
| 18 |
+
"step": 1
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 0.019115890083632018,
|
| 22 |
+
"grad_norm": 4.7660908699035645,
|
| 23 |
+
"learning_rate": 1.5625e-06,
|
| 24 |
+
"loss": 1.2133,
|
| 25 |
+
"num_tokens": 6211852.0,
|
| 26 |
+
"step": 2
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.02867383512544803,
|
| 30 |
+
"grad_norm": 4.726694107055664,
|
| 31 |
+
"learning_rate": 3.125e-06,
|
| 32 |
+
"loss": 1.2106,
|
| 33 |
+
"num_tokens": 9308149.0,
|
| 34 |
+
"step": 3
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.038231780167264036,
|
| 38 |
+
"grad_norm": 4.008660793304443,
|
| 39 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 40 |
+
"loss": 1.1473,
|
| 41 |
+
"num_tokens": 12406212.0,
|
| 42 |
+
"step": 4
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.04778972520908005,
|
| 46 |
+
"grad_norm": 2.9268953800201416,
|
| 47 |
+
"learning_rate": 6.25e-06,
|
| 48 |
+
"loss": 1.0694,
|
| 49 |
+
"num_tokens": 15498886.0,
|
| 50 |
+
"step": 5
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.05734767025089606,
|
| 54 |
+
"grad_norm": 1.6218466758728027,
|
| 55 |
+
"learning_rate": 7.8125e-06,
|
| 56 |
+
"loss": 0.9229,
|
| 57 |
+
"num_tokens": 18594568.0,
|
| 58 |
+
"step": 6
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.06690561529271206,
|
| 62 |
+
"grad_norm": 1.232273817062378,
|
| 63 |
+
"learning_rate": 9.375000000000001e-06,
|
| 64 |
+
"loss": 0.8977,
|
| 65 |
+
"num_tokens": 21685747.0,
|
| 66 |
+
"step": 7
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.07646356033452807,
|
| 70 |
+
"grad_norm": 1.352134346961975,
|
| 71 |
+
"learning_rate": 1.09375e-05,
|
| 72 |
+
"loss": 0.8468,
|
| 73 |
+
"num_tokens": 24802133.0,
|
| 74 |
+
"step": 8
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.08602150537634409,
|
| 78 |
+
"grad_norm": 1.1753616333007812,
|
| 79 |
+
"learning_rate": 1.25e-05,
|
| 80 |
+
"loss": 0.8356,
|
| 81 |
+
"num_tokens": 27921758.0,
|
| 82 |
+
"step": 9
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.0955794504181601,
|
| 86 |
+
"grad_norm": 0.841299831867218,
|
| 87 |
+
"learning_rate": 1.4062500000000001e-05,
|
| 88 |
+
"loss": 0.8045,
|
| 89 |
+
"num_tokens": 30982499.0,
|
| 90 |
+
"step": 10
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.10513739545997611,
|
| 94 |
+
"grad_norm": 0.7823367118835449,
|
| 95 |
+
"learning_rate": 1.5625e-05,
|
| 96 |
+
"loss": 0.7772,
|
| 97 |
+
"num_tokens": 34120523.0,
|
| 98 |
+
"step": 11
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.11469534050179211,
|
| 102 |
+
"grad_norm": 0.6168179512023926,
|
| 103 |
+
"learning_rate": 1.71875e-05,
|
| 104 |
+
"loss": 0.7493,
|
| 105 |
+
"num_tokens": 37263940.0,
|
| 106 |
+
"step": 12
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.12425328554360812,
|
| 110 |
+
"grad_norm": 0.4352397918701172,
|
| 111 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 112 |
+
"loss": 0.7391,
|
| 113 |
+
"num_tokens": 40419105.0,
|
| 114 |
+
"step": 13
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.13381123058542413,
|
| 118 |
+
"grad_norm": 0.4141225516796112,
|
| 119 |
+
"learning_rate": 2.0312500000000002e-05,
|
| 120 |
+
"loss": 0.7168,
|
| 121 |
+
"num_tokens": 43544209.0,
|
| 122 |
+
"step": 14
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.14336917562724014,
|
| 126 |
+
"grad_norm": 0.3596664071083069,
|
| 127 |
+
"learning_rate": 2.1875e-05,
|
| 128 |
+
"loss": 0.7096,
|
| 129 |
+
"num_tokens": 46640184.0,
|
| 130 |
+
"step": 15
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.15292712066905614,
|
| 134 |
+
"grad_norm": 0.32062292098999023,
|
| 135 |
+
"learning_rate": 2.34375e-05,
|
| 136 |
+
"loss": 0.6878,
|
| 137 |
+
"num_tokens": 49723879.0,
|
| 138 |
+
"step": 16
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.16248506571087215,
|
| 142 |
+
"grad_norm": 0.2776451110839844,
|
| 143 |
+
"learning_rate": 2.5e-05,
|
| 144 |
+
"loss": 0.685,
|
| 145 |
+
"num_tokens": 52840221.0,
|
| 146 |
+
"step": 17
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.17204301075268819,
|
| 150 |
+
"grad_norm": 0.2792057693004608,
|
| 151 |
+
"learning_rate": 2.6562500000000002e-05,
|
| 152 |
+
"loss": 0.6683,
|
| 153 |
+
"num_tokens": 55938115.0,
|
| 154 |
+
"step": 18
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.1816009557945042,
|
| 158 |
+
"grad_norm": 0.24952948093414307,
|
| 159 |
+
"learning_rate": 2.8125000000000003e-05,
|
| 160 |
+
"loss": 0.6643,
|
| 161 |
+
"num_tokens": 59088421.0,
|
| 162 |
+
"step": 19
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 0.1911589008363202,
|
| 166 |
+
"grad_norm": 0.23966556787490845,
|
| 167 |
+
"learning_rate": 2.96875e-05,
|
| 168 |
+
"loss": 0.6568,
|
| 169 |
+
"num_tokens": 62120009.0,
|
| 170 |
+
"step": 20
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.2007168458781362,
|
| 174 |
+
"grad_norm": 0.29473549127578735,
|
| 175 |
+
"learning_rate": 3.125e-05,
|
| 176 |
+
"loss": 0.6443,
|
| 177 |
+
"num_tokens": 65202270.0,
|
| 178 |
+
"step": 21
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.21027479091995221,
|
| 182 |
+
"grad_norm": 0.25088849663734436,
|
| 183 |
+
"learning_rate": 3.2812500000000005e-05,
|
| 184 |
+
"loss": 0.64,
|
| 185 |
+
"num_tokens": 68278086.0,
|
| 186 |
+
"step": 22
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.21983273596176822,
|
| 190 |
+
"grad_norm": 0.26761022210121155,
|
| 191 |
+
"learning_rate": 3.4375e-05,
|
| 192 |
+
"loss": 0.6406,
|
| 193 |
+
"num_tokens": 71423006.0,
|
| 194 |
+
"step": 23
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.22939068100358423,
|
| 198 |
+
"grad_norm": 0.2646687626838684,
|
| 199 |
+
"learning_rate": 3.59375e-05,
|
| 200 |
+
"loss": 0.6287,
|
| 201 |
+
"num_tokens": 74542778.0,
|
| 202 |
+
"step": 24
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.23894862604540024,
|
| 206 |
+
"grad_norm": 0.28765901923179626,
|
| 207 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 208 |
+
"loss": 0.6316,
|
| 209 |
+
"num_tokens": 77676355.0,
|
| 210 |
+
"step": 25
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 0.24850657108721624,
|
| 214 |
+
"grad_norm": 0.28572410345077515,
|
| 215 |
+
"learning_rate": 3.90625e-05,
|
| 216 |
+
"loss": 0.6208,
|
| 217 |
+
"num_tokens": 80757397.0,
|
| 218 |
+
"step": 26
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.25806451612903225,
|
| 222 |
+
"grad_norm": 0.3025137782096863,
|
| 223 |
+
"learning_rate": 4.0625000000000005e-05,
|
| 224 |
+
"loss": 0.6206,
|
| 225 |
+
"num_tokens": 83848581.0,
|
| 226 |
+
"step": 27
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.26762246117084826,
|
| 230 |
+
"grad_norm": 0.30501461029052734,
|
| 231 |
+
"learning_rate": 4.21875e-05,
|
| 232 |
+
"loss": 0.61,
|
| 233 |
+
"num_tokens": 86949103.0,
|
| 234 |
+
"step": 28
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.27718040621266427,
|
| 238 |
+
"grad_norm": 0.24303048849105835,
|
| 239 |
+
"learning_rate": 4.375e-05,
|
| 240 |
+
"loss": 0.6062,
|
| 241 |
+
"num_tokens": 90043461.0,
|
| 242 |
+
"step": 29
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.2867383512544803,
|
| 246 |
+
"grad_norm": 0.4097803235054016,
|
| 247 |
+
"learning_rate": 4.5312500000000004e-05,
|
| 248 |
+
"loss": 0.608,
|
| 249 |
+
"num_tokens": 93175833.0,
|
| 250 |
+
"step": 30
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.2962962962962963,
|
| 254 |
+
"grad_norm": 0.22506467998027802,
|
| 255 |
+
"learning_rate": 4.6875e-05,
|
| 256 |
+
"loss": 0.6043,
|
| 257 |
+
"num_tokens": 96298838.0,
|
| 258 |
+
"step": 31
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.3058542413381123,
|
| 262 |
+
"grad_norm": 0.4554276168346405,
|
| 263 |
+
"learning_rate": 4.8437500000000005e-05,
|
| 264 |
+
"loss": 0.6039,
|
| 265 |
+
"num_tokens": 99387594.0,
|
| 266 |
+
"step": 32
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.3154121863799283,
|
| 270 |
+
"grad_norm": 0.24191643297672272,
|
| 271 |
+
"learning_rate": 5e-05,
|
| 272 |
+
"loss": 0.5919,
|
| 273 |
+
"num_tokens": 102495943.0,
|
| 274 |
+
"step": 33
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.3249701314217443,
|
| 278 |
+
"grad_norm": 0.497525155544281,
|
| 279 |
+
"learning_rate": 5e-05,
|
| 280 |
+
"loss": 0.5919,
|
| 281 |
+
"num_tokens": 105605746.0,
|
| 282 |
+
"step": 34
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.3345280764635603,
|
| 286 |
+
"grad_norm": 0.39946335554122925,
|
| 287 |
+
"learning_rate": 5e-05,
|
| 288 |
+
"loss": 0.5966,
|
| 289 |
+
"num_tokens": 108704371.0,
|
| 290 |
+
"step": 35
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.34408602150537637,
|
| 294 |
+
"grad_norm": 0.3315200209617615,
|
| 295 |
+
"learning_rate": 5e-05,
|
| 296 |
+
"loss": 0.58,
|
| 297 |
+
"num_tokens": 111803239.0,
|
| 298 |
+
"step": 36
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.3536439665471924,
|
| 302 |
+
"grad_norm": 0.38535887002944946,
|
| 303 |
+
"learning_rate": 5e-05,
|
| 304 |
+
"loss": 0.5836,
|
| 305 |
+
"num_tokens": 114901473.0,
|
| 306 |
+
"step": 37
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.3632019115890084,
|
| 310 |
+
"grad_norm": 0.30798089504241943,
|
| 311 |
+
"learning_rate": 5e-05,
|
| 312 |
+
"loss": 0.5821,
|
| 313 |
+
"num_tokens": 117998143.0,
|
| 314 |
+
"step": 38
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.3727598566308244,
|
| 318 |
+
"grad_norm": 0.3112103343009949,
|
| 319 |
+
"learning_rate": 5e-05,
|
| 320 |
+
"loss": 0.5789,
|
| 321 |
+
"num_tokens": 121151007.0,
|
| 322 |
+
"step": 39
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.3823178016726404,
|
| 326 |
+
"grad_norm": 0.28211474418640137,
|
| 327 |
+
"learning_rate": 5e-05,
|
| 328 |
+
"loss": 0.576,
|
| 329 |
+
"num_tokens": 124244420.0,
|
| 330 |
+
"step": 40
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.3918757467144564,
|
| 334 |
+
"grad_norm": 0.29719066619873047,
|
| 335 |
+
"learning_rate": 5e-05,
|
| 336 |
+
"loss": 0.5776,
|
| 337 |
+
"num_tokens": 127347282.0,
|
| 338 |
+
"step": 41
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.4014336917562724,
|
| 342 |
+
"grad_norm": 0.28514203429222107,
|
| 343 |
+
"learning_rate": 5e-05,
|
| 344 |
+
"loss": 0.5739,
|
| 345 |
+
"num_tokens": 130425105.0,
|
| 346 |
+
"step": 42
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.4109916367980884,
|
| 350 |
+
"grad_norm": 0.24759361147880554,
|
| 351 |
+
"learning_rate": 5e-05,
|
| 352 |
+
"loss": 0.5614,
|
| 353 |
+
"num_tokens": 133566882.0,
|
| 354 |
+
"step": 43
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.42054958183990443,
|
| 358 |
+
"grad_norm": 0.3673174977302551,
|
| 359 |
+
"learning_rate": 5e-05,
|
| 360 |
+
"loss": 0.5755,
|
| 361 |
+
"num_tokens": 136693576.0,
|
| 362 |
+
"step": 44
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.43010752688172044,
|
| 366 |
+
"grad_norm": 0.2918493151664734,
|
| 367 |
+
"learning_rate": 5e-05,
|
| 368 |
+
"loss": 0.5741,
|
| 369 |
+
"num_tokens": 139836069.0,
|
| 370 |
+
"step": 45
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.43966547192353644,
|
| 374 |
+
"grad_norm": 0.20241530239582062,
|
| 375 |
+
"learning_rate": 5e-05,
|
| 376 |
+
"loss": 0.5659,
|
| 377 |
+
"num_tokens": 142903826.0,
|
| 378 |
+
"step": 46
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.44922341696535245,
|
| 382 |
+
"grad_norm": 0.36211881041526794,
|
| 383 |
+
"learning_rate": 5e-05,
|
| 384 |
+
"loss": 0.5652,
|
| 385 |
+
"num_tokens": 146013234.0,
|
| 386 |
+
"step": 47
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.45878136200716846,
|
| 390 |
+
"grad_norm": 0.27178287506103516,
|
| 391 |
+
"learning_rate": 5e-05,
|
| 392 |
+
"loss": 0.5639,
|
| 393 |
+
"num_tokens": 149106780.0,
|
| 394 |
+
"step": 48
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.46833930704898447,
|
| 398 |
+
"grad_norm": 0.22404040396213531,
|
| 399 |
+
"learning_rate": 5e-05,
|
| 400 |
+
"loss": 0.5588,
|
| 401 |
+
"num_tokens": 152240290.0,
|
| 402 |
+
"step": 49
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.4778972520908005,
|
| 406 |
+
"grad_norm": 0.267503023147583,
|
| 407 |
+
"learning_rate": 5e-05,
|
| 408 |
+
"loss": 0.5634,
|
| 409 |
+
"num_tokens": 155343192.0,
|
| 410 |
+
"step": 50
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.4874551971326165,
|
| 414 |
+
"grad_norm": 0.22685855627059937,
|
| 415 |
+
"learning_rate": 5e-05,
|
| 416 |
+
"loss": 0.5624,
|
| 417 |
+
"num_tokens": 158483136.0,
|
| 418 |
+
"step": 51
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.4970131421744325,
|
| 422 |
+
"grad_norm": 0.3110021948814392,
|
| 423 |
+
"learning_rate": 5e-05,
|
| 424 |
+
"loss": 0.5543,
|
| 425 |
+
"num_tokens": 161533766.0,
|
| 426 |
+
"step": 52
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.5065710872162486,
|
| 430 |
+
"grad_norm": 0.23207558691501617,
|
| 431 |
+
"learning_rate": 5e-05,
|
| 432 |
+
"loss": 0.546,
|
| 433 |
+
"num_tokens": 164664119.0,
|
| 434 |
+
"step": 53
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.5161290322580645,
|
| 438 |
+
"grad_norm": 0.33425262570381165,
|
| 439 |
+
"learning_rate": 5e-05,
|
| 440 |
+
"loss": 0.5544,
|
| 441 |
+
"num_tokens": 167808973.0,
|
| 442 |
+
"step": 54
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.5256869772998806,
|
| 446 |
+
"grad_norm": 0.2662634253501892,
|
| 447 |
+
"learning_rate": 5e-05,
|
| 448 |
+
"loss": 0.5526,
|
| 449 |
+
"num_tokens": 170887441.0,
|
| 450 |
+
"step": 55
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.5352449223416965,
|
| 454 |
+
"grad_norm": 0.28920575976371765,
|
| 455 |
+
"learning_rate": 5e-05,
|
| 456 |
+
"loss": 0.5536,
|
| 457 |
+
"num_tokens": 174005920.0,
|
| 458 |
+
"step": 56
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.5448028673835126,
|
| 462 |
+
"grad_norm": 0.26591673493385315,
|
| 463 |
+
"learning_rate": 5e-05,
|
| 464 |
+
"loss": 0.5481,
|
| 465 |
+
"num_tokens": 177097398.0,
|
| 466 |
+
"step": 57
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.5543608124253285,
|
| 470 |
+
"grad_norm": 0.2847670614719391,
|
| 471 |
+
"learning_rate": 5e-05,
|
| 472 |
+
"loss": 0.5536,
|
| 473 |
+
"num_tokens": 180221288.0,
|
| 474 |
+
"step": 58
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.5639187574671446,
|
| 478 |
+
"grad_norm": 0.2541261613368988,
|
| 479 |
+
"learning_rate": 5e-05,
|
| 480 |
+
"loss": 0.542,
|
| 481 |
+
"num_tokens": 183289863.0,
|
| 482 |
+
"step": 59
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.5734767025089605,
|
| 486 |
+
"grad_norm": 0.275193989276886,
|
| 487 |
+
"learning_rate": 5e-05,
|
| 488 |
+
"loss": 0.5456,
|
| 489 |
+
"num_tokens": 186406120.0,
|
| 490 |
+
"step": 60
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.5830346475507766,
|
| 494 |
+
"grad_norm": 0.26497912406921387,
|
| 495 |
+
"learning_rate": 5e-05,
|
| 496 |
+
"loss": 0.5425,
|
| 497 |
+
"num_tokens": 189471040.0,
|
| 498 |
+
"step": 61
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.5925925925925926,
|
| 502 |
+
"grad_norm": 0.27101513743400574,
|
| 503 |
+
"learning_rate": 5e-05,
|
| 504 |
+
"loss": 0.5386,
|
| 505 |
+
"num_tokens": 192575364.0,
|
| 506 |
+
"step": 62
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.6021505376344086,
|
| 510 |
+
"grad_norm": 0.26009494066238403,
|
| 511 |
+
"learning_rate": 5e-05,
|
| 512 |
+
"loss": 0.5318,
|
| 513 |
+
"num_tokens": 195684975.0,
|
| 514 |
+
"step": 63
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.6117084826762246,
|
| 518 |
+
"grad_norm": 0.22097371518611908,
|
| 519 |
+
"learning_rate": 5e-05,
|
| 520 |
+
"loss": 0.5411,
|
| 521 |
+
"num_tokens": 198779716.0,
|
| 522 |
+
"step": 64
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.6212664277180406,
|
| 526 |
+
"grad_norm": 0.3351891040802002,
|
| 527 |
+
"learning_rate": 5e-05,
|
| 528 |
+
"loss": 0.5458,
|
| 529 |
+
"num_tokens": 201906900.0,
|
| 530 |
+
"step": 65
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.6308243727598566,
|
| 534 |
+
"grad_norm": 0.23704691231250763,
|
| 535 |
+
"learning_rate": 5e-05,
|
| 536 |
+
"loss": 0.5397,
|
| 537 |
+
"num_tokens": 205031188.0,
|
| 538 |
+
"step": 66
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.6403823178016727,
|
| 542 |
+
"grad_norm": 0.32154443860054016,
|
| 543 |
+
"learning_rate": 5e-05,
|
| 544 |
+
"loss": 0.5445,
|
| 545 |
+
"num_tokens": 208116939.0,
|
| 546 |
+
"step": 67
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 0.6499402628434886,
|
| 550 |
+
"grad_norm": 0.2630949020385742,
|
| 551 |
+
"learning_rate": 5e-05,
|
| 552 |
+
"loss": 0.5322,
|
| 553 |
+
"num_tokens": 211211227.0,
|
| 554 |
+
"step": 68
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.6594982078853047,
|
| 558 |
+
"grad_norm": 0.31168922781944275,
|
| 559 |
+
"learning_rate": 5e-05,
|
| 560 |
+
"loss": 0.5324,
|
| 561 |
+
"num_tokens": 214301570.0,
|
| 562 |
+
"step": 69
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.6690561529271206,
|
| 566 |
+
"grad_norm": 0.2585446834564209,
|
| 567 |
+
"learning_rate": 5e-05,
|
| 568 |
+
"loss": 0.5386,
|
| 569 |
+
"num_tokens": 217393704.0,
|
| 570 |
+
"step": 70
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.6786140979689367,
|
| 574 |
+
"grad_norm": 0.25293031334877014,
|
| 575 |
+
"learning_rate": 5e-05,
|
| 576 |
+
"loss": 0.5223,
|
| 577 |
+
"num_tokens": 220482293.0,
|
| 578 |
+
"step": 71
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.6881720430107527,
|
| 582 |
+
"grad_norm": 0.3039219379425049,
|
| 583 |
+
"learning_rate": 5e-05,
|
| 584 |
+
"loss": 0.5337,
|
| 585 |
+
"num_tokens": 223590652.0,
|
| 586 |
+
"step": 72
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.6977299880525687,
|
| 590 |
+
"grad_norm": 0.2043919861316681,
|
| 591 |
+
"learning_rate": 5e-05,
|
| 592 |
+
"loss": 0.5251,
|
| 593 |
+
"num_tokens": 226675856.0,
|
| 594 |
+
"step": 73
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.7072879330943848,
|
| 598 |
+
"grad_norm": 0.290919691324234,
|
| 599 |
+
"learning_rate": 5e-05,
|
| 600 |
+
"loss": 0.5343,
|
| 601 |
+
"num_tokens": 229778501.0,
|
| 602 |
+
"step": 74
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.7168458781362007,
|
| 606 |
+
"grad_norm": 0.24795055389404297,
|
| 607 |
+
"learning_rate": 5e-05,
|
| 608 |
+
"loss": 0.5356,
|
| 609 |
+
"num_tokens": 232829595.0,
|
| 610 |
+
"step": 75
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.7264038231780168,
|
| 614 |
+
"grad_norm": 0.25901439785957336,
|
| 615 |
+
"learning_rate": 5e-05,
|
| 616 |
+
"loss": 0.5258,
|
| 617 |
+
"num_tokens": 235940446.0,
|
| 618 |
+
"step": 76
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.7359617682198327,
|
| 622 |
+
"grad_norm": 0.30405953526496887,
|
| 623 |
+
"learning_rate": 5e-05,
|
| 624 |
+
"loss": 0.5361,
|
| 625 |
+
"num_tokens": 239028046.0,
|
| 626 |
+
"step": 77
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.7455197132616488,
|
| 630 |
+
"grad_norm": 0.23060917854309082,
|
| 631 |
+
"learning_rate": 5e-05,
|
| 632 |
+
"loss": 0.5361,
|
| 633 |
+
"num_tokens": 242205870.0,
|
| 634 |
+
"step": 78
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.7550776583034647,
|
| 638 |
+
"grad_norm": 0.24737045168876648,
|
| 639 |
+
"learning_rate": 5e-05,
|
| 640 |
+
"loss": 0.5204,
|
| 641 |
+
"num_tokens": 245233676.0,
|
| 642 |
+
"step": 79
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.7646356033452808,
|
| 646 |
+
"grad_norm": 0.22408932447433472,
|
| 647 |
+
"learning_rate": 5e-05,
|
| 648 |
+
"loss": 0.5202,
|
| 649 |
+
"num_tokens": 248356444.0,
|
| 650 |
+
"step": 80
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.7741935483870968,
|
| 654 |
+
"grad_norm": 0.328327476978302,
|
| 655 |
+
"learning_rate": 5e-05,
|
| 656 |
+
"loss": 0.5311,
|
| 657 |
+
"num_tokens": 251468079.0,
|
| 658 |
+
"step": 81
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.7837514934289128,
|
| 662 |
+
"grad_norm": 0.20574553310871124,
|
| 663 |
+
"learning_rate": 5e-05,
|
| 664 |
+
"loss": 0.5231,
|
| 665 |
+
"num_tokens": 254594703.0,
|
| 666 |
+
"step": 82
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.7933094384707288,
|
| 670 |
+
"grad_norm": 0.305171936750412,
|
| 671 |
+
"learning_rate": 5e-05,
|
| 672 |
+
"loss": 0.5264,
|
| 673 |
+
"num_tokens": 257723496.0,
|
| 674 |
+
"step": 83
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.8028673835125448,
|
| 678 |
+
"grad_norm": 0.20093268156051636,
|
| 679 |
+
"learning_rate": 5e-05,
|
| 680 |
+
"loss": 0.5274,
|
| 681 |
+
"num_tokens": 260768286.0,
|
| 682 |
+
"step": 84
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.8124253285543608,
|
| 686 |
+
"grad_norm": 0.3125307559967041,
|
| 687 |
+
"learning_rate": 5e-05,
|
| 688 |
+
"loss": 0.5313,
|
| 689 |
+
"num_tokens": 263875942.0,
|
| 690 |
+
"step": 85
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.8219832735961768,
|
| 694 |
+
"grad_norm": 0.20460698008537292,
|
| 695 |
+
"learning_rate": 5e-05,
|
| 696 |
+
"loss": 0.5217,
|
| 697 |
+
"num_tokens": 266985727.0,
|
| 698 |
+
"step": 86
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.8315412186379928,
|
| 702 |
+
"grad_norm": 0.2367510348558426,
|
| 703 |
+
"learning_rate": 5e-05,
|
| 704 |
+
"loss": 0.5163,
|
| 705 |
+
"num_tokens": 270075318.0,
|
| 706 |
+
"step": 87
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.8410991636798089,
|
| 710 |
+
"grad_norm": 0.2568177580833435,
|
| 711 |
+
"learning_rate": 5e-05,
|
| 712 |
+
"loss": 0.5241,
|
| 713 |
+
"num_tokens": 273156355.0,
|
| 714 |
+
"step": 88
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.8506571087216248,
|
| 718 |
+
"grad_norm": 0.2784541845321655,
|
| 719 |
+
"learning_rate": 5e-05,
|
| 720 |
+
"loss": 0.5223,
|
| 721 |
+
"num_tokens": 276259662.0,
|
| 722 |
+
"step": 89
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.8602150537634409,
|
| 726 |
+
"grad_norm": 0.25873690843582153,
|
| 727 |
+
"learning_rate": 5e-05,
|
| 728 |
+
"loss": 0.5256,
|
| 729 |
+
"num_tokens": 279394317.0,
|
| 730 |
+
"step": 90
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.8697729988052568,
|
| 734 |
+
"grad_norm": 0.2528490126132965,
|
| 735 |
+
"learning_rate": 5e-05,
|
| 736 |
+
"loss": 0.5199,
|
| 737 |
+
"num_tokens": 282498565.0,
|
| 738 |
+
"step": 91
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.8793309438470729,
|
| 742 |
+
"grad_norm": 0.24740688502788544,
|
| 743 |
+
"learning_rate": 5e-05,
|
| 744 |
+
"loss": 0.5141,
|
| 745 |
+
"num_tokens": 285610462.0,
|
| 746 |
+
"step": 92
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.8888888888888888,
|
| 750 |
+
"grad_norm": 0.24610859155654907,
|
| 751 |
+
"learning_rate": 5e-05,
|
| 752 |
+
"loss": 0.5197,
|
| 753 |
+
"num_tokens": 288771247.0,
|
| 754 |
+
"step": 93
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 0.8984468339307049,
|
| 758 |
+
"grad_norm": 0.24038155376911163,
|
| 759 |
+
"learning_rate": 5e-05,
|
| 760 |
+
"loss": 0.5191,
|
| 761 |
+
"num_tokens": 291911091.0,
|
| 762 |
+
"step": 94
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.9080047789725209,
|
| 766 |
+
"grad_norm": 0.2939693033695221,
|
| 767 |
+
"learning_rate": 5e-05,
|
| 768 |
+
"loss": 0.5147,
|
| 769 |
+
"num_tokens": 294998183.0,
|
| 770 |
+
"step": 95
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.9175627240143369,
|
| 774 |
+
"grad_norm": 0.19252680242061615,
|
| 775 |
+
"learning_rate": 5e-05,
|
| 776 |
+
"loss": 0.5138,
|
| 777 |
+
"num_tokens": 298050683.0,
|
| 778 |
+
"step": 96
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.927120669056153,
|
| 782 |
+
"grad_norm": 0.23349586129188538,
|
| 783 |
+
"learning_rate": 5e-05,
|
| 784 |
+
"loss": 0.5172,
|
| 785 |
+
"num_tokens": 301164218.0,
|
| 786 |
+
"step": 97
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.9366786140979689,
|
| 790 |
+
"grad_norm": 0.20470231771469116,
|
| 791 |
+
"learning_rate": 5e-05,
|
| 792 |
+
"loss": 0.5132,
|
| 793 |
+
"num_tokens": 304261920.0,
|
| 794 |
+
"step": 98
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.946236559139785,
|
| 798 |
+
"grad_norm": 0.27620136737823486,
|
| 799 |
+
"learning_rate": 5e-05,
|
| 800 |
+
"loss": 0.5195,
|
| 801 |
+
"num_tokens": 307412237.0,
|
| 802 |
+
"step": 99
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 0.955794504181601,
|
| 806 |
+
"grad_norm": 0.21717092394828796,
|
| 807 |
+
"learning_rate": 5e-05,
|
| 808 |
+
"loss": 0.5172,
|
| 809 |
+
"num_tokens": 310507993.0,
|
| 810 |
+
"step": 100
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.965352449223417,
|
| 814 |
+
"grad_norm": 0.29017373919487,
|
| 815 |
+
"learning_rate": 5e-05,
|
| 816 |
+
"loss": 0.5107,
|
| 817 |
+
"num_tokens": 313624257.0,
|
| 818 |
+
"step": 101
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.974910394265233,
|
| 822 |
+
"grad_norm": 0.28068530559539795,
|
| 823 |
+
"learning_rate": 5e-05,
|
| 824 |
+
"loss": 0.5176,
|
| 825 |
+
"num_tokens": 316722666.0,
|
| 826 |
+
"step": 102
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 0.984468339307049,
|
| 830 |
+
"grad_norm": 0.2316262573003769,
|
| 831 |
+
"learning_rate": 5e-05,
|
| 832 |
+
"loss": 0.5162,
|
| 833 |
+
"num_tokens": 319820496.0,
|
| 834 |
+
"step": 103
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.994026284348865,
|
| 838 |
+
"grad_norm": 0.30137985944747925,
|
| 839 |
+
"learning_rate": 5e-05,
|
| 840 |
+
"loss": 0.5102,
|
| 841 |
+
"num_tokens": 322901137.0,
|
| 842 |
+
"step": 104
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.0,
|
| 846 |
+
"grad_norm": 0.30137985944747925,
|
| 847 |
+
"learning_rate": 5e-05,
|
| 848 |
+
"loss": 0.5069,
|
| 849 |
+
"num_tokens": 324860579.0,
|
| 850 |
+
"step": 105
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 1.009557945041816,
|
| 854 |
+
"grad_norm": 0.32997432351112366,
|
| 855 |
+
"learning_rate": 5e-05,
|
| 856 |
+
"loss": 0.4761,
|
| 857 |
+
"num_tokens": 327927654.0,
|
| 858 |
+
"step": 106
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.0191158900836321,
|
| 862 |
+
"grad_norm": 0.32110321521759033,
|
| 863 |
+
"learning_rate": 5e-05,
|
| 864 |
+
"loss": 0.4785,
|
| 865 |
+
"num_tokens": 331066424.0,
|
| 866 |
+
"step": 107
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.028673835125448,
|
| 870 |
+
"grad_norm": 0.20342878997325897,
|
| 871 |
+
"learning_rate": 5e-05,
|
| 872 |
+
"loss": 0.4845,
|
| 873 |
+
"num_tokens": 334162426.0,
|
| 874 |
+
"step": 108
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.038231780167264,
|
| 878 |
+
"grad_norm": 0.3325869143009186,
|
| 879 |
+
"learning_rate": 5e-05,
|
| 880 |
+
"loss": 0.4822,
|
| 881 |
+
"num_tokens": 337302238.0,
|
| 882 |
+
"step": 109
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.04778972520908,
|
| 886 |
+
"grad_norm": 0.22826018929481506,
|
| 887 |
+
"learning_rate": 5e-05,
|
| 888 |
+
"loss": 0.4824,
|
| 889 |
+
"num_tokens": 340447280.0,
|
| 890 |
+
"step": 110
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.0573476702508962,
|
| 894 |
+
"grad_norm": 0.3223550021648407,
|
| 895 |
+
"learning_rate": 5e-05,
|
| 896 |
+
"loss": 0.4696,
|
| 897 |
+
"num_tokens": 343568761.0,
|
| 898 |
+
"step": 111
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.066905615292712,
|
| 902 |
+
"grad_norm": 0.21368011832237244,
|
| 903 |
+
"learning_rate": 5e-05,
|
| 904 |
+
"loss": 0.4756,
|
| 905 |
+
"num_tokens": 346718666.0,
|
| 906 |
+
"step": 112
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.076463560334528,
|
| 910 |
+
"grad_norm": 0.24224598705768585,
|
| 911 |
+
"learning_rate": 5e-05,
|
| 912 |
+
"loss": 0.4765,
|
| 913 |
+
"num_tokens": 349842177.0,
|
| 914 |
+
"step": 113
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 1.086021505376344,
|
| 918 |
+
"grad_norm": 0.2377299815416336,
|
| 919 |
+
"learning_rate": 5e-05,
|
| 920 |
+
"loss": 0.4728,
|
| 921 |
+
"num_tokens": 352932093.0,
|
| 922 |
+
"step": 114
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 1.0955794504181602,
|
| 926 |
+
"grad_norm": 0.23792093992233276,
|
| 927 |
+
"learning_rate": 5e-05,
|
| 928 |
+
"loss": 0.4777,
|
| 929 |
+
"num_tokens": 356033224.0,
|
| 930 |
+
"step": 115
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.1051373954599761,
|
| 934 |
+
"grad_norm": 0.26680827140808105,
|
| 935 |
+
"learning_rate": 5e-05,
|
| 936 |
+
"loss": 0.4778,
|
| 937 |
+
"num_tokens": 359191089.0,
|
| 938 |
+
"step": 116
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 1.114695340501792,
|
| 942 |
+
"grad_norm": 0.26498955488204956,
|
| 943 |
+
"learning_rate": 5e-05,
|
| 944 |
+
"loss": 0.4739,
|
| 945 |
+
"num_tokens": 362282504.0,
|
| 946 |
+
"step": 117
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 1.124253285543608,
|
| 950 |
+
"grad_norm": 0.2889910936355591,
|
| 951 |
+
"learning_rate": 5e-05,
|
| 952 |
+
"loss": 0.4751,
|
| 953 |
+
"num_tokens": 365375038.0,
|
| 954 |
+
"step": 118
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.1338112305854242,
|
| 958 |
+
"grad_norm": 0.2168579399585724,
|
| 959 |
+
"learning_rate": 5e-05,
|
| 960 |
+
"loss": 0.4615,
|
| 961 |
+
"num_tokens": 368405060.0,
|
| 962 |
+
"step": 119
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 1.1433691756272402,
|
| 966 |
+
"grad_norm": 0.2727675437927246,
|
| 967 |
+
"learning_rate": 5e-05,
|
| 968 |
+
"loss": 0.4738,
|
| 969 |
+
"num_tokens": 371483975.0,
|
| 970 |
+
"step": 120
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.1529271206690561,
|
| 974 |
+
"grad_norm": 0.36416906118392944,
|
| 975 |
+
"learning_rate": 5e-05,
|
| 976 |
+
"loss": 0.4791,
|
| 977 |
+
"num_tokens": 374634435.0,
|
| 978 |
+
"step": 121
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.162485065710872,
|
| 982 |
+
"grad_norm": 0.18200910091400146,
|
| 983 |
+
"learning_rate": 5e-05,
|
| 984 |
+
"loss": 0.4731,
|
| 985 |
+
"num_tokens": 377757663.0,
|
| 986 |
+
"step": 122
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.1720430107526882,
|
| 990 |
+
"grad_norm": 0.386223167181015,
|
| 991 |
+
"learning_rate": 5e-05,
|
| 992 |
+
"loss": 0.4777,
|
| 993 |
+
"num_tokens": 380874183.0,
|
| 994 |
+
"step": 123
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 1.1816009557945042,
|
| 998 |
+
"grad_norm": 0.25184914469718933,
|
| 999 |
+
"learning_rate": 5e-05,
|
| 1000 |
+
"loss": 0.4793,
|
| 1001 |
+
"num_tokens": 383990561.0,
|
| 1002 |
+
"step": 124
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.1911589008363201,
|
| 1006 |
+
"grad_norm": 0.37080973386764526,
|
| 1007 |
+
"learning_rate": 5e-05,
|
| 1008 |
+
"loss": 0.4764,
|
| 1009 |
+
"num_tokens": 387086225.0,
|
| 1010 |
+
"step": 125
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.2007168458781363,
|
| 1014 |
+
"grad_norm": 0.27940818667411804,
|
| 1015 |
+
"learning_rate": 5e-05,
|
| 1016 |
+
"loss": 0.4718,
|
| 1017 |
+
"num_tokens": 390204819.0,
|
| 1018 |
+
"step": 126
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 1.2102747909199523,
|
| 1022 |
+
"grad_norm": 0.28235924243927,
|
| 1023 |
+
"learning_rate": 5e-05,
|
| 1024 |
+
"loss": 0.4713,
|
| 1025 |
+
"num_tokens": 393287937.0,
|
| 1026 |
+
"step": 127
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.2198327359617682,
|
| 1030 |
+
"grad_norm": 0.28537049889564514,
|
| 1031 |
+
"learning_rate": 5e-05,
|
| 1032 |
+
"loss": 0.4746,
|
| 1033 |
+
"num_tokens": 396363663.0,
|
| 1034 |
+
"step": 128
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.2293906810035842,
|
| 1038 |
+
"grad_norm": 0.3287442922592163,
|
| 1039 |
+
"learning_rate": 5e-05,
|
| 1040 |
+
"loss": 0.4796,
|
| 1041 |
+
"num_tokens": 399462801.0,
|
| 1042 |
+
"step": 129
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 1.2389486260454001,
|
| 1046 |
+
"grad_norm": 0.23722870647907257,
|
| 1047 |
+
"learning_rate": 5e-05,
|
| 1048 |
+
"loss": 0.4674,
|
| 1049 |
+
"num_tokens": 402561690.0,
|
| 1050 |
+
"step": 130
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 1.2485065710872163,
|
| 1054 |
+
"grad_norm": 0.26579034328460693,
|
| 1055 |
+
"learning_rate": 5e-05,
|
| 1056 |
+
"loss": 0.46,
|
| 1057 |
+
"num_tokens": 405671986.0,
|
| 1058 |
+
"step": 131
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.2580645161290323,
|
| 1062 |
+
"grad_norm": 0.21499669551849365,
|
| 1063 |
+
"learning_rate": 5e-05,
|
| 1064 |
+
"loss": 0.4633,
|
| 1065 |
+
"num_tokens": 408745166.0,
|
| 1066 |
+
"step": 132
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.2676224611708482,
|
| 1070 |
+
"grad_norm": 0.26388466358184814,
|
| 1071 |
+
"learning_rate": 5e-05,
|
| 1072 |
+
"loss": 0.4736,
|
| 1073 |
+
"num_tokens": 411848560.0,
|
| 1074 |
+
"step": 133
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 1.2771804062126644,
|
| 1078 |
+
"grad_norm": 0.22143350541591644,
|
| 1079 |
+
"learning_rate": 5e-05,
|
| 1080 |
+
"loss": 0.4624,
|
| 1081 |
+
"num_tokens": 414961042.0,
|
| 1082 |
+
"step": 134
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 1.2867383512544803,
|
| 1086 |
+
"grad_norm": 0.27563315629959106,
|
| 1087 |
+
"learning_rate": 5e-05,
|
| 1088 |
+
"loss": 0.4645,
|
| 1089 |
+
"num_tokens": 418038833.0,
|
| 1090 |
+
"step": 135
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 1.2962962962962963,
|
| 1094 |
+
"grad_norm": 0.2135905772447586,
|
| 1095 |
+
"learning_rate": 5e-05,
|
| 1096 |
+
"loss": 0.4732,
|
| 1097 |
+
"num_tokens": 421176052.0,
|
| 1098 |
+
"step": 136
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 1.3058542413381122,
|
| 1102 |
+
"grad_norm": 0.24489735066890717,
|
| 1103 |
+
"learning_rate": 5e-05,
|
| 1104 |
+
"loss": 0.472,
|
| 1105 |
+
"num_tokens": 424278020.0,
|
| 1106 |
+
"step": 137
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 1.3154121863799282,
|
| 1110 |
+
"grad_norm": 0.22754637897014618,
|
| 1111 |
+
"learning_rate": 5e-05,
|
| 1112 |
+
"loss": 0.4691,
|
| 1113 |
+
"num_tokens": 427397110.0,
|
| 1114 |
+
"step": 138
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 1.3249701314217444,
|
| 1118 |
+
"grad_norm": 0.24249699711799622,
|
| 1119 |
+
"learning_rate": 5e-05,
|
| 1120 |
+
"loss": 0.4757,
|
| 1121 |
+
"num_tokens": 430504573.0,
|
| 1122 |
+
"step": 139
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.3345280764635603,
|
| 1126 |
+
"grad_norm": 0.23354573547840118,
|
| 1127 |
+
"learning_rate": 5e-05,
|
| 1128 |
+
"loss": 0.4605,
|
| 1129 |
+
"num_tokens": 433626191.0,
|
| 1130 |
+
"step": 140
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.3440860215053765,
|
| 1134 |
+
"grad_norm": 0.2180928736925125,
|
| 1135 |
+
"learning_rate": 5e-05,
|
| 1136 |
+
"loss": 0.4679,
|
| 1137 |
+
"num_tokens": 436727625.0,
|
| 1138 |
+
"step": 141
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 1.3536439665471924,
|
| 1142 |
+
"grad_norm": 0.2624548077583313,
|
| 1143 |
+
"learning_rate": 5e-05,
|
| 1144 |
+
"loss": 0.4704,
|
| 1145 |
+
"num_tokens": 439824215.0,
|
| 1146 |
+
"step": 142
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 1.3632019115890084,
|
| 1150 |
+
"grad_norm": 0.22936105728149414,
|
| 1151 |
+
"learning_rate": 5e-05,
|
| 1152 |
+
"loss": 0.4638,
|
| 1153 |
+
"num_tokens": 442903885.0,
|
| 1154 |
+
"step": 143
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 1.3727598566308243,
|
| 1158 |
+
"grad_norm": 0.2361782044172287,
|
| 1159 |
+
"learning_rate": 5e-05,
|
| 1160 |
+
"loss": 0.4722,
|
| 1161 |
+
"num_tokens": 446049216.0,
|
| 1162 |
+
"step": 144
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 1.3823178016726403,
|
| 1166 |
+
"grad_norm": 0.22354678809642792,
|
| 1167 |
+
"learning_rate": 5e-05,
|
| 1168 |
+
"loss": 0.4702,
|
| 1169 |
+
"num_tokens": 449164822.0,
|
| 1170 |
+
"step": 145
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.3918757467144565,
|
| 1174 |
+
"grad_norm": 0.2597694396972656,
|
| 1175 |
+
"learning_rate": 5e-05,
|
| 1176 |
+
"loss": 0.4659,
|
| 1177 |
+
"num_tokens": 452296016.0,
|
| 1178 |
+
"step": 146
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.4014336917562724,
|
| 1182 |
+
"grad_norm": 0.18445123732089996,
|
| 1183 |
+
"learning_rate": 5e-05,
|
| 1184 |
+
"loss": 0.4675,
|
| 1185 |
+
"num_tokens": 455409198.0,
|
| 1186 |
+
"step": 147
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 1.4109916367980884,
|
| 1190 |
+
"grad_norm": 0.24752481281757355,
|
| 1191 |
+
"learning_rate": 5e-05,
|
| 1192 |
+
"loss": 0.4665,
|
| 1193 |
+
"num_tokens": 458552498.0,
|
| 1194 |
+
"step": 148
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.4205495818399045,
|
| 1198 |
+
"grad_norm": 0.23295103013515472,
|
| 1199 |
+
"learning_rate": 5e-05,
|
| 1200 |
+
"loss": 0.4642,
|
| 1201 |
+
"num_tokens": 461598172.0,
|
| 1202 |
+
"step": 149
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 1.4301075268817205,
|
| 1206 |
+
"grad_norm": 0.26925209164619446,
|
| 1207 |
+
"learning_rate": 5e-05,
|
| 1208 |
+
"loss": 0.4618,
|
| 1209 |
+
"num_tokens": 464763731.0,
|
| 1210 |
+
"step": 150
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 1.4396654719235364,
|
| 1214 |
+
"grad_norm": 0.1971091479063034,
|
| 1215 |
+
"learning_rate": 5e-05,
|
| 1216 |
+
"loss": 0.4639,
|
| 1217 |
+
"num_tokens": 467905236.0,
|
| 1218 |
+
"step": 151
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 1.4492234169653524,
|
| 1222 |
+
"grad_norm": 0.2771715819835663,
|
| 1223 |
+
"learning_rate": 5e-05,
|
| 1224 |
+
"loss": 0.4645,
|
| 1225 |
+
"num_tokens": 471038706.0,
|
| 1226 |
+
"step": 152
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 1.4587813620071683,
|
| 1230 |
+
"grad_norm": 0.2006852775812149,
|
| 1231 |
+
"learning_rate": 5e-05,
|
| 1232 |
+
"loss": 0.4664,
|
| 1233 |
+
"num_tokens": 474170594.0,
|
| 1234 |
+
"step": 153
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.4683393070489845,
|
| 1238 |
+
"grad_norm": 0.27502164244651794,
|
| 1239 |
+
"learning_rate": 5e-05,
|
| 1240 |
+
"loss": 0.4581,
|
| 1241 |
+
"num_tokens": 477293327.0,
|
| 1242 |
+
"step": 154
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.4778972520908005,
|
| 1246 |
+
"grad_norm": 0.19383655488491058,
|
| 1247 |
+
"learning_rate": 5e-05,
|
| 1248 |
+
"loss": 0.4641,
|
| 1249 |
+
"num_tokens": 480356338.0,
|
| 1250 |
+
"step": 155
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 1.4874551971326164,
|
| 1254 |
+
"grad_norm": 0.29519346356391907,
|
| 1255 |
+
"learning_rate": 5e-05,
|
| 1256 |
+
"loss": 0.4636,
|
| 1257 |
+
"num_tokens": 483466822.0,
|
| 1258 |
+
"step": 156
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 1.4970131421744326,
|
| 1262 |
+
"grad_norm": 0.24049213528633118,
|
| 1263 |
+
"learning_rate": 5e-05,
|
| 1264 |
+
"loss": 0.4708,
|
| 1265 |
+
"num_tokens": 486598025.0,
|
| 1266 |
+
"step": 157
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 1.5065710872162486,
|
| 1270 |
+
"grad_norm": 0.23467357456684113,
|
| 1271 |
+
"learning_rate": 5e-05,
|
| 1272 |
+
"loss": 0.4629,
|
| 1273 |
+
"num_tokens": 489710268.0,
|
| 1274 |
+
"step": 158
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 1.5161290322580645,
|
| 1278 |
+
"grad_norm": 0.23279620707035065,
|
| 1279 |
+
"learning_rate": 5e-05,
|
| 1280 |
+
"loss": 0.4642,
|
| 1281 |
+
"num_tokens": 492749301.0,
|
| 1282 |
+
"step": 159
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.5256869772998805,
|
| 1286 |
+
"grad_norm": 0.21989446878433228,
|
| 1287 |
+
"learning_rate": 5e-05,
|
| 1288 |
+
"loss": 0.4609,
|
| 1289 |
+
"num_tokens": 495858425.0,
|
| 1290 |
+
"step": 160
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.5352449223416964,
|
| 1294 |
+
"grad_norm": 0.20997576415538788,
|
| 1295 |
+
"learning_rate": 5e-05,
|
| 1296 |
+
"loss": 0.4615,
|
| 1297 |
+
"num_tokens": 499002124.0,
|
| 1298 |
+
"step": 161
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 1.5448028673835126,
|
| 1302 |
+
"grad_norm": 0.2443646341562271,
|
| 1303 |
+
"learning_rate": 5e-05,
|
| 1304 |
+
"loss": 0.4599,
|
| 1305 |
+
"num_tokens": 502158127.0,
|
| 1306 |
+
"step": 162
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 1.5543608124253285,
|
| 1310 |
+
"grad_norm": 0.2202104926109314,
|
| 1311 |
+
"learning_rate": 5e-05,
|
| 1312 |
+
"loss": 0.4685,
|
| 1313 |
+
"num_tokens": 505273653.0,
|
| 1314 |
+
"step": 163
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 1.5639187574671447,
|
| 1318 |
+
"grad_norm": 0.2284436672925949,
|
| 1319 |
+
"learning_rate": 5e-05,
|
| 1320 |
+
"loss": 0.4666,
|
| 1321 |
+
"num_tokens": 508383175.0,
|
| 1322 |
+
"step": 164
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 1.5734767025089607,
|
| 1326 |
+
"grad_norm": 0.21461881697177887,
|
| 1327 |
+
"learning_rate": 5e-05,
|
| 1328 |
+
"loss": 0.4649,
|
| 1329 |
+
"num_tokens": 511456188.0,
|
| 1330 |
+
"step": 165
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 1.5830346475507766,
|
| 1334 |
+
"grad_norm": 0.24341613054275513,
|
| 1335 |
+
"learning_rate": 5e-05,
|
| 1336 |
+
"loss": 0.4726,
|
| 1337 |
+
"num_tokens": 514504919.0,
|
| 1338 |
+
"step": 166
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.5925925925925926,
|
| 1342 |
+
"grad_norm": 0.2432275414466858,
|
| 1343 |
+
"learning_rate": 5e-05,
|
| 1344 |
+
"loss": 0.462,
|
| 1345 |
+
"num_tokens": 517638097.0,
|
| 1346 |
+
"step": 167
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 1.6021505376344085,
|
| 1350 |
+
"grad_norm": 0.1924099624156952,
|
| 1351 |
+
"learning_rate": 5e-05,
|
| 1352 |
+
"loss": 0.4574,
|
| 1353 |
+
"num_tokens": 520744764.0,
|
| 1354 |
+
"step": 168
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.6117084826762245,
|
| 1358 |
+
"grad_norm": 0.2210276573896408,
|
| 1359 |
+
"learning_rate": 5e-05,
|
| 1360 |
+
"loss": 0.4652,
|
| 1361 |
+
"num_tokens": 523847555.0,
|
| 1362 |
+
"step": 169
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 1.6212664277180406,
|
| 1366 |
+
"grad_norm": 0.24831326305866241,
|
| 1367 |
+
"learning_rate": 5e-05,
|
| 1368 |
+
"loss": 0.4622,
|
| 1369 |
+
"num_tokens": 526938439.0,
|
| 1370 |
+
"step": 170
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 1.6308243727598566,
|
| 1374 |
+
"grad_norm": 0.20027406513690948,
|
| 1375 |
+
"learning_rate": 5e-05,
|
| 1376 |
+
"loss": 0.4625,
|
| 1377 |
+
"num_tokens": 530100826.0,
|
| 1378 |
+
"step": 171
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 1.6403823178016728,
|
| 1382 |
+
"grad_norm": 0.2210628092288971,
|
| 1383 |
+
"learning_rate": 5e-05,
|
| 1384 |
+
"loss": 0.4543,
|
| 1385 |
+
"num_tokens": 533217368.0,
|
| 1386 |
+
"step": 172
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 1.6499402628434887,
|
| 1390 |
+
"grad_norm": 0.21531818807125092,
|
| 1391 |
+
"learning_rate": 5e-05,
|
| 1392 |
+
"loss": 0.464,
|
| 1393 |
+
"num_tokens": 536266028.0,
|
| 1394 |
+
"step": 173
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.6594982078853047,
|
| 1398 |
+
"grad_norm": 0.24346983432769775,
|
| 1399 |
+
"learning_rate": 5e-05,
|
| 1400 |
+
"loss": 0.4582,
|
| 1401 |
+
"num_tokens": 539353780.0,
|
| 1402 |
+
"step": 174
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.6690561529271206,
|
| 1406 |
+
"grad_norm": 0.18500176072120667,
|
| 1407 |
+
"learning_rate": 5e-05,
|
| 1408 |
+
"loss": 0.4611,
|
| 1409 |
+
"num_tokens": 542436030.0,
|
| 1410 |
+
"step": 175
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 1.6786140979689366,
|
| 1414 |
+
"grad_norm": 0.21900725364685059,
|
| 1415 |
+
"learning_rate": 5e-05,
|
| 1416 |
+
"loss": 0.4567,
|
| 1417 |
+
"num_tokens": 545556259.0,
|
| 1418 |
+
"step": 176
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 1.6881720430107527,
|
| 1422 |
+
"grad_norm": 0.26041945815086365,
|
| 1423 |
+
"learning_rate": 5e-05,
|
| 1424 |
+
"loss": 0.4609,
|
| 1425 |
+
"num_tokens": 548651036.0,
|
| 1426 |
+
"step": 177
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 1.6977299880525687,
|
| 1430 |
+
"grad_norm": 0.1868957281112671,
|
| 1431 |
+
"learning_rate": 5e-05,
|
| 1432 |
+
"loss": 0.4578,
|
| 1433 |
+
"num_tokens": 551763960.0,
|
| 1434 |
+
"step": 178
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 1.7072879330943849,
|
| 1438 |
+
"grad_norm": 0.30002743005752563,
|
| 1439 |
+
"learning_rate": 5e-05,
|
| 1440 |
+
"loss": 0.4621,
|
| 1441 |
+
"num_tokens": 554859971.0,
|
| 1442 |
+
"step": 179
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 1.7168458781362008,
|
| 1446 |
+
"grad_norm": 0.15785135328769684,
|
| 1447 |
+
"learning_rate": 5e-05,
|
| 1448 |
+
"loss": 0.4531,
|
| 1449 |
+
"num_tokens": 557960344.0,
|
| 1450 |
+
"step": 180
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.7264038231780168,
|
| 1454 |
+
"grad_norm": 0.21860018372535706,
|
| 1455 |
+
"learning_rate": 5e-05,
|
| 1456 |
+
"loss": 0.456,
|
| 1457 |
+
"num_tokens": 561070346.0,
|
| 1458 |
+
"step": 181
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 1.7359617682198327,
|
| 1462 |
+
"grad_norm": 0.21262319386005402,
|
| 1463 |
+
"learning_rate": 5e-05,
|
| 1464 |
+
"loss": 0.4548,
|
| 1465 |
+
"num_tokens": 564124513.0,
|
| 1466 |
+
"step": 182
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 1.7455197132616487,
|
| 1470 |
+
"grad_norm": 0.208367258310318,
|
| 1471 |
+
"learning_rate": 5e-05,
|
| 1472 |
+
"loss": 0.4597,
|
| 1473 |
+
"num_tokens": 567229917.0,
|
| 1474 |
+
"step": 183
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 1.7550776583034646,
|
| 1478 |
+
"grad_norm": 0.23050186038017273,
|
| 1479 |
+
"learning_rate": 5e-05,
|
| 1480 |
+
"loss": 0.4609,
|
| 1481 |
+
"num_tokens": 570377549.0,
|
| 1482 |
+
"step": 184
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 1.7646356033452808,
|
| 1486 |
+
"grad_norm": 0.19711245596408844,
|
| 1487 |
+
"learning_rate": 5e-05,
|
| 1488 |
+
"loss": 0.4596,
|
| 1489 |
+
"num_tokens": 573423384.0,
|
| 1490 |
+
"step": 185
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 1.7741935483870968,
|
| 1494 |
+
"grad_norm": 0.25033050775527954,
|
| 1495 |
+
"learning_rate": 5e-05,
|
| 1496 |
+
"loss": 0.4536,
|
| 1497 |
+
"num_tokens": 576529785.0,
|
| 1498 |
+
"step": 186
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 1.783751493428913,
|
| 1502 |
+
"grad_norm": 0.23019000887870789,
|
| 1503 |
+
"learning_rate": 5e-05,
|
| 1504 |
+
"loss": 0.457,
|
| 1505 |
+
"num_tokens": 579611243.0,
|
| 1506 |
+
"step": 187
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.7933094384707289,
|
| 1510 |
+
"grad_norm": 0.24459266662597656,
|
| 1511 |
+
"learning_rate": 5e-05,
|
| 1512 |
+
"loss": 0.4563,
|
| 1513 |
+
"num_tokens": 582684585.0,
|
| 1514 |
+
"step": 188
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 1.8028673835125448,
|
| 1518 |
+
"grad_norm": 0.2335624098777771,
|
| 1519 |
+
"learning_rate": 5e-05,
|
| 1520 |
+
"loss": 0.456,
|
| 1521 |
+
"num_tokens": 585749574.0,
|
| 1522 |
+
"step": 189
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 1.8124253285543608,
|
| 1526 |
+
"grad_norm": 0.21826566755771637,
|
| 1527 |
+
"learning_rate": 5e-05,
|
| 1528 |
+
"loss": 0.4545,
|
| 1529 |
+
"num_tokens": 588793971.0,
|
| 1530 |
+
"step": 190
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 1.8219832735961767,
|
| 1534 |
+
"grad_norm": 0.2118789553642273,
|
| 1535 |
+
"learning_rate": 5e-05,
|
| 1536 |
+
"loss": 0.4561,
|
| 1537 |
+
"num_tokens": 591863271.0,
|
| 1538 |
+
"step": 191
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 1.8315412186379927,
|
| 1542 |
+
"grad_norm": 0.25364288687705994,
|
| 1543 |
+
"learning_rate": 5e-05,
|
| 1544 |
+
"loss": 0.4582,
|
| 1545 |
+
"num_tokens": 594955013.0,
|
| 1546 |
+
"step": 192
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 1.8410991636798089,
|
| 1550 |
+
"grad_norm": 0.18223045766353607,
|
| 1551 |
+
"learning_rate": 5e-05,
|
| 1552 |
+
"loss": 0.4586,
|
| 1553 |
+
"num_tokens": 598054762.0,
|
| 1554 |
+
"step": 193
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 1.8506571087216248,
|
| 1558 |
+
"grad_norm": 0.19568181037902832,
|
| 1559 |
+
"learning_rate": 5e-05,
|
| 1560 |
+
"loss": 0.4542,
|
| 1561 |
+
"num_tokens": 601156311.0,
|
| 1562 |
+
"step": 194
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.860215053763441,
|
| 1566 |
+
"grad_norm": 0.21927742660045624,
|
| 1567 |
+
"learning_rate": 5e-05,
|
| 1568 |
+
"loss": 0.4593,
|
| 1569 |
+
"num_tokens": 604264490.0,
|
| 1570 |
+
"step": 195
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 1.869772998805257,
|
| 1574 |
+
"grad_norm": 0.22864626348018646,
|
| 1575 |
+
"learning_rate": 5e-05,
|
| 1576 |
+
"loss": 0.4582,
|
| 1577 |
+
"num_tokens": 607369801.0,
|
| 1578 |
+
"step": 196
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 1.8793309438470729,
|
| 1582 |
+
"grad_norm": 0.2220122367143631,
|
| 1583 |
+
"learning_rate": 5e-05,
|
| 1584 |
+
"loss": 0.4586,
|
| 1585 |
+
"num_tokens": 610464157.0,
|
| 1586 |
+
"step": 197
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 1.8888888888888888,
|
| 1590 |
+
"grad_norm": 0.20654238760471344,
|
| 1591 |
+
"learning_rate": 5e-05,
|
| 1592 |
+
"loss": 0.4572,
|
| 1593 |
+
"num_tokens": 613600074.0,
|
| 1594 |
+
"step": 198
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 1.8984468339307048,
|
| 1598 |
+
"grad_norm": 0.24055001139640808,
|
| 1599 |
+
"learning_rate": 5e-05,
|
| 1600 |
+
"loss": 0.447,
|
| 1601 |
+
"num_tokens": 616692135.0,
|
| 1602 |
+
"step": 199
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 1.9080047789725207,
|
| 1606 |
+
"grad_norm": 0.2102368324995041,
|
| 1607 |
+
"learning_rate": 5e-05,
|
| 1608 |
+
"loss": 0.4616,
|
| 1609 |
+
"num_tokens": 619865728.0,
|
| 1610 |
+
"step": 200
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 1.917562724014337,
|
| 1614 |
+
"grad_norm": 0.1804041713476181,
|
| 1615 |
+
"learning_rate": 5e-05,
|
| 1616 |
+
"loss": 0.456,
|
| 1617 |
+
"num_tokens": 622937270.0,
|
| 1618 |
+
"step": 201
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.927120669056153,
|
| 1622 |
+
"grad_norm": 0.20141775906085968,
|
| 1623 |
+
"learning_rate": 5e-05,
|
| 1624 |
+
"loss": 0.4573,
|
| 1625 |
+
"num_tokens": 626025595.0,
|
| 1626 |
+
"step": 202
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 1.936678614097969,
|
| 1630 |
+
"grad_norm": 0.23730237782001495,
|
| 1631 |
+
"learning_rate": 5e-05,
|
| 1632 |
+
"loss": 0.4588,
|
| 1633 |
+
"num_tokens": 629102565.0,
|
| 1634 |
+
"step": 203
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 1.946236559139785,
|
| 1638 |
+
"grad_norm": 0.20814980566501617,
|
| 1639 |
+
"learning_rate": 5e-05,
|
| 1640 |
+
"loss": 0.4596,
|
| 1641 |
+
"num_tokens": 632248389.0,
|
| 1642 |
+
"step": 204
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 1.955794504181601,
|
| 1646 |
+
"grad_norm": 0.23954477906227112,
|
| 1647 |
+
"learning_rate": 5e-05,
|
| 1648 |
+
"loss": 0.4608,
|
| 1649 |
+
"num_tokens": 635381252.0,
|
| 1650 |
+
"step": 205
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 1.965352449223417,
|
| 1654 |
+
"grad_norm": 0.2159876823425293,
|
| 1655 |
+
"learning_rate": 5e-05,
|
| 1656 |
+
"loss": 0.4568,
|
| 1657 |
+
"num_tokens": 638503290.0,
|
| 1658 |
+
"step": 206
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 1.9749103942652328,
|
| 1662 |
+
"grad_norm": 0.2248695194721222,
|
| 1663 |
+
"learning_rate": 5e-05,
|
| 1664 |
+
"loss": 0.4494,
|
| 1665 |
+
"num_tokens": 641566168.0,
|
| 1666 |
+
"step": 207
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 1.984468339307049,
|
| 1670 |
+
"grad_norm": 0.18469811975955963,
|
| 1671 |
+
"learning_rate": 5e-05,
|
| 1672 |
+
"loss": 0.4553,
|
| 1673 |
+
"num_tokens": 644664768.0,
|
| 1674 |
+
"step": 208
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.994026284348865,
|
| 1678 |
+
"grad_norm": 0.2029660940170288,
|
| 1679 |
+
"learning_rate": 5e-05,
|
| 1680 |
+
"loss": 0.458,
|
| 1681 |
+
"num_tokens": 647777196.0,
|
| 1682 |
+
"step": 209
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 2.0,
|
| 1686 |
+
"grad_norm": 0.2680661678314209,
|
| 1687 |
+
"learning_rate": 5e-05,
|
| 1688 |
+
"loss": 0.4472,
|
| 1689 |
+
"num_tokens": 649717884.0,
|
| 1690 |
+
"step": 210
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 2.009557945041816,
|
| 1694 |
+
"grad_norm": 0.2700228691101074,
|
| 1695 |
+
"learning_rate": 5e-05,
|
| 1696 |
+
"loss": 0.4049,
|
| 1697 |
+
"num_tokens": 652835514.0,
|
| 1698 |
+
"step": 211
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 2.019115890083632,
|
| 1702 |
+
"grad_norm": 0.21437576413154602,
|
| 1703 |
+
"learning_rate": 5e-05,
|
| 1704 |
+
"loss": 0.4077,
|
| 1705 |
+
"num_tokens": 655939553.0,
|
| 1706 |
+
"step": 212
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 2.028673835125448,
|
| 1710 |
+
"grad_norm": 0.2812529504299164,
|
| 1711 |
+
"learning_rate": 5e-05,
|
| 1712 |
+
"loss": 0.4043,
|
| 1713 |
+
"num_tokens": 659052255.0,
|
| 1714 |
+
"step": 213
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 2.0382317801672643,
|
| 1718 |
+
"grad_norm": 0.19706906378269196,
|
| 1719 |
+
"learning_rate": 5e-05,
|
| 1720 |
+
"loss": 0.4069,
|
| 1721 |
+
"num_tokens": 662144062.0,
|
| 1722 |
+
"step": 214
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 2.04778972520908,
|
| 1726 |
+
"grad_norm": 0.23425546288490295,
|
| 1727 |
+
"learning_rate": 5e-05,
|
| 1728 |
+
"loss": 0.3947,
|
| 1729 |
+
"num_tokens": 665243818.0,
|
| 1730 |
+
"step": 215
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 2.057347670250896,
|
| 1734 |
+
"grad_norm": 0.2341323047876358,
|
| 1735 |
+
"learning_rate": 5e-05,
|
| 1736 |
+
"loss": 0.4005,
|
| 1737 |
+
"num_tokens": 668343069.0,
|
| 1738 |
+
"step": 216
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 2.066905615292712,
|
| 1742 |
+
"grad_norm": 0.26267462968826294,
|
| 1743 |
+
"learning_rate": 5e-05,
|
| 1744 |
+
"loss": 0.4076,
|
| 1745 |
+
"num_tokens": 671491544.0,
|
| 1746 |
+
"step": 217
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 2.076463560334528,
|
| 1750 |
+
"grad_norm": 0.22123785316944122,
|
| 1751 |
+
"learning_rate": 5e-05,
|
| 1752 |
+
"loss": 0.3994,
|
| 1753 |
+
"num_tokens": 674603425.0,
|
| 1754 |
+
"step": 218
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 2.086021505376344,
|
| 1758 |
+
"grad_norm": 0.22117979824543,
|
| 1759 |
+
"learning_rate": 5e-05,
|
| 1760 |
+
"loss": 0.3999,
|
| 1761 |
+
"num_tokens": 677683109.0,
|
| 1762 |
+
"step": 219
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 2.09557945041816,
|
| 1766 |
+
"grad_norm": 0.25451090931892395,
|
| 1767 |
+
"learning_rate": 5e-05,
|
| 1768 |
+
"loss": 0.4035,
|
| 1769 |
+
"num_tokens": 680765161.0,
|
| 1770 |
+
"step": 220
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 2.105137395459976,
|
| 1774 |
+
"grad_norm": 0.21930952370166779,
|
| 1775 |
+
"learning_rate": 5e-05,
|
| 1776 |
+
"loss": 0.4027,
|
| 1777 |
+
"num_tokens": 683877744.0,
|
| 1778 |
+
"step": 221
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 2.1146953405017923,
|
| 1782 |
+
"grad_norm": 0.2268250733613968,
|
| 1783 |
+
"learning_rate": 5e-05,
|
| 1784 |
+
"loss": 0.4017,
|
| 1785 |
+
"num_tokens": 687007762.0,
|
| 1786 |
+
"step": 222
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 2.1242532855436083,
|
| 1790 |
+
"grad_norm": 0.21051234006881714,
|
| 1791 |
+
"learning_rate": 5e-05,
|
| 1792 |
+
"loss": 0.4083,
|
| 1793 |
+
"num_tokens": 690110389.0,
|
| 1794 |
+
"step": 223
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 2.133811230585424,
|
| 1798 |
+
"grad_norm": 0.22137832641601562,
|
| 1799 |
+
"learning_rate": 5e-05,
|
| 1800 |
+
"loss": 0.3908,
|
| 1801 |
+
"num_tokens": 693193272.0,
|
| 1802 |
+
"step": 224
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 2.14336917562724,
|
| 1806 |
+
"grad_norm": 0.24819853901863098,
|
| 1807 |
+
"learning_rate": 5e-05,
|
| 1808 |
+
"loss": 0.4037,
|
| 1809 |
+
"num_tokens": 696332399.0,
|
| 1810 |
+
"step": 225
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 2.152927120669056,
|
| 1814 |
+
"grad_norm": 0.2136155366897583,
|
| 1815 |
+
"learning_rate": 5e-05,
|
| 1816 |
+
"loss": 0.3972,
|
| 1817 |
+
"num_tokens": 699449474.0,
|
| 1818 |
+
"step": 226
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 2.162485065710872,
|
| 1822 |
+
"grad_norm": 0.21676744520664215,
|
| 1823 |
+
"learning_rate": 5e-05,
|
| 1824 |
+
"loss": 0.4009,
|
| 1825 |
+
"num_tokens": 702535376.0,
|
| 1826 |
+
"step": 227
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 2.172043010752688,
|
| 1830 |
+
"grad_norm": 0.21379505097866058,
|
| 1831 |
+
"learning_rate": 5e-05,
|
| 1832 |
+
"loss": 0.4048,
|
| 1833 |
+
"num_tokens": 705594342.0,
|
| 1834 |
+
"step": 228
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.1816009557945044,
|
| 1838 |
+
"grad_norm": 0.24907568097114563,
|
| 1839 |
+
"learning_rate": 5e-05,
|
| 1840 |
+
"loss": 0.4016,
|
| 1841 |
+
"num_tokens": 708681631.0,
|
| 1842 |
+
"step": 229
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 2.1911589008363204,
|
| 1846 |
+
"grad_norm": 0.23827004432678223,
|
| 1847 |
+
"learning_rate": 5e-05,
|
| 1848 |
+
"loss": 0.3897,
|
| 1849 |
+
"num_tokens": 711768161.0,
|
| 1850 |
+
"step": 230
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 2.2007168458781363,
|
| 1854 |
+
"grad_norm": 0.19475553929805756,
|
| 1855 |
+
"learning_rate": 5e-05,
|
| 1856 |
+
"loss": 0.3984,
|
| 1857 |
+
"num_tokens": 714866856.0,
|
| 1858 |
+
"step": 231
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 2.2102747909199523,
|
| 1862 |
+
"grad_norm": 0.22524000704288483,
|
| 1863 |
+
"learning_rate": 5e-05,
|
| 1864 |
+
"loss": 0.3997,
|
| 1865 |
+
"num_tokens": 717989725.0,
|
| 1866 |
+
"step": 232
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 2.219832735961768,
|
| 1870 |
+
"grad_norm": 0.22430852055549622,
|
| 1871 |
+
"learning_rate": 5e-05,
|
| 1872 |
+
"loss": 0.3981,
|
| 1873 |
+
"num_tokens": 721080176.0,
|
| 1874 |
+
"step": 233
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 2.229390681003584,
|
| 1878 |
+
"grad_norm": 0.2231917530298233,
|
| 1879 |
+
"learning_rate": 5e-05,
|
| 1880 |
+
"loss": 0.4044,
|
| 1881 |
+
"num_tokens": 724211432.0,
|
| 1882 |
+
"step": 234
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 2.2389486260454,
|
| 1886 |
+
"grad_norm": 0.24612534046173096,
|
| 1887 |
+
"learning_rate": 5e-05,
|
| 1888 |
+
"loss": 0.3955,
|
| 1889 |
+
"num_tokens": 727270220.0,
|
| 1890 |
+
"step": 235
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 2.248506571087216,
|
| 1894 |
+
"grad_norm": 0.2350219190120697,
|
| 1895 |
+
"learning_rate": 5e-05,
|
| 1896 |
+
"loss": 0.4062,
|
| 1897 |
+
"num_tokens": 730444444.0,
|
| 1898 |
+
"step": 236
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 2.258064516129032,
|
| 1902 |
+
"grad_norm": 0.24301743507385254,
|
| 1903 |
+
"learning_rate": 5e-05,
|
| 1904 |
+
"loss": 0.4025,
|
| 1905 |
+
"num_tokens": 733557450.0,
|
| 1906 |
+
"step": 237
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 2.2676224611708484,
|
| 1910 |
+
"grad_norm": 0.20786051452159882,
|
| 1911 |
+
"learning_rate": 5e-05,
|
| 1912 |
+
"loss": 0.4005,
|
| 1913 |
+
"num_tokens": 736681750.0,
|
| 1914 |
+
"step": 238
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 2.2771804062126644,
|
| 1918 |
+
"grad_norm": 0.33719518780708313,
|
| 1919 |
+
"learning_rate": 5e-05,
|
| 1920 |
+
"loss": 0.393,
|
| 1921 |
+
"num_tokens": 739789232.0,
|
| 1922 |
+
"step": 239
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 2.2867383512544803,
|
| 1926 |
+
"grad_norm": 0.17687377333641052,
|
| 1927 |
+
"learning_rate": 5e-05,
|
| 1928 |
+
"loss": 0.4038,
|
| 1929 |
+
"num_tokens": 742891601.0,
|
| 1930 |
+
"step": 240
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 2.2962962962962963,
|
| 1934 |
+
"grad_norm": 0.342306524515152,
|
| 1935 |
+
"learning_rate": 5e-05,
|
| 1936 |
+
"loss": 0.4019,
|
| 1937 |
+
"num_tokens": 746006954.0,
|
| 1938 |
+
"step": 241
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 2.3058542413381122,
|
| 1942 |
+
"grad_norm": 0.16845592856407166,
|
| 1943 |
+
"learning_rate": 5e-05,
|
| 1944 |
+
"loss": 0.403,
|
| 1945 |
+
"num_tokens": 749147037.0,
|
| 1946 |
+
"step": 242
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 2.315412186379928,
|
| 1950 |
+
"grad_norm": 0.32256966829299927,
|
| 1951 |
+
"learning_rate": 5e-05,
|
| 1952 |
+
"loss": 0.4034,
|
| 1953 |
+
"num_tokens": 752229629.0,
|
| 1954 |
+
"step": 243
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 2.324970131421744,
|
| 1958 |
+
"grad_norm": 0.20878097414970398,
|
| 1959 |
+
"learning_rate": 5e-05,
|
| 1960 |
+
"loss": 0.4026,
|
| 1961 |
+
"num_tokens": 755278777.0,
|
| 1962 |
+
"step": 244
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.3345280764635605,
|
| 1966 |
+
"grad_norm": 0.2807317078113556,
|
| 1967 |
+
"learning_rate": 5e-05,
|
| 1968 |
+
"loss": 0.406,
|
| 1969 |
+
"num_tokens": 758363222.0,
|
| 1970 |
+
"step": 245
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 2.3440860215053765,
|
| 1974 |
+
"grad_norm": 0.2278064638376236,
|
| 1975 |
+
"learning_rate": 5e-05,
|
| 1976 |
+
"loss": 0.401,
|
| 1977 |
+
"num_tokens": 761474147.0,
|
| 1978 |
+
"step": 246
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 2.3536439665471924,
|
| 1982 |
+
"grad_norm": 0.2612324357032776,
|
| 1983 |
+
"learning_rate": 5e-05,
|
| 1984 |
+
"loss": 0.395,
|
| 1985 |
+
"num_tokens": 764551769.0,
|
| 1986 |
+
"step": 247
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 2.3632019115890084,
|
| 1990 |
+
"grad_norm": 0.23537400364875793,
|
| 1991 |
+
"learning_rate": 5e-05,
|
| 1992 |
+
"loss": 0.3948,
|
| 1993 |
+
"num_tokens": 767614488.0,
|
| 1994 |
+
"step": 248
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 2.3727598566308243,
|
| 1998 |
+
"grad_norm": 0.21469944715499878,
|
| 1999 |
+
"learning_rate": 5e-05,
|
| 2000 |
+
"loss": 0.403,
|
| 2001 |
+
"num_tokens": 770735177.0,
|
| 2002 |
+
"step": 249
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 2.3823178016726403,
|
| 2006 |
+
"grad_norm": 0.2674992084503174,
|
| 2007 |
+
"learning_rate": 5e-05,
|
| 2008 |
+
"loss": 0.3933,
|
| 2009 |
+
"num_tokens": 773851787.0,
|
| 2010 |
+
"step": 250
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 2.3918757467144562,
|
| 2014 |
+
"grad_norm": 0.18953253328800201,
|
| 2015 |
+
"learning_rate": 5e-05,
|
| 2016 |
+
"loss": 0.3994,
|
| 2017 |
+
"num_tokens": 776935088.0,
|
| 2018 |
+
"step": 251
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 2.4014336917562726,
|
| 2022 |
+
"grad_norm": 0.2478407323360443,
|
| 2023 |
+
"learning_rate": 5e-05,
|
| 2024 |
+
"loss": 0.4052,
|
| 2025 |
+
"num_tokens": 780049122.0,
|
| 2026 |
+
"step": 252
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 2.4109916367980886,
|
| 2030 |
+
"grad_norm": 0.195692777633667,
|
| 2031 |
+
"learning_rate": 5e-05,
|
| 2032 |
+
"loss": 0.4084,
|
| 2033 |
+
"num_tokens": 783178312.0,
|
| 2034 |
+
"step": 253
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 2.4205495818399045,
|
| 2038 |
+
"grad_norm": 0.20106397569179535,
|
| 2039 |
+
"learning_rate": 5e-05,
|
| 2040 |
+
"loss": 0.4022,
|
| 2041 |
+
"num_tokens": 786281027.0,
|
| 2042 |
+
"step": 254
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 2.4301075268817205,
|
| 2046 |
+
"grad_norm": 0.19922764599323273,
|
| 2047 |
+
"learning_rate": 5e-05,
|
| 2048 |
+
"loss": 0.4039,
|
| 2049 |
+
"num_tokens": 789414028.0,
|
| 2050 |
+
"step": 255
|
| 2051 |
+
},
|
| 2052 |
+
{
|
| 2053 |
+
"epoch": 2.4396654719235364,
|
| 2054 |
+
"grad_norm": 0.2001892626285553,
|
| 2055 |
+
"learning_rate": 5e-05,
|
| 2056 |
+
"loss": 0.4024,
|
| 2057 |
+
"num_tokens": 792520500.0,
|
| 2058 |
+
"step": 256
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 2.4492234169653524,
|
| 2062 |
+
"grad_norm": 0.22683829069137573,
|
| 2063 |
+
"learning_rate": 5e-05,
|
| 2064 |
+
"loss": 0.4022,
|
| 2065 |
+
"num_tokens": 795635985.0,
|
| 2066 |
+
"step": 257
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 2.4587813620071683,
|
| 2070 |
+
"grad_norm": 0.20693090558052063,
|
| 2071 |
+
"learning_rate": 5e-05,
|
| 2072 |
+
"loss": 0.3995,
|
| 2073 |
+
"num_tokens": 798748918.0,
|
| 2074 |
+
"step": 258
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 2.4683393070489843,
|
| 2078 |
+
"grad_norm": 0.22366446256637573,
|
| 2079 |
+
"learning_rate": 5e-05,
|
| 2080 |
+
"loss": 0.3988,
|
| 2081 |
+
"num_tokens": 801831194.0,
|
| 2082 |
+
"step": 259
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 2.4778972520908003,
|
| 2086 |
+
"grad_norm": 0.21484145522117615,
|
| 2087 |
+
"learning_rate": 5e-05,
|
| 2088 |
+
"loss": 0.3991,
|
| 2089 |
+
"num_tokens": 804928214.0,
|
| 2090 |
+
"step": 260
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 2.4874551971326166,
|
| 2094 |
+
"grad_norm": 0.22603361308574677,
|
| 2095 |
+
"learning_rate": 5e-05,
|
| 2096 |
+
"loss": 0.4033,
|
| 2097 |
+
"num_tokens": 807987120.0,
|
| 2098 |
+
"step": 261
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 2.4970131421744326,
|
| 2102 |
+
"grad_norm": 0.23147359490394592,
|
| 2103 |
+
"learning_rate": 5e-05,
|
| 2104 |
+
"loss": 0.4022,
|
| 2105 |
+
"num_tokens": 811098180.0,
|
| 2106 |
+
"step": 262
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 2.5065710872162486,
|
| 2110 |
+
"grad_norm": 0.1985515058040619,
|
| 2111 |
+
"learning_rate": 5e-05,
|
| 2112 |
+
"loss": 0.4022,
|
| 2113 |
+
"num_tokens": 814198566.0,
|
| 2114 |
+
"step": 263
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 2.5161290322580645,
|
| 2118 |
+
"grad_norm": 0.20071357488632202,
|
| 2119 |
+
"learning_rate": 5e-05,
|
| 2120 |
+
"loss": 0.4034,
|
| 2121 |
+
"num_tokens": 817257072.0,
|
| 2122 |
+
"step": 264
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 2.5256869772998805,
|
| 2126 |
+
"grad_norm": 0.25529035925865173,
|
| 2127 |
+
"learning_rate": 5e-05,
|
| 2128 |
+
"loss": 0.4067,
|
| 2129 |
+
"num_tokens": 820406825.0,
|
| 2130 |
+
"step": 265
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 2.5352449223416964,
|
| 2134 |
+
"grad_norm": 0.20370543003082275,
|
| 2135 |
+
"learning_rate": 5e-05,
|
| 2136 |
+
"loss": 0.4008,
|
| 2137 |
+
"num_tokens": 823512817.0,
|
| 2138 |
+
"step": 266
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 2.5448028673835124,
|
| 2142 |
+
"grad_norm": 0.20760004222393036,
|
| 2143 |
+
"learning_rate": 5e-05,
|
| 2144 |
+
"loss": 0.4034,
|
| 2145 |
+
"num_tokens": 826610050.0,
|
| 2146 |
+
"step": 267
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 2.5543608124253288,
|
| 2150 |
+
"grad_norm": 0.20357725024223328,
|
| 2151 |
+
"learning_rate": 5e-05,
|
| 2152 |
+
"loss": 0.398,
|
| 2153 |
+
"num_tokens": 829753903.0,
|
| 2154 |
+
"step": 268
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 2.5639187574671447,
|
| 2158 |
+
"grad_norm": 0.22291164100170135,
|
| 2159 |
+
"learning_rate": 5e-05,
|
| 2160 |
+
"loss": 0.3995,
|
| 2161 |
+
"num_tokens": 832868477.0,
|
| 2162 |
+
"step": 269
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 2.5734767025089607,
|
| 2166 |
+
"grad_norm": 0.1923750936985016,
|
| 2167 |
+
"learning_rate": 5e-05,
|
| 2168 |
+
"loss": 0.402,
|
| 2169 |
+
"num_tokens": 835937080.0,
|
| 2170 |
+
"step": 270
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 2.5830346475507766,
|
| 2174 |
+
"grad_norm": 0.22883036732673645,
|
| 2175 |
+
"learning_rate": 5e-05,
|
| 2176 |
+
"loss": 0.3997,
|
| 2177 |
+
"num_tokens": 839073075.0,
|
| 2178 |
+
"step": 271
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 2.5925925925925926,
|
| 2182 |
+
"grad_norm": 0.21349798142910004,
|
| 2183 |
+
"learning_rate": 5e-05,
|
| 2184 |
+
"loss": 0.403,
|
| 2185 |
+
"num_tokens": 842191197.0,
|
| 2186 |
+
"step": 272
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 2.6021505376344085,
|
| 2190 |
+
"grad_norm": 0.22202804684638977,
|
| 2191 |
+
"learning_rate": 5e-05,
|
| 2192 |
+
"loss": 0.4026,
|
| 2193 |
+
"num_tokens": 845276551.0,
|
| 2194 |
+
"step": 273
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 2.6117084826762245,
|
| 2198 |
+
"grad_norm": 0.20453181862831116,
|
| 2199 |
+
"learning_rate": 5e-05,
|
| 2200 |
+
"loss": 0.3986,
|
| 2201 |
+
"num_tokens": 848352451.0,
|
| 2202 |
+
"step": 274
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 2.621266427718041,
|
| 2206 |
+
"grad_norm": 0.21245239675045013,
|
| 2207 |
+
"learning_rate": 5e-05,
|
| 2208 |
+
"loss": 0.3993,
|
| 2209 |
+
"num_tokens": 851459757.0,
|
| 2210 |
+
"step": 275
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 2.6308243727598564,
|
| 2214 |
+
"grad_norm": 0.20784446597099304,
|
| 2215 |
+
"learning_rate": 5e-05,
|
| 2216 |
+
"loss": 0.4029,
|
| 2217 |
+
"num_tokens": 854632749.0,
|
| 2218 |
+
"step": 276
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 2.6403823178016728,
|
| 2222 |
+
"grad_norm": 0.20124511420726776,
|
| 2223 |
+
"learning_rate": 5e-05,
|
| 2224 |
+
"loss": 0.4018,
|
| 2225 |
+
"num_tokens": 857704440.0,
|
| 2226 |
+
"step": 277
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 2.6499402628434887,
|
| 2230 |
+
"grad_norm": 0.19951878488063812,
|
| 2231 |
+
"learning_rate": 5e-05,
|
| 2232 |
+
"loss": 0.4096,
|
| 2233 |
+
"num_tokens": 860825716.0,
|
| 2234 |
+
"step": 278
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 2.6594982078853047,
|
| 2238 |
+
"grad_norm": 0.19267506897449493,
|
| 2239 |
+
"learning_rate": 5e-05,
|
| 2240 |
+
"loss": 0.3974,
|
| 2241 |
+
"num_tokens": 863922145.0,
|
| 2242 |
+
"step": 279
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 2.6690561529271206,
|
| 2246 |
+
"grad_norm": 0.21417343616485596,
|
| 2247 |
+
"learning_rate": 5e-05,
|
| 2248 |
+
"loss": 0.4002,
|
| 2249 |
+
"num_tokens": 866991544.0,
|
| 2250 |
+
"step": 280
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 2.6786140979689366,
|
| 2254 |
+
"grad_norm": 0.21816791594028473,
|
| 2255 |
+
"learning_rate": 5e-05,
|
| 2256 |
+
"loss": 0.3997,
|
| 2257 |
+
"num_tokens": 870069694.0,
|
| 2258 |
+
"step": 281
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 2.688172043010753,
|
| 2262 |
+
"grad_norm": 0.21573670208454132,
|
| 2263 |
+
"learning_rate": 5e-05,
|
| 2264 |
+
"loss": 0.4064,
|
| 2265 |
+
"num_tokens": 873180617.0,
|
| 2266 |
+
"step": 282
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 2.6977299880525685,
|
| 2270 |
+
"grad_norm": 0.18886543810367584,
|
| 2271 |
+
"learning_rate": 5e-05,
|
| 2272 |
+
"loss": 0.3971,
|
| 2273 |
+
"num_tokens": 876262507.0,
|
| 2274 |
+
"step": 283
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 2.707287933094385,
|
| 2278 |
+
"grad_norm": 0.18656791746616364,
|
| 2279 |
+
"learning_rate": 5e-05,
|
| 2280 |
+
"loss": 0.4038,
|
| 2281 |
+
"num_tokens": 879380128.0,
|
| 2282 |
+
"step": 284
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 2.716845878136201,
|
| 2286 |
+
"grad_norm": 0.197296142578125,
|
| 2287 |
+
"learning_rate": 5e-05,
|
| 2288 |
+
"loss": 0.3986,
|
| 2289 |
+
"num_tokens": 882473178.0,
|
| 2290 |
+
"step": 285
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 2.7264038231780168,
|
| 2294 |
+
"grad_norm": 0.23007364571094513,
|
| 2295 |
+
"learning_rate": 5e-05,
|
| 2296 |
+
"loss": 0.4036,
|
| 2297 |
+
"num_tokens": 885565233.0,
|
| 2298 |
+
"step": 286
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 2.7359617682198327,
|
| 2302 |
+
"grad_norm": 0.19327418506145477,
|
| 2303 |
+
"learning_rate": 5e-05,
|
| 2304 |
+
"loss": 0.4134,
|
| 2305 |
+
"num_tokens": 888688634.0,
|
| 2306 |
+
"step": 287
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 2.7455197132616487,
|
| 2310 |
+
"grad_norm": 0.1903911530971527,
|
| 2311 |
+
"learning_rate": 5e-05,
|
| 2312 |
+
"loss": 0.3996,
|
| 2313 |
+
"num_tokens": 891794908.0,
|
| 2314 |
+
"step": 288
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 2.7550776583034646,
|
| 2318 |
+
"grad_norm": 0.18659088015556335,
|
| 2319 |
+
"learning_rate": 5e-05,
|
| 2320 |
+
"loss": 0.4044,
|
| 2321 |
+
"num_tokens": 894950413.0,
|
| 2322 |
+
"step": 289
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 2.7646356033452806,
|
| 2326 |
+
"grad_norm": 0.22853848338127136,
|
| 2327 |
+
"learning_rate": 5e-05,
|
| 2328 |
+
"loss": 0.4085,
|
| 2329 |
+
"num_tokens": 898028493.0,
|
| 2330 |
+
"step": 290
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 2.774193548387097,
|
| 2334 |
+
"grad_norm": 0.2249801754951477,
|
| 2335 |
+
"learning_rate": 5e-05,
|
| 2336 |
+
"loss": 0.4063,
|
| 2337 |
+
"num_tokens": 901215396.0,
|
| 2338 |
+
"step": 291
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 2.783751493428913,
|
| 2342 |
+
"grad_norm": 0.2070578783750534,
|
| 2343 |
+
"learning_rate": 5e-05,
|
| 2344 |
+
"loss": 0.3988,
|
| 2345 |
+
"num_tokens": 904344510.0,
|
| 2346 |
+
"step": 292
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 2.793309438470729,
|
| 2350 |
+
"grad_norm": 0.1973543018102646,
|
| 2351 |
+
"learning_rate": 5e-05,
|
| 2352 |
+
"loss": 0.4042,
|
| 2353 |
+
"num_tokens": 907446044.0,
|
| 2354 |
+
"step": 293
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 2.802867383512545,
|
| 2358 |
+
"grad_norm": 0.18739643692970276,
|
| 2359 |
+
"learning_rate": 5e-05,
|
| 2360 |
+
"loss": 0.4118,
|
| 2361 |
+
"num_tokens": 910525112.0,
|
| 2362 |
+
"step": 294
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 2.812425328554361,
|
| 2366 |
+
"grad_norm": 0.18655946850776672,
|
| 2367 |
+
"learning_rate": 5e-05,
|
| 2368 |
+
"loss": 0.3976,
|
| 2369 |
+
"num_tokens": 913646064.0,
|
| 2370 |
+
"step": 295
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 2.8219832735961767,
|
| 2374 |
+
"grad_norm": 0.19254498183727264,
|
| 2375 |
+
"learning_rate": 5e-05,
|
| 2376 |
+
"loss": 0.4044,
|
| 2377 |
+
"num_tokens": 916750615.0,
|
| 2378 |
+
"step": 296
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 2.8315412186379927,
|
| 2382 |
+
"grad_norm": 0.20286458730697632,
|
| 2383 |
+
"learning_rate": 5e-05,
|
| 2384 |
+
"loss": 0.4026,
|
| 2385 |
+
"num_tokens": 919815959.0,
|
| 2386 |
+
"step": 297
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 2.841099163679809,
|
| 2390 |
+
"grad_norm": 0.2116730809211731,
|
| 2391 |
+
"learning_rate": 5e-05,
|
| 2392 |
+
"loss": 0.3999,
|
| 2393 |
+
"num_tokens": 922854402.0,
|
| 2394 |
+
"step": 298
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 2.8506571087216246,
|
| 2398 |
+
"grad_norm": 0.22266528010368347,
|
| 2399 |
+
"learning_rate": 5e-05,
|
| 2400 |
+
"loss": 0.4038,
|
| 2401 |
+
"num_tokens": 925926441.0,
|
| 2402 |
+
"step": 299
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 2.860215053763441,
|
| 2406 |
+
"grad_norm": 0.18784675002098083,
|
| 2407 |
+
"learning_rate": 5e-05,
|
| 2408 |
+
"loss": 0.4046,
|
| 2409 |
+
"num_tokens": 929091816.0,
|
| 2410 |
+
"step": 300
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 2.869772998805257,
|
| 2414 |
+
"grad_norm": 0.18861991167068481,
|
| 2415 |
+
"learning_rate": 5e-05,
|
| 2416 |
+
"loss": 0.4058,
|
| 2417 |
+
"num_tokens": 932245034.0,
|
| 2418 |
+
"step": 301
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 2.879330943847073,
|
| 2422 |
+
"grad_norm": 0.18745040893554688,
|
| 2423 |
+
"learning_rate": 5e-05,
|
| 2424 |
+
"loss": 0.4053,
|
| 2425 |
+
"num_tokens": 935337789.0,
|
| 2426 |
+
"step": 302
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 2.888888888888889,
|
| 2430 |
+
"grad_norm": 0.21158824861049652,
|
| 2431 |
+
"learning_rate": 5e-05,
|
| 2432 |
+
"loss": 0.398,
|
| 2433 |
+
"num_tokens": 938469999.0,
|
| 2434 |
+
"step": 303
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 2.898446833930705,
|
| 2438 |
+
"grad_norm": 0.21590955555438995,
|
| 2439 |
+
"learning_rate": 5e-05,
|
| 2440 |
+
"loss": 0.4024,
|
| 2441 |
+
"num_tokens": 941548793.0,
|
| 2442 |
+
"step": 304
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 2.9080047789725207,
|
| 2446 |
+
"grad_norm": 0.206715390086174,
|
| 2447 |
+
"learning_rate": 5e-05,
|
| 2448 |
+
"loss": 0.4048,
|
| 2449 |
+
"num_tokens": 944687830.0,
|
| 2450 |
+
"step": 305
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 2.9175627240143367,
|
| 2454 |
+
"grad_norm": 0.2118476778268814,
|
| 2455 |
+
"learning_rate": 5e-05,
|
| 2456 |
+
"loss": 0.4093,
|
| 2457 |
+
"num_tokens": 947789759.0,
|
| 2458 |
+
"step": 306
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 2.927120669056153,
|
| 2462 |
+
"grad_norm": 0.18484579026699066,
|
| 2463 |
+
"learning_rate": 5e-05,
|
| 2464 |
+
"loss": 0.3981,
|
| 2465 |
+
"num_tokens": 950899733.0,
|
| 2466 |
+
"step": 307
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 2.936678614097969,
|
| 2470 |
+
"grad_norm": 0.17388074100017548,
|
| 2471 |
+
"learning_rate": 5e-05,
|
| 2472 |
+
"loss": 0.4065,
|
| 2473 |
+
"num_tokens": 954039453.0,
|
| 2474 |
+
"step": 308
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 2.946236559139785,
|
| 2478 |
+
"grad_norm": 0.16568942368030548,
|
| 2479 |
+
"learning_rate": 5e-05,
|
| 2480 |
+
"loss": 0.3993,
|
| 2481 |
+
"num_tokens": 957164986.0,
|
| 2482 |
+
"step": 309
|
| 2483 |
+
},
|
| 2484 |
+
{
|
| 2485 |
+
"epoch": 2.955794504181601,
|
| 2486 |
+
"grad_norm": 0.17550726234912872,
|
| 2487 |
+
"learning_rate": 5e-05,
|
| 2488 |
+
"loss": 0.4014,
|
| 2489 |
+
"num_tokens": 960246209.0,
|
| 2490 |
+
"step": 310
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 2.965352449223417,
|
| 2494 |
+
"grad_norm": 0.1897525042295456,
|
| 2495 |
+
"learning_rate": 5e-05,
|
| 2496 |
+
"loss": 0.3993,
|
| 2497 |
+
"num_tokens": 963312993.0,
|
| 2498 |
+
"step": 311
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 2.974910394265233,
|
| 2502 |
+
"grad_norm": 0.1897035390138626,
|
| 2503 |
+
"learning_rate": 5e-05,
|
| 2504 |
+
"loss": 0.3997,
|
| 2505 |
+
"num_tokens": 966396496.0,
|
| 2506 |
+
"step": 312
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 2.974910394265233,
|
| 2510 |
+
"step": 312,
|
| 2511 |
+
"total_flos": 3.993677328755458e+19,
|
| 2512 |
+
"train_loss": 0.4935996499963296,
|
| 2513 |
+
"train_runtime": 22258.2821,
|
| 2514 |
+
"train_samples_per_second": 12.633,
|
| 2515 |
+
"train_steps_per_second": 0.014
|
| 2516 |
+
}
|
| 2517 |
+
],
|
| 2518 |
+
"logging_steps": 1,
|
| 2519 |
+
"max_steps": 312,
|
| 2520 |
+
"num_input_tokens_seen": 0,
|
| 2521 |
+
"num_train_epochs": 3,
|
| 2522 |
+
"save_steps": 16,
|
| 2523 |
+
"stateful_callbacks": {
|
| 2524 |
+
"TrainerControl": {
|
| 2525 |
+
"args": {
|
| 2526 |
+
"should_epoch_stop": false,
|
| 2527 |
+
"should_evaluate": false,
|
| 2528 |
+
"should_log": false,
|
| 2529 |
+
"should_save": true,
|
| 2530 |
+
"should_training_stop": true
|
| 2531 |
+
},
|
| 2532 |
+
"attributes": {}
|
| 2533 |
+
}
|
| 2534 |
+
},
|
| 2535 |
+
"total_flos": 3.993677328755458e+19,
|
| 2536 |
+
"train_batch_size": 16,
|
| 2537 |
+
"trial_name": null,
|
| 2538 |
+
"trial_params": null
|
| 2539 |
+
}
|