Model save
Browse files- README.md +67 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1176 -0
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-7b
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: gemma-7b-orpo-noisy-5e-5
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-orpo-noisy-5e-5
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b).
|
| 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="silviasapora/gemma-7b-orpo-noisy-5e-5", 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/silvias/huggingface/runs/nlvt2go3)
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
This model was trained with ORPO, a method introduced in [ORPO: Monolithic Preference Optimization without Reference Model](https://huggingface.co/papers/2403.07691).
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.13.0
|
| 38 |
+
- Transformers: 4.46.1
|
| 39 |
+
- Pytorch: 2.4.0
|
| 40 |
+
- Datasets: 3.1.0
|
| 41 |
+
- Tokenizers: 0.20.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite ORPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{hong2024orpo,
|
| 49 |
+
title = {{ORPO: Monolithic Preference Optimization without Reference Model}},
|
| 50 |
+
author = {Jiwoo Hong and Noah Lee and James Thorne},
|
| 51 |
+
year = 2024,
|
| 52 |
+
eprint = {arXiv:2403.07691}
|
| 53 |
+
}
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
Cite TRL as:
|
| 57 |
+
|
| 58 |
+
```bibtex
|
| 59 |
+
@misc{vonwerra2022trl,
|
| 60 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 61 |
+
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 Gallouédec},
|
| 62 |
+
year = 2020,
|
| 63 |
+
journal = {GitHub repository},
|
| 64 |
+
publisher = {GitHub},
|
| 65 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 66 |
+
}
|
| 67 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.986666666666667,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 50.88312741234189,
|
| 5 |
+
"train_runtime": 6896.118,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.936,
|
| 8 |
+
"train_steps_per_second": 0.046
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.986666666666667,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 50.88312741234189,
|
| 5 |
+
"train_runtime": 6896.118,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.936,
|
| 8 |
+
"train_steps_per_second": 0.046
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.986666666666667,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 315,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.047407407407407405,
|
| 13 |
+
"grad_norm": 604.0,
|
| 14 |
+
"learning_rate": 7.8125e-06,
|
| 15 |
+
"log_odds_chosen": 4.997040271759033,
|
| 16 |
+
"log_odds_ratio": -9.76394271850586,
|
| 17 |
+
"logits/chosen": 138.64193725585938,
|
| 18 |
+
"logits/rejected": 152.19479370117188,
|
| 19 |
+
"logps/chosen": -20.550189971923828,
|
| 20 |
+
"logps/rejected": -25.547183990478516,
|
| 21 |
+
"loss": 284.5915,
|
| 22 |
+
"nll_loss": 8.063753128051758,
|
| 23 |
+
"rewards/accuracies": 0.543749988079071,
|
| 24 |
+
"rewards/chosen": -1.0275095701217651,
|
| 25 |
+
"rewards/margins": 0.24984964728355408,
|
| 26 |
+
"rewards/rejected": -1.2773592472076416,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 350.0,
|
| 32 |
+
"learning_rate": 1.5625e-05,
|
| 33 |
+
"log_odds_chosen": 2.356145143508911,
|
| 34 |
+
"log_odds_ratio": -7.16825008392334,
|
| 35 |
+
"logits/chosen": 131.21401977539062,
|
| 36 |
+
"logits/rejected": 160.0535430908203,
|
| 37 |
+
"logps/chosen": -15.259765625,
|
| 38 |
+
"logps/rejected": -17.6148738861084,
|
| 39 |
+
"loss": 251.6382,
|
| 40 |
+
"nll_loss": 6.878444671630859,
|
| 41 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 42 |
+
"rewards/chosen": -0.7629883289337158,
|
| 43 |
+
"rewards/margins": 0.11775543540716171,
|
| 44 |
+
"rewards/rejected": -0.880743682384491,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 314.0,
|
| 50 |
+
"learning_rate": 2.34375e-05,
|
| 51 |
+
"log_odds_chosen": 4.337428092956543,
|
| 52 |
+
"log_odds_ratio": -7.543389797210693,
|
| 53 |
+
"logits/chosen": 123.3839340209961,
|
| 54 |
+
"logits/rejected": 150.43328857421875,
|
| 55 |
+
"logps/chosen": -19.47274398803711,
|
| 56 |
+
"logps/rejected": -23.80842399597168,
|
| 57 |
+
"loss": 257.0097,
|
| 58 |
+
"nll_loss": 8.022147178649902,
|
| 59 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 60 |
+
"rewards/chosen": -0.9736372828483582,
|
| 61 |
+
"rewards/margins": 0.2167840451002121,
|
| 62 |
+
"rewards/rejected": -1.190421223640442,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 1352.0,
|
| 68 |
+
"learning_rate": 3.125e-05,
|
| 69 |
+
"log_odds_chosen": -0.34767332673072815,
|
| 70 |
+
"log_odds_ratio": -5.567352771759033,
|
| 71 |
+
"logits/chosen": 149.3828125,
|
| 72 |
+
"logits/rejected": 153.21987915039062,
|
| 73 |
+
"logps/chosen": -13.090100288391113,
|
| 74 |
+
"logps/rejected": -12.743819236755371,
|
| 75 |
+
"loss": 207.4182,
|
| 76 |
+
"nll_loss": 6.128692150115967,
|
| 77 |
+
"rewards/accuracies": 0.543749988079071,
|
| 78 |
+
"rewards/chosen": -0.6545050740242004,
|
| 79 |
+
"rewards/margins": -0.01731405407190323,
|
| 80 |
+
"rewards/rejected": -0.6371909976005554,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 294.0,
|
| 86 |
+
"learning_rate": 3.90625e-05,
|
| 87 |
+
"log_odds_chosen": 1.0955836772918701,
|
| 88 |
+
"log_odds_ratio": -1.4455451965332031,
|
| 89 |
+
"logits/chosen": 165.73220825195312,
|
| 90 |
+
"logits/rejected": 174.95925903320312,
|
| 91 |
+
"logps/chosen": -3.5200371742248535,
|
| 92 |
+
"logps/rejected": -4.602563858032227,
|
| 93 |
+
"loss": 75.9138,
|
| 94 |
+
"nll_loss": 2.368574380874634,
|
| 95 |
+
"rewards/accuracies": 0.53125,
|
| 96 |
+
"rewards/chosen": -0.17600186169147491,
|
| 97 |
+
"rewards/margins": 0.05412634462118149,
|
| 98 |
+
"rewards/rejected": -0.230128213763237,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 153.0,
|
| 104 |
+
"learning_rate": 4.6875e-05,
|
| 105 |
+
"log_odds_chosen": 0.15411178767681122,
|
| 106 |
+
"log_odds_ratio": -0.7986702919006348,
|
| 107 |
+
"logits/chosen": 197.4137725830078,
|
| 108 |
+
"logits/rejected": 219.61807250976562,
|
| 109 |
+
"logps/chosen": -1.626283884048462,
|
| 110 |
+
"logps/rejected": -1.7547603845596313,
|
| 111 |
+
"loss": 62.0478,
|
| 112 |
+
"nll_loss": 1.865581750869751,
|
| 113 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 114 |
+
"rewards/chosen": -0.08131420612335205,
|
| 115 |
+
"rewards/margins": 0.006423823535442352,
|
| 116 |
+
"rewards/rejected": -0.0877380222082138,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 166.0,
|
| 122 |
+
"learning_rate": 4.998613757348784e-05,
|
| 123 |
+
"log_odds_chosen": 0.004070779774338007,
|
| 124 |
+
"log_odds_ratio": -0.8579562902450562,
|
| 125 |
+
"logits/chosen": 234.51419067382812,
|
| 126 |
+
"logits/rejected": 230.0107421875,
|
| 127 |
+
"logps/chosen": -1.581650972366333,
|
| 128 |
+
"logps/rejected": -1.591902494430542,
|
| 129 |
+
"loss": 57.0513,
|
| 130 |
+
"nll_loss": 1.7507221698760986,
|
| 131 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 132 |
+
"rewards/chosen": -0.07908254116773605,
|
| 133 |
+
"rewards/margins": 0.0005125798052176833,
|
| 134 |
+
"rewards/rejected": -0.07959512621164322,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 161.0,
|
| 140 |
+
"learning_rate": 4.990147841143462e-05,
|
| 141 |
+
"log_odds_chosen": 0.2214292585849762,
|
| 142 |
+
"log_odds_ratio": -0.6974400281906128,
|
| 143 |
+
"logits/chosen": 228.30813598632812,
|
| 144 |
+
"logits/rejected": 230.08468627929688,
|
| 145 |
+
"logps/chosen": -1.3625082969665527,
|
| 146 |
+
"logps/rejected": -1.5368319749832153,
|
| 147 |
+
"loss": 54.0731,
|
| 148 |
+
"nll_loss": 1.6455962657928467,
|
| 149 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 150 |
+
"rewards/chosen": -0.0681254118680954,
|
| 151 |
+
"rewards/margins": 0.008716189302504063,
|
| 152 |
+
"rewards/rejected": -0.07684160768985748,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 78.0,
|
| 158 |
+
"learning_rate": 4.97401218720448e-05,
|
| 159 |
+
"log_odds_chosen": 0.18503662943840027,
|
| 160 |
+
"log_odds_ratio": -0.7793182730674744,
|
| 161 |
+
"logits/chosen": 209.5319366455078,
|
| 162 |
+
"logits/rejected": 208.72811889648438,
|
| 163 |
+
"logps/chosen": -1.389093041419983,
|
| 164 |
+
"logps/rejected": -1.543234944343567,
|
| 165 |
+
"loss": 52.3797,
|
| 166 |
+
"nll_loss": 1.6294572353363037,
|
| 167 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 168 |
+
"rewards/chosen": -0.06945464760065079,
|
| 169 |
+
"rewards/margins": 0.00770709291100502,
|
| 170 |
+
"rewards/rejected": -0.0771617442369461,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 68.0,
|
| 176 |
+
"learning_rate": 4.9502564938797946e-05,
|
| 177 |
+
"log_odds_chosen": 0.1809634566307068,
|
| 178 |
+
"log_odds_ratio": -0.7305103540420532,
|
| 179 |
+
"logits/chosen": 201.18109130859375,
|
| 180 |
+
"logits/rejected": 201.95797729492188,
|
| 181 |
+
"logps/chosen": -1.3455302715301514,
|
| 182 |
+
"logps/rejected": -1.4954122304916382,
|
| 183 |
+
"loss": 54.161,
|
| 184 |
+
"nll_loss": 1.6752201318740845,
|
| 185 |
+
"rewards/accuracies": 0.5625,
|
| 186 |
+
"rewards/chosen": -0.06727652251720428,
|
| 187 |
+
"rewards/margins": 0.007494096644222736,
|
| 188 |
+
"rewards/rejected": -0.07477061450481415,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 70.0,
|
| 194 |
+
"learning_rate": 4.918953929490768e-05,
|
| 195 |
+
"log_odds_chosen": 0.14337705075740814,
|
| 196 |
+
"log_odds_ratio": -0.7200931310653687,
|
| 197 |
+
"logits/chosen": 194.4894256591797,
|
| 198 |
+
"logits/rejected": 193.81051635742188,
|
| 199 |
+
"logps/chosen": -1.2621675729751587,
|
| 200 |
+
"logps/rejected": -1.3799371719360352,
|
| 201 |
+
"loss": 49.4217,
|
| 202 |
+
"nll_loss": 1.5453771352767944,
|
| 203 |
+
"rewards/accuracies": 0.5625,
|
| 204 |
+
"rewards/chosen": -0.06310837715864182,
|
| 205 |
+
"rewards/margins": 0.005888479296118021,
|
| 206 |
+
"rewards/rejected": -0.068996861577034,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 63.25,
|
| 212 |
+
"learning_rate": 4.88020090697132e-05,
|
| 213 |
+
"log_odds_chosen": 0.221818208694458,
|
| 214 |
+
"log_odds_ratio": -0.6779998540878296,
|
| 215 |
+
"logits/chosen": 194.46102905273438,
|
| 216 |
+
"logits/rejected": 190.25668334960938,
|
| 217 |
+
"logps/chosen": -1.1540250778198242,
|
| 218 |
+
"logps/rejected": -1.3141117095947266,
|
| 219 |
+
"loss": 47.4294,
|
| 220 |
+
"nll_loss": 1.4274753332138062,
|
| 221 |
+
"rewards/accuracies": 0.581250011920929,
|
| 222 |
+
"rewards/chosen": -0.05770125240087509,
|
| 223 |
+
"rewards/margins": 0.008004339411854744,
|
| 224 |
+
"rewards/rejected": -0.06570558995008469,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 83.0,
|
| 230 |
+
"learning_rate": 4.834116786912897e-05,
|
| 231 |
+
"log_odds_chosen": 0.1881437748670578,
|
| 232 |
+
"log_odds_ratio": -0.7192609906196594,
|
| 233 |
+
"logits/chosen": 202.4822235107422,
|
| 234 |
+
"logits/rejected": 204.4671173095703,
|
| 235 |
+
"logps/chosen": -1.2269448041915894,
|
| 236 |
+
"logps/rejected": -1.3740522861480713,
|
| 237 |
+
"loss": 47.3934,
|
| 238 |
+
"nll_loss": 1.461767315864563,
|
| 239 |
+
"rewards/accuracies": 0.53125,
|
| 240 |
+
"rewards/chosen": -0.06134723871946335,
|
| 241 |
+
"rewards/margins": 0.00735537800937891,
|
| 242 |
+
"rewards/rejected": -0.06870262324810028,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 114.0,
|
| 248 |
+
"learning_rate": 4.7808435099299045e-05,
|
| 249 |
+
"log_odds_chosen": 0.30721157789230347,
|
| 250 |
+
"log_odds_ratio": -0.663455069065094,
|
| 251 |
+
"logits/chosen": 202.92465209960938,
|
| 252 |
+
"logits/rejected": 196.5619354248047,
|
| 253 |
+
"logps/chosen": -1.1372950077056885,
|
| 254 |
+
"logps/rejected": -1.3888975381851196,
|
| 255 |
+
"loss": 46.1289,
|
| 256 |
+
"nll_loss": 1.4218207597732544,
|
| 257 |
+
"rewards/accuracies": 0.581250011920929,
|
| 258 |
+
"rewards/chosen": -0.056864749640226364,
|
| 259 |
+
"rewards/margins": 0.012580126523971558,
|
| 260 |
+
"rewards/rejected": -0.06944488734006882,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 294.0,
|
| 266 |
+
"learning_rate": 4.720545159477922e-05,
|
| 267 |
+
"log_odds_chosen": 0.3042272627353668,
|
| 268 |
+
"log_odds_ratio": -0.6515589952468872,
|
| 269 |
+
"logits/chosen": 203.1158447265625,
|
| 270 |
+
"logits/rejected": 205.7737274169922,
|
| 271 |
+
"logps/chosen": -1.100188136100769,
|
| 272 |
+
"logps/rejected": -1.3240959644317627,
|
| 273 |
+
"loss": 46.1877,
|
| 274 |
+
"nll_loss": 1.4243879318237305,
|
| 275 |
+
"rewards/accuracies": 0.637499988079071,
|
| 276 |
+
"rewards/chosen": -0.05500940605998039,
|
| 277 |
+
"rewards/margins": 0.011195378378033638,
|
| 278 |
+
"rewards/rejected": -0.06620479375123978,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 114.5,
|
| 284 |
+
"learning_rate": 4.653407456471222e-05,
|
| 285 |
+
"log_odds_chosen": 0.2768544554710388,
|
| 286 |
+
"log_odds_ratio": -0.6946061849594116,
|
| 287 |
+
"logits/chosen": 209.74020385742188,
|
| 288 |
+
"logits/rejected": 206.70449829101562,
|
| 289 |
+
"logps/chosen": -1.1427994966506958,
|
| 290 |
+
"logps/rejected": -1.3516017198562622,
|
| 291 |
+
"loss": 45.0629,
|
| 292 |
+
"nll_loss": 1.4209954738616943,
|
| 293 |
+
"rewards/accuracies": 0.59375,
|
| 294 |
+
"rewards/chosen": -0.05713997408747673,
|
| 295 |
+
"rewards/margins": 0.010440108366310596,
|
| 296 |
+
"rewards/rejected": -0.06758008152246475,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 72.0,
|
| 302 |
+
"learning_rate": 4.579637187256222e-05,
|
| 303 |
+
"log_odds_chosen": 0.2393846958875656,
|
| 304 |
+
"log_odds_ratio": -0.6808741688728333,
|
| 305 |
+
"logits/chosen": 204.6654815673828,
|
| 306 |
+
"logits/rejected": 199.04537963867188,
|
| 307 |
+
"logps/chosen": -1.0734455585479736,
|
| 308 |
+
"logps/rejected": -1.2461655139923096,
|
| 309 |
+
"loss": 44.4836,
|
| 310 |
+
"nll_loss": 1.3528298139572144,
|
| 311 |
+
"rewards/accuracies": 0.59375,
|
| 312 |
+
"rewards/chosen": -0.05367228388786316,
|
| 313 |
+
"rewards/margins": 0.00863600242882967,
|
| 314 |
+
"rewards/rejected": -0.062308281660079956,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 74.5,
|
| 320 |
+
"learning_rate": 4.499461566702685e-05,
|
| 321 |
+
"log_odds_chosen": 0.26840656995773315,
|
| 322 |
+
"log_odds_ratio": -0.660419762134552,
|
| 323 |
+
"logits/chosen": 203.23355102539062,
|
| 324 |
+
"logits/rejected": 202.61805725097656,
|
| 325 |
+
"logps/chosen": -1.0564864873886108,
|
| 326 |
+
"logps/rejected": -1.2361832857131958,
|
| 327 |
+
"loss": 45.7849,
|
| 328 |
+
"nll_loss": 1.381618857383728,
|
| 329 |
+
"rewards/accuracies": 0.625,
|
| 330 |
+
"rewards/chosen": -0.05282432958483696,
|
| 331 |
+
"rewards/margins": 0.008984836749732494,
|
| 332 |
+
"rewards/rejected": -0.06180916354060173,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 119.0,
|
| 338 |
+
"learning_rate": 4.413127538374411e-05,
|
| 339 |
+
"log_odds_chosen": 0.37480229139328003,
|
| 340 |
+
"log_odds_ratio": -0.5969882011413574,
|
| 341 |
+
"logits/chosen": 198.45016479492188,
|
| 342 |
+
"logits/rejected": 195.573974609375,
|
| 343 |
+
"logps/chosen": -0.9739300608634949,
|
| 344 |
+
"logps/rejected": -1.2221287488937378,
|
| 345 |
+
"loss": 44.2264,
|
| 346 |
+
"nll_loss": 1.2837311029434204,
|
| 347 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 348 |
+
"rewards/chosen": -0.048696503043174744,
|
| 349 |
+
"rewards/margins": 0.012409940361976624,
|
| 350 |
+
"rewards/rejected": -0.06110644340515137,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 54.0,
|
| 356 |
+
"learning_rate": 4.320901013934887e-05,
|
| 357 |
+
"log_odds_chosen": 0.1513899266719818,
|
| 358 |
+
"log_odds_ratio": -0.7243643403053284,
|
| 359 |
+
"logits/chosen": 206.537109375,
|
| 360 |
+
"logits/rejected": 200.01821899414062,
|
| 361 |
+
"logps/chosen": -1.0972650051116943,
|
| 362 |
+
"logps/rejected": -1.22805917263031,
|
| 363 |
+
"loss": 44.4168,
|
| 364 |
+
"nll_loss": 1.393330693244934,
|
| 365 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 366 |
+
"rewards/chosen": -0.054863251745700836,
|
| 367 |
+
"rewards/margins": 0.006539707537740469,
|
| 368 |
+
"rewards/rejected": -0.06140295788645744,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 110.0,
|
| 374 |
+
"learning_rate": 4.223066054130568e-05,
|
| 375 |
+
"log_odds_chosen": 0.30877920985221863,
|
| 376 |
+
"log_odds_ratio": -0.6278973817825317,
|
| 377 |
+
"logits/chosen": 207.3557891845703,
|
| 378 |
+
"logits/rejected": 204.31600952148438,
|
| 379 |
+
"logps/chosen": -1.0028679370880127,
|
| 380 |
+
"logps/rejected": -1.2167755365371704,
|
| 381 |
+
"loss": 43.3043,
|
| 382 |
+
"nll_loss": 1.304164171218872,
|
| 383 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 384 |
+
"rewards/chosen": -0.05014339089393616,
|
| 385 |
+
"rewards/margins": 0.010695383884012699,
|
| 386 |
+
"rewards/rejected": -0.06083878129720688,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.0429629629629629,
|
| 391 |
+
"grad_norm": 56.25,
|
| 392 |
+
"learning_rate": 4.1199239938743797e-05,
|
| 393 |
+
"log_odds_chosen": 0.5618379712104797,
|
| 394 |
+
"log_odds_ratio": -0.5704431533813477,
|
| 395 |
+
"logits/chosen": 198.82461547851562,
|
| 396 |
+
"logits/rejected": 193.56935119628906,
|
| 397 |
+
"logps/chosen": -0.8706051707267761,
|
| 398 |
+
"logps/rejected": -1.2394962310791016,
|
| 399 |
+
"loss": 37.9309,
|
| 400 |
+
"nll_loss": 1.133053183555603,
|
| 401 |
+
"rewards/accuracies": 0.706250011920929,
|
| 402 |
+
"rewards/chosen": -0.043530263006687164,
|
| 403 |
+
"rewards/margins": 0.018444549292325974,
|
| 404 |
+
"rewards/rejected": -0.06197481229901314,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0903703703703704,
|
| 409 |
+
"grad_norm": 101.0,
|
| 410 |
+
"learning_rate": 4.0117925141242174e-05,
|
| 411 |
+
"log_odds_chosen": 0.7508487701416016,
|
| 412 |
+
"log_odds_ratio": -0.48277807235717773,
|
| 413 |
+
"logits/chosen": 194.4977569580078,
|
| 414 |
+
"logits/rejected": 192.06838989257812,
|
| 415 |
+
"logps/chosen": -0.8287516832351685,
|
| 416 |
+
"logps/rejected": -1.3157087564468384,
|
| 417 |
+
"loss": 36.5937,
|
| 418 |
+
"nll_loss": 1.1013622283935547,
|
| 419 |
+
"rewards/accuracies": 0.75,
|
| 420 |
+
"rewards/chosen": -0.04143758863210678,
|
| 421 |
+
"rewards/margins": 0.02434784732758999,
|
| 422 |
+
"rewards/rejected": -0.06578543782234192,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1377777777777778,
|
| 427 |
+
"grad_norm": 59.0,
|
| 428 |
+
"learning_rate": 3.899004663415084e-05,
|
| 429 |
+
"log_odds_chosen": 0.8108431696891785,
|
| 430 |
+
"log_odds_ratio": -0.4502410888671875,
|
| 431 |
+
"logits/chosen": 194.10855102539062,
|
| 432 |
+
"logits/rejected": 193.88357543945312,
|
| 433 |
+
"logps/chosen": -0.8010668754577637,
|
| 434 |
+
"logps/rejected": -1.2895253896713257,
|
| 435 |
+
"loss": 37.1465,
|
| 436 |
+
"nll_loss": 1.1015279293060303,
|
| 437 |
+
"rewards/accuracies": 0.800000011920929,
|
| 438 |
+
"rewards/chosen": -0.0400533489882946,
|
| 439 |
+
"rewards/margins": 0.02442292682826519,
|
| 440 |
+
"rewards/rejected": -0.06447627395391464,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1851851851851851,
|
| 445 |
+
"grad_norm": 76.0,
|
| 446 |
+
"learning_rate": 3.781907832058587e-05,
|
| 447 |
+
"log_odds_chosen": 0.6857287883758545,
|
| 448 |
+
"log_odds_ratio": -0.5154197216033936,
|
| 449 |
+
"logits/chosen": 192.4232635498047,
|
| 450 |
+
"logits/rejected": 189.86856079101562,
|
| 451 |
+
"logps/chosen": -0.8564508557319641,
|
| 452 |
+
"logps/rejected": -1.2979596853256226,
|
| 453 |
+
"loss": 34.4135,
|
| 454 |
+
"nll_loss": 1.067600965499878,
|
| 455 |
+
"rewards/accuracies": 0.737500011920929,
|
| 456 |
+
"rewards/chosen": -0.042822543531656265,
|
| 457 |
+
"rewards/margins": 0.02207544818520546,
|
| 458 |
+
"rewards/rejected": -0.06489799171686172,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2325925925925927,
|
| 463 |
+
"grad_norm": 49.75,
|
| 464 |
+
"learning_rate": 3.660862682169282e-05,
|
| 465 |
+
"log_odds_chosen": 0.5871692895889282,
|
| 466 |
+
"log_odds_ratio": -0.544813334941864,
|
| 467 |
+
"logits/chosen": 197.4970703125,
|
| 468 |
+
"logits/rejected": 195.34658813476562,
|
| 469 |
+
"logps/chosen": -0.8550790548324585,
|
| 470 |
+
"logps/rejected": -1.1946574449539185,
|
| 471 |
+
"loss": 36.1597,
|
| 472 |
+
"nll_loss": 1.1329249143600464,
|
| 473 |
+
"rewards/accuracies": 0.731249988079071,
|
| 474 |
+
"rewards/chosen": -0.04275395721197128,
|
| 475 |
+
"rewards/margins": 0.0169789157807827,
|
| 476 |
+
"rewards/rejected": -0.059732865542173386,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.28,
|
| 481 |
+
"grad_norm": 56.75,
|
| 482 |
+
"learning_rate": 3.5362420368134356e-05,
|
| 483 |
+
"log_odds_chosen": 0.696057915687561,
|
| 484 |
+
"log_odds_ratio": -0.500568151473999,
|
| 485 |
+
"logits/chosen": 191.46945190429688,
|
| 486 |
+
"logits/rejected": 193.9897918701172,
|
| 487 |
+
"logps/chosen": -0.7812957763671875,
|
| 488 |
+
"logps/rejected": -1.2018015384674072,
|
| 489 |
+
"loss": 35.7705,
|
| 490 |
+
"nll_loss": 1.0668926239013672,
|
| 491 |
+
"rewards/accuracies": 0.75,
|
| 492 |
+
"rewards/chosen": -0.039064787328243256,
|
| 493 |
+
"rewards/margins": 0.021025285124778748,
|
| 494 |
+
"rewards/rejected": -0.0600900761783123,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3274074074074074,
|
| 499 |
+
"grad_norm": 50.5,
|
| 500 |
+
"learning_rate": 3.408429731701635e-05,
|
| 501 |
+
"log_odds_chosen": 0.6391748189926147,
|
| 502 |
+
"log_odds_ratio": -0.5416852235794067,
|
| 503 |
+
"logits/chosen": 186.97299194335938,
|
| 504 |
+
"logits/rejected": 188.9523162841797,
|
| 505 |
+
"logps/chosen": -0.8818444013595581,
|
| 506 |
+
"logps/rejected": -1.2928279638290405,
|
| 507 |
+
"loss": 37.211,
|
| 508 |
+
"nll_loss": 1.1587209701538086,
|
| 509 |
+
"rewards/accuracies": 0.737500011920929,
|
| 510 |
+
"rewards/chosen": -0.044092223048210144,
|
| 511 |
+
"rewards/margins": 0.02054918184876442,
|
| 512 |
+
"rewards/rejected": -0.06464140117168427,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.374814814814815,
|
| 517 |
+
"grad_norm": 62.75,
|
| 518 |
+
"learning_rate": 3.2778194329621104e-05,
|
| 519 |
+
"log_odds_chosen": 0.6546341180801392,
|
| 520 |
+
"log_odds_ratio": -0.5061019062995911,
|
| 521 |
+
"logits/chosen": 193.48361206054688,
|
| 522 |
+
"logits/rejected": 189.97891235351562,
|
| 523 |
+
"logps/chosen": -0.8745110630989075,
|
| 524 |
+
"logps/rejected": -1.3096916675567627,
|
| 525 |
+
"loss": 36.6068,
|
| 526 |
+
"nll_loss": 1.1216226816177368,
|
| 527 |
+
"rewards/accuracies": 0.731249988079071,
|
| 528 |
+
"rewards/chosen": -0.043725550174713135,
|
| 529 |
+
"rewards/margins": 0.021759027615189552,
|
| 530 |
+
"rewards/rejected": -0.06548458337783813,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.4222222222222223,
|
| 535 |
+
"grad_norm": 57.5,
|
| 536 |
+
"learning_rate": 3.144813424636031e-05,
|
| 537 |
+
"log_odds_chosen": 0.6039055585861206,
|
| 538 |
+
"log_odds_ratio": -0.5407282114028931,
|
| 539 |
+
"logits/chosen": 190.10671997070312,
|
| 540 |
+
"logits/rejected": 191.7057342529297,
|
| 541 |
+
"logps/chosen": -0.8144019246101379,
|
| 542 |
+
"logps/rejected": -1.1773698329925537,
|
| 543 |
+
"loss": 35.5667,
|
| 544 |
+
"nll_loss": 1.0835626125335693,
|
| 545 |
+
"rewards/accuracies": 0.71875,
|
| 546 |
+
"rewards/chosen": -0.040720097720623016,
|
| 547 |
+
"rewards/margins": 0.01814839243888855,
|
| 548 |
+
"rewards/rejected": -0.058868490159511566,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4696296296296296,
|
| 553 |
+
"grad_norm": 43.25,
|
| 554 |
+
"learning_rate": 3.0098213696293542e-05,
|
| 555 |
+
"log_odds_chosen": 0.5702832937240601,
|
| 556 |
+
"log_odds_ratio": -0.5412538051605225,
|
| 557 |
+
"logits/chosen": 195.86476135253906,
|
| 558 |
+
"logits/rejected": 193.18934631347656,
|
| 559 |
+
"logps/chosen": -0.8274780511856079,
|
| 560 |
+
"logps/rejected": -1.1580636501312256,
|
| 561 |
+
"loss": 36.6038,
|
| 562 |
+
"nll_loss": 1.1131330728530884,
|
| 563 |
+
"rewards/accuracies": 0.71875,
|
| 564 |
+
"rewards/chosen": -0.041373901069164276,
|
| 565 |
+
"rewards/margins": 0.016529280692338943,
|
| 566 |
+
"rewards/rejected": -0.05790318176150322,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.5170370370370372,
|
| 571 |
+
"grad_norm": 72.0,
|
| 572 |
+
"learning_rate": 2.8732590479375165e-05,
|
| 573 |
+
"log_odds_chosen": 0.6357023119926453,
|
| 574 |
+
"log_odds_ratio": -0.5035272836685181,
|
| 575 |
+
"logits/chosen": 199.20773315429688,
|
| 576 |
+
"logits/rejected": 201.07427978515625,
|
| 577 |
+
"logps/chosen": -0.898838996887207,
|
| 578 |
+
"logps/rejected": -1.3042376041412354,
|
| 579 |
+
"loss": 37.2702,
|
| 580 |
+
"nll_loss": 1.169263482093811,
|
| 581 |
+
"rewards/accuracies": 0.762499988079071,
|
| 582 |
+
"rewards/chosen": -0.04494195431470871,
|
| 583 |
+
"rewards/margins": 0.020269926637411118,
|
| 584 |
+
"rewards/rejected": -0.06521187722682953,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5644444444444443,
|
| 589 |
+
"grad_norm": 63.0,
|
| 590 |
+
"learning_rate": 2.7355470760292956e-05,
|
| 591 |
+
"log_odds_chosen": 0.7361176013946533,
|
| 592 |
+
"log_odds_ratio": -0.49207085371017456,
|
| 593 |
+
"logits/chosen": 196.7345428466797,
|
| 594 |
+
"logits/rejected": 201.42710876464844,
|
| 595 |
+
"logps/chosen": -0.8479239344596863,
|
| 596 |
+
"logps/rejected": -1.30262291431427,
|
| 597 |
+
"loss": 35.9781,
|
| 598 |
+
"nll_loss": 1.0971934795379639,
|
| 599 |
+
"rewards/accuracies": 0.768750011920929,
|
| 600 |
+
"rewards/chosen": -0.042396195232868195,
|
| 601 |
+
"rewards/margins": 0.022734951227903366,
|
| 602 |
+
"rewards/rejected": -0.06513114273548126,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6118518518518519,
|
| 607 |
+
"grad_norm": 44.25,
|
| 608 |
+
"learning_rate": 2.597109611334169e-05,
|
| 609 |
+
"log_odds_chosen": 0.7843419313430786,
|
| 610 |
+
"log_odds_ratio": -0.47543781995773315,
|
| 611 |
+
"logits/chosen": 195.94033813476562,
|
| 612 |
+
"logits/rejected": 194.622314453125,
|
| 613 |
+
"logps/chosen": -0.8161409497261047,
|
| 614 |
+
"logps/rejected": -1.3115676641464233,
|
| 615 |
+
"loss": 35.8517,
|
| 616 |
+
"nll_loss": 1.0809214115142822,
|
| 617 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 618 |
+
"rewards/chosen": -0.040807049721479416,
|
| 619 |
+
"rewards/margins": 0.02477133832871914,
|
| 620 |
+
"rewards/rejected": -0.0655783861875534,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6592592592592592,
|
| 625 |
+
"grad_norm": 51.5,
|
| 626 |
+
"learning_rate": 2.458373045823404e-05,
|
| 627 |
+
"log_odds_chosen": 0.6568929553031921,
|
| 628 |
+
"log_odds_ratio": -0.5108085870742798,
|
| 629 |
+
"logits/chosen": 197.05380249023438,
|
| 630 |
+
"logits/rejected": 193.593994140625,
|
| 631 |
+
"logps/chosen": -0.8179122805595398,
|
| 632 |
+
"logps/rejected": -1.2255433797836304,
|
| 633 |
+
"loss": 35.4159,
|
| 634 |
+
"nll_loss": 1.1078870296478271,
|
| 635 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 636 |
+
"rewards/chosen": -0.04089561477303505,
|
| 637 |
+
"rewards/margins": 0.02038155123591423,
|
| 638 |
+
"rewards/rejected": -0.06127716228365898,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7066666666666666,
|
| 643 |
+
"grad_norm": 44.75,
|
| 644 |
+
"learning_rate": 2.3197646927086697e-05,
|
| 645 |
+
"log_odds_chosen": 0.5102953314781189,
|
| 646 |
+
"log_odds_ratio": -0.5745335817337036,
|
| 647 |
+
"logits/chosen": 189.71182250976562,
|
| 648 |
+
"logits/rejected": 189.39669799804688,
|
| 649 |
+
"logps/chosen": -0.8822305798530579,
|
| 650 |
+
"logps/rejected": -1.1973659992218018,
|
| 651 |
+
"loss": 36.8984,
|
| 652 |
+
"nll_loss": 1.1632763147354126,
|
| 653 |
+
"rewards/accuracies": 0.675000011920929,
|
| 654 |
+
"rewards/chosen": -0.04411153122782707,
|
| 655 |
+
"rewards/margins": 0.015756776556372643,
|
| 656 |
+
"rewards/rejected": -0.059868305921554565,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.7540740740740741,
|
| 661 |
+
"grad_norm": 34.75,
|
| 662 |
+
"learning_rate": 2.1817114703032176e-05,
|
| 663 |
+
"log_odds_chosen": 0.6035521030426025,
|
| 664 |
+
"log_odds_ratio": -0.5137989521026611,
|
| 665 |
+
"logits/chosen": 192.17518615722656,
|
| 666 |
+
"logits/rejected": 191.57754516601562,
|
| 667 |
+
"logps/chosen": -0.8429533839225769,
|
| 668 |
+
"logps/rejected": -1.228217601776123,
|
| 669 |
+
"loss": 35.5081,
|
| 670 |
+
"nll_loss": 1.0759392976760864,
|
| 671 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 672 |
+
"rewards/chosen": -0.042147669941186905,
|
| 673 |
+
"rewards/margins": 0.019263211637735367,
|
| 674 |
+
"rewards/rejected": -0.06141088157892227,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.8014814814814815,
|
| 679 |
+
"grad_norm": 58.25,
|
| 680 |
+
"learning_rate": 2.0446385870993467e-05,
|
| 681 |
+
"log_odds_chosen": 0.7391771078109741,
|
| 682 |
+
"log_odds_ratio": -0.4915809631347656,
|
| 683 |
+
"logits/chosen": 193.21046447753906,
|
| 684 |
+
"logits/rejected": 190.1485595703125,
|
| 685 |
+
"logps/chosen": -0.8688086271286011,
|
| 686 |
+
"logps/rejected": -1.3316614627838135,
|
| 687 |
+
"loss": 37.3014,
|
| 688 |
+
"nll_loss": 1.1229429244995117,
|
| 689 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 690 |
+
"rewards/chosen": -0.043440431356430054,
|
| 691 |
+
"rewards/margins": 0.023142646998167038,
|
| 692 |
+
"rewards/rejected": -0.06658308207988739,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8488888888888888,
|
| 697 |
+
"grad_norm": 45.25,
|
| 698 |
+
"learning_rate": 1.9089682321121834e-05,
|
| 699 |
+
"log_odds_chosen": 0.7794807553291321,
|
| 700 |
+
"log_odds_ratio": -0.47752276062965393,
|
| 701 |
+
"logits/chosen": 191.25637817382812,
|
| 702 |
+
"logits/rejected": 185.43508911132812,
|
| 703 |
+
"logps/chosen": -0.8267251253128052,
|
| 704 |
+
"logps/rejected": -1.3205393552780151,
|
| 705 |
+
"loss": 35.5594,
|
| 706 |
+
"nll_loss": 1.077235460281372,
|
| 707 |
+
"rewards/accuracies": 0.762499988079071,
|
| 708 |
+
"rewards/chosen": -0.04133625701069832,
|
| 709 |
+
"rewards/margins": 0.024690713733434677,
|
| 710 |
+
"rewards/rejected": -0.0660269632935524,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8962962962962964,
|
| 715 |
+
"grad_norm": 40.25,
|
| 716 |
+
"learning_rate": 1.775118274523545e-05,
|
| 717 |
+
"log_odds_chosen": 0.5731266140937805,
|
| 718 |
+
"log_odds_ratio": -0.5659371614456177,
|
| 719 |
+
"logits/chosen": 192.36257934570312,
|
| 720 |
+
"logits/rejected": 191.9702911376953,
|
| 721 |
+
"logps/chosen": -0.8584796786308289,
|
| 722 |
+
"logps/rejected": -1.2092317342758179,
|
| 723 |
+
"loss": 37.8225,
|
| 724 |
+
"nll_loss": 1.114450216293335,
|
| 725 |
+
"rewards/accuracies": 0.6875,
|
| 726 |
+
"rewards/chosen": -0.04292398318648338,
|
| 727 |
+
"rewards/margins": 0.01753760501742363,
|
| 728 |
+
"rewards/rejected": -0.06046159192919731,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9437037037037037,
|
| 733 |
+
"grad_norm": 47.0,
|
| 734 |
+
"learning_rate": 1.643500976631037e-05,
|
| 735 |
+
"log_odds_chosen": 0.6254440546035767,
|
| 736 |
+
"log_odds_ratio": -0.5245063304901123,
|
| 737 |
+
"logits/chosen": 195.681640625,
|
| 738 |
+
"logits/rejected": 188.51748657226562,
|
| 739 |
+
"logps/chosen": -0.8348779678344727,
|
| 740 |
+
"logps/rejected": -1.2449554204940796,
|
| 741 |
+
"loss": 33.9795,
|
| 742 |
+
"nll_loss": 1.04995858669281,
|
| 743 |
+
"rewards/accuracies": 0.71875,
|
| 744 |
+
"rewards/chosen": -0.04174389690160751,
|
| 745 |
+
"rewards/margins": 0.020503869280219078,
|
| 746 |
+
"rewards/rejected": -0.06224777549505234,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.991111111111111,
|
| 751 |
+
"grad_norm": 57.75,
|
| 752 |
+
"learning_rate": 1.514521724066537e-05,
|
| 753 |
+
"log_odds_chosen": 0.5750250816345215,
|
| 754 |
+
"log_odds_ratio": -0.5381630659103394,
|
| 755 |
+
"logits/chosen": 189.287353515625,
|
| 756 |
+
"logits/rejected": 191.6526641845703,
|
| 757 |
+
"logps/chosen": -0.7991558313369751,
|
| 758 |
+
"logps/rejected": -1.1348118782043457,
|
| 759 |
+
"loss": 34.3563,
|
| 760 |
+
"nll_loss": 1.0541894435882568,
|
| 761 |
+
"rewards/accuracies": 0.706250011920929,
|
| 762 |
+
"rewards/chosen": -0.039957791566848755,
|
| 763 |
+
"rewards/margins": 0.01678280346095562,
|
| 764 |
+
"rewards/rejected": -0.056740593165159225,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0385185185185186,
|
| 769 |
+
"grad_norm": 48.0,
|
| 770 |
+
"learning_rate": 1.3885777771950348e-05,
|
| 771 |
+
"log_odds_chosen": 1.051464319229126,
|
| 772 |
+
"log_odds_ratio": -0.3930102288722992,
|
| 773 |
+
"logits/chosen": 184.33993530273438,
|
| 774 |
+
"logits/rejected": 184.17428588867188,
|
| 775 |
+
"logps/chosen": -0.7056841850280762,
|
| 776 |
+
"logps/rejected": -1.287292242050171,
|
| 777 |
+
"loss": 30.5978,
|
| 778 |
+
"nll_loss": 0.9244067072868347,
|
| 779 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 780 |
+
"rewards/chosen": -0.03528420999646187,
|
| 781 |
+
"rewards/margins": 0.029080402106046677,
|
| 782 |
+
"rewards/rejected": -0.06436461955308914,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.0859259259259257,
|
| 787 |
+
"grad_norm": 65.0,
|
| 788 |
+
"learning_rate": 1.2660570475395683e-05,
|
| 789 |
+
"log_odds_chosen": 1.3456884622573853,
|
| 790 |
+
"log_odds_ratio": -0.33826351165771484,
|
| 791 |
+
"logits/chosen": 171.13723754882812,
|
| 792 |
+
"logits/rejected": 175.174072265625,
|
| 793 |
+
"logps/chosen": -0.6312381029129028,
|
| 794 |
+
"logps/rejected": -1.3982974290847778,
|
| 795 |
+
"loss": 27.8654,
|
| 796 |
+
"nll_loss": 0.8854498863220215,
|
| 797 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 798 |
+
"rewards/chosen": -0.03156190365552902,
|
| 799 |
+
"rewards/margins": 0.03835297003388405,
|
| 800 |
+
"rewards/rejected": -0.06991487741470337,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1333333333333333,
|
| 805 |
+
"grad_norm": 39.0,
|
| 806 |
+
"learning_rate": 1.1473369030008974e-05,
|
| 807 |
+
"log_odds_chosen": 1.20364248752594,
|
| 808 |
+
"log_odds_ratio": -0.3720916509628296,
|
| 809 |
+
"logits/chosen": 179.11940002441406,
|
| 810 |
+
"logits/rejected": 177.05686950683594,
|
| 811 |
+
"logps/chosen": -0.643406093120575,
|
| 812 |
+
"logps/rejected": -1.30879545211792,
|
| 813 |
+
"loss": 29.0556,
|
| 814 |
+
"nll_loss": 0.8873918652534485,
|
| 815 |
+
"rewards/accuracies": 0.875,
|
| 816 |
+
"rewards/chosen": -0.03217030316591263,
|
| 817 |
+
"rewards/margins": 0.033269476145505905,
|
| 818 |
+
"rewards/rejected": -0.06543977558612823,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.180740740740741,
|
| 823 |
+
"grad_norm": 56.5,
|
| 824 |
+
"learning_rate": 1.0327830055518842e-05,
|
| 825 |
+
"log_odds_chosen": 1.3064587116241455,
|
| 826 |
+
"log_odds_ratio": -0.3238556981086731,
|
| 827 |
+
"logits/chosen": 175.1729736328125,
|
| 828 |
+
"logits/rejected": 176.44863891601562,
|
| 829 |
+
"logps/chosen": -0.6326884627342224,
|
| 830 |
+
"logps/rejected": -1.3729749917984009,
|
| 831 |
+
"loss": 27.8203,
|
| 832 |
+
"nll_loss": 0.854932427406311,
|
| 833 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 834 |
+
"rewards/chosen": -0.03163442760705948,
|
| 835 |
+
"rewards/margins": 0.03701432794332504,
|
| 836 |
+
"rewards/rejected": -0.06864875555038452,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.228148148148148,
|
| 841 |
+
"grad_norm": 47.0,
|
| 842 |
+
"learning_rate": 9.227481849865235e-06,
|
| 843 |
+
"log_odds_chosen": 1.1249010562896729,
|
| 844 |
+
"log_odds_ratio": -0.3926791250705719,
|
| 845 |
+
"logits/chosen": 173.18649291992188,
|
| 846 |
+
"logits/rejected": 177.1532440185547,
|
| 847 |
+
"logps/chosen": -0.6803759336471558,
|
| 848 |
+
"logps/rejected": -1.2795072793960571,
|
| 849 |
+
"loss": 28.6684,
|
| 850 |
+
"nll_loss": 0.8934875726699829,
|
| 851 |
+
"rewards/accuracies": 0.84375,
|
| 852 |
+
"rewards/chosen": -0.03401879593729973,
|
| 853 |
+
"rewards/margins": 0.02995656430721283,
|
| 854 |
+
"rewards/rejected": -0.06397536396980286,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2755555555555556,
|
| 859 |
+
"grad_norm": 43.75,
|
| 860 |
+
"learning_rate": 8.175713521924978e-06,
|
| 861 |
+
"log_odds_chosen": 1.3639342784881592,
|
| 862 |
+
"log_odds_ratio": -0.3280356526374817,
|
| 863 |
+
"logits/chosen": 170.26976013183594,
|
| 864 |
+
"logits/rejected": 175.56011962890625,
|
| 865 |
+
"logps/chosen": -0.6179399490356445,
|
| 866 |
+
"logps/rejected": -1.3654717206954956,
|
| 867 |
+
"loss": 28.216,
|
| 868 |
+
"nll_loss": 0.8750711679458618,
|
| 869 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 870 |
+
"rewards/chosen": -0.030897000804543495,
|
| 871 |
+
"rewards/margins": 0.037376586347818375,
|
| 872 |
+
"rewards/rejected": -0.06827358901500702,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.322962962962963,
|
| 877 |
+
"grad_norm": 36.25,
|
| 878 |
+
"learning_rate": 7.1757645529443665e-06,
|
| 879 |
+
"log_odds_chosen": 1.3733258247375488,
|
| 880 |
+
"log_odds_ratio": -0.33095091581344604,
|
| 881 |
+
"logits/chosen": 170.61001586914062,
|
| 882 |
+
"logits/rejected": 173.3466339111328,
|
| 883 |
+
"logps/chosen": -0.643506646156311,
|
| 884 |
+
"logps/rejected": -1.4111663103103638,
|
| 885 |
+
"loss": 27.9636,
|
| 886 |
+
"nll_loss": 0.8818845748901367,
|
| 887 |
+
"rewards/accuracies": 0.875,
|
| 888 |
+
"rewards/chosen": -0.03217533975839615,
|
| 889 |
+
"rewards/margins": 0.03838297724723816,
|
| 890 |
+
"rewards/rejected": -0.07055831700563431,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3703703703703702,
|
| 895 |
+
"grad_norm": 36.25,
|
| 896 |
+
"learning_rate": 6.230714818829733e-06,
|
| 897 |
+
"log_odds_chosen": 1.4122543334960938,
|
| 898 |
+
"log_odds_ratio": -0.3063311278820038,
|
| 899 |
+
"logits/chosen": 166.13916015625,
|
| 900 |
+
"logits/rejected": 172.06869506835938,
|
| 901 |
+
"logps/chosen": -0.6153512001037598,
|
| 902 |
+
"logps/rejected": -1.4051437377929688,
|
| 903 |
+
"loss": 27.1152,
|
| 904 |
+
"nll_loss": 0.8230810165405273,
|
| 905 |
+
"rewards/accuracies": 0.893750011920929,
|
| 906 |
+
"rewards/chosen": -0.030767563730478287,
|
| 907 |
+
"rewards/margins": 0.03948962315917015,
|
| 908 |
+
"rewards/rejected": -0.07025718688964844,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.417777777777778,
|
| 913 |
+
"grad_norm": 39.75,
|
| 914 |
+
"learning_rate": 5.343475104027743e-06,
|
| 915 |
+
"log_odds_chosen": 1.5338956117630005,
|
| 916 |
+
"log_odds_ratio": -0.3078998625278473,
|
| 917 |
+
"logits/chosen": 163.81158447265625,
|
| 918 |
+
"logits/rejected": 167.9901580810547,
|
| 919 |
+
"logps/chosen": -0.5699383020401001,
|
| 920 |
+
"logps/rejected": -1.4373152256011963,
|
| 921 |
+
"loss": 27.1707,
|
| 922 |
+
"nll_loss": 0.8324142694473267,
|
| 923 |
+
"rewards/accuracies": 0.90625,
|
| 924 |
+
"rewards/chosen": -0.028496911749243736,
|
| 925 |
+
"rewards/margins": 0.043368853628635406,
|
| 926 |
+
"rewards/rejected": -0.0718657597899437,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.4651851851851854,
|
| 931 |
+
"grad_norm": 47.5,
|
| 932 |
+
"learning_rate": 4.516778136213037e-06,
|
| 933 |
+
"log_odds_chosen": 1.4830719232559204,
|
| 934 |
+
"log_odds_ratio": -0.31063929200172424,
|
| 935 |
+
"logits/chosen": 163.97885131835938,
|
| 936 |
+
"logits/rejected": 169.63653564453125,
|
| 937 |
+
"logps/chosen": -0.606620728969574,
|
| 938 |
+
"logps/rejected": -1.4316565990447998,
|
| 939 |
+
"loss": 27.166,
|
| 940 |
+
"nll_loss": 0.8371620178222656,
|
| 941 |
+
"rewards/accuracies": 0.875,
|
| 942 |
+
"rewards/chosen": -0.03033103607594967,
|
| 943 |
+
"rewards/margins": 0.04125178977847099,
|
| 944 |
+
"rewards/rejected": -0.07158282399177551,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5125925925925925,
|
| 949 |
+
"grad_norm": 45.0,
|
| 950 |
+
"learning_rate": 3.7531701693965554e-06,
|
| 951 |
+
"log_odds_chosen": 1.2772926092147827,
|
| 952 |
+
"log_odds_ratio": -0.3710404634475708,
|
| 953 |
+
"logits/chosen": 166.98779296875,
|
| 954 |
+
"logits/rejected": 168.69686889648438,
|
| 955 |
+
"logps/chosen": -0.6445072889328003,
|
| 956 |
+
"logps/rejected": -1.3713839054107666,
|
| 957 |
+
"loss": 27.2089,
|
| 958 |
+
"nll_loss": 0.8413775563240051,
|
| 959 |
+
"rewards/accuracies": 0.875,
|
| 960 |
+
"rewards/chosen": -0.032225366681814194,
|
| 961 |
+
"rewards/margins": 0.036343835294246674,
|
| 962 |
+
"rewards/rejected": -0.06856919825077057,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.56,
|
| 967 |
+
"grad_norm": 38.75,
|
| 968 |
+
"learning_rate": 3.055003141378948e-06,
|
| 969 |
+
"log_odds_chosen": 1.4745795726776123,
|
| 970 |
+
"log_odds_ratio": -0.29757434129714966,
|
| 971 |
+
"logits/chosen": 167.68179321289062,
|
| 972 |
+
"logits/rejected": 172.54640197753906,
|
| 973 |
+
"logps/chosen": -0.6402829885482788,
|
| 974 |
+
"logps/rejected": -1.4750518798828125,
|
| 975 |
+
"loss": 27.9266,
|
| 976 |
+
"nll_loss": 0.8752562403678894,
|
| 977 |
+
"rewards/accuracies": 0.90625,
|
| 978 |
+
"rewards/chosen": -0.0320141427218914,
|
| 979 |
+
"rewards/margins": 0.041738446801900864,
|
| 980 |
+
"rewards/rejected": -0.07375258952379227,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.6074074074074076,
|
| 985 |
+
"grad_norm": 47.75,
|
| 986 |
+
"learning_rate": 2.424427429704365e-06,
|
| 987 |
+
"log_odds_chosen": 1.4180128574371338,
|
| 988 |
+
"log_odds_ratio": -0.3266395926475525,
|
| 989 |
+
"logits/chosen": 165.8551483154297,
|
| 990 |
+
"logits/rejected": 167.0991668701172,
|
| 991 |
+
"logps/chosen": -0.6501474976539612,
|
| 992 |
+
"logps/rejected": -1.4511449337005615,
|
| 993 |
+
"loss": 28.3881,
|
| 994 |
+
"nll_loss": 0.8757912516593933,
|
| 995 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 996 |
+
"rewards/chosen": -0.03250737115740776,
|
| 997 |
+
"rewards/margins": 0.04004988074302673,
|
| 998 |
+
"rewards/rejected": -0.07255725562572479,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.6548148148148147,
|
| 1003 |
+
"grad_norm": 42.5,
|
| 1004 |
+
"learning_rate": 1.8633852284264508e-06,
|
| 1005 |
+
"log_odds_chosen": 1.3011739253997803,
|
| 1006 |
+
"log_odds_ratio": -0.3382631838321686,
|
| 1007 |
+
"logits/chosen": 167.21853637695312,
|
| 1008 |
+
"logits/rejected": 169.163330078125,
|
| 1009 |
+
"logps/chosen": -0.6230785250663757,
|
| 1010 |
+
"logps/rejected": -1.3411012887954712,
|
| 1011 |
+
"loss": 26.7019,
|
| 1012 |
+
"nll_loss": 0.8079374432563782,
|
| 1013 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1014 |
+
"rewards/chosen": -0.031153922900557518,
|
| 1015 |
+
"rewards/margins": 0.035901136696338654,
|
| 1016 |
+
"rewards/rejected": -0.06705506145954132,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.7022222222222223,
|
| 1021 |
+
"grad_norm": 40.0,
|
| 1022 |
+
"learning_rate": 1.3736045660864034e-06,
|
| 1023 |
+
"log_odds_chosen": 1.3024629354476929,
|
| 1024 |
+
"log_odds_ratio": -0.3390798270702362,
|
| 1025 |
+
"logits/chosen": 169.4490203857422,
|
| 1026 |
+
"logits/rejected": 169.3362274169922,
|
| 1027 |
+
"logps/chosen": -0.6201122403144836,
|
| 1028 |
+
"logps/rejected": -1.3454174995422363,
|
| 1029 |
+
"loss": 28.0871,
|
| 1030 |
+
"nll_loss": 0.8465447425842285,
|
| 1031 |
+
"rewards/accuracies": 0.875,
|
| 1032 |
+
"rewards/chosen": -0.03100561536848545,
|
| 1033 |
+
"rewards/margins": 0.036265257745981216,
|
| 1034 |
+
"rewards/rejected": -0.06727087497711182,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.74962962962963,
|
| 1039 |
+
"grad_norm": 44.75,
|
| 1040 |
+
"learning_rate": 9.565939833279192e-07,
|
| 1041 |
+
"log_odds_chosen": 1.383882761001587,
|
| 1042 |
+
"log_odds_ratio": -0.31699490547180176,
|
| 1043 |
+
"logits/chosen": 167.99237060546875,
|
| 1044 |
+
"logits/rejected": 173.28675842285156,
|
| 1045 |
+
"logps/chosen": -0.6342524290084839,
|
| 1046 |
+
"logps/rejected": -1.401726484298706,
|
| 1047 |
+
"loss": 28.0037,
|
| 1048 |
+
"nll_loss": 0.8696128726005554,
|
| 1049 |
+
"rewards/accuracies": 0.893750011920929,
|
| 1050 |
+
"rewards/chosen": -0.03171262517571449,
|
| 1051 |
+
"rewards/margins": 0.03837370499968529,
|
| 1052 |
+
"rewards/rejected": -0.07008632272481918,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.797037037037037,
|
| 1057 |
+
"grad_norm": 51.5,
|
| 1058 |
+
"learning_rate": 6.136378865420872e-07,
|
| 1059 |
+
"log_odds_chosen": 1.4249391555786133,
|
| 1060 |
+
"log_odds_ratio": -0.35160475969314575,
|
| 1061 |
+
"logits/chosen": 167.20567321777344,
|
| 1062 |
+
"logits/rejected": 170.51210021972656,
|
| 1063 |
+
"logps/chosen": -0.6513175964355469,
|
| 1064 |
+
"logps/rejected": -1.4825408458709717,
|
| 1065 |
+
"loss": 27.9843,
|
| 1066 |
+
"nll_loss": 0.8786695599555969,
|
| 1067 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1068 |
+
"rewards/chosen": -0.0325658805668354,
|
| 1069 |
+
"rewards/margins": 0.04156116023659706,
|
| 1070 |
+
"rewards/rejected": -0.07412704080343246,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.8444444444444446,
|
| 1075 |
+
"grad_norm": 43.5,
|
| 1076 |
+
"learning_rate": 3.45792591853214e-07,
|
| 1077 |
+
"log_odds_chosen": 1.237529993057251,
|
| 1078 |
+
"log_odds_ratio": -0.36313003301620483,
|
| 1079 |
+
"logits/chosen": 175.26461791992188,
|
| 1080 |
+
"logits/rejected": 173.83885192871094,
|
| 1081 |
+
"logps/chosen": -0.6848838925361633,
|
| 1082 |
+
"logps/rejected": -1.3957815170288086,
|
| 1083 |
+
"loss": 28.3691,
|
| 1084 |
+
"nll_loss": 0.9239709973335266,
|
| 1085 |
+
"rewards/accuracies": 0.875,
|
| 1086 |
+
"rewards/chosen": -0.034244198352098465,
|
| 1087 |
+
"rewards/margins": 0.035544879734516144,
|
| 1088 |
+
"rewards/rejected": -0.06978908181190491,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.891851851851852,
|
| 1093 |
+
"grad_norm": 40.0,
|
| 1094 |
+
"learning_rate": 1.538830716302092e-07,
|
| 1095 |
+
"log_odds_chosen": 1.354568362236023,
|
| 1096 |
+
"log_odds_ratio": -0.33406147360801697,
|
| 1097 |
+
"logits/chosen": 168.0280303955078,
|
| 1098 |
+
"logits/rejected": 169.37564086914062,
|
| 1099 |
+
"logps/chosen": -0.6664038300514221,
|
| 1100 |
+
"logps/rejected": -1.40469491481781,
|
| 1101 |
+
"loss": 26.5858,
|
| 1102 |
+
"nll_loss": 0.817513108253479,
|
| 1103 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1104 |
+
"rewards/chosen": -0.033320192247629166,
|
| 1105 |
+
"rewards/margins": 0.036914560943841934,
|
| 1106 |
+
"rewards/rejected": -0.0702347606420517,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9392592592592592,
|
| 1111 |
+
"grad_norm": 52.5,
|
| 1112 |
+
"learning_rate": 3.8500413544415025e-08,
|
| 1113 |
+
"log_odds_chosen": 1.4047539234161377,
|
| 1114 |
+
"log_odds_ratio": -0.31650859117507935,
|
| 1115 |
+
"logits/chosen": 168.70623779296875,
|
| 1116 |
+
"logits/rejected": 170.0711669921875,
|
| 1117 |
+
"logps/chosen": -0.6374054551124573,
|
| 1118 |
+
"logps/rejected": -1.395747184753418,
|
| 1119 |
+
"loss": 27.6658,
|
| 1120 |
+
"nll_loss": 0.8348110318183899,
|
| 1121 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1122 |
+
"rewards/chosen": -0.031870268285274506,
|
| 1123 |
+
"rewards/margins": 0.037917084991931915,
|
| 1124 |
+
"rewards/rejected": -0.06978735327720642,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.986666666666667,
|
| 1129 |
+
"grad_norm": 61.5,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 1.5434544086456299,
|
| 1132 |
+
"log_odds_ratio": -0.29583439230918884,
|
| 1133 |
+
"logits/chosen": 165.3496856689453,
|
| 1134 |
+
"logits/rejected": 166.1381378173828,
|
| 1135 |
+
"logps/chosen": -0.5781136155128479,
|
| 1136 |
+
"logps/rejected": -1.430616021156311,
|
| 1137 |
+
"loss": 27.0079,
|
| 1138 |
+
"nll_loss": 0.8401652574539185,
|
| 1139 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1140 |
+
"rewards/chosen": -0.028905680403113365,
|
| 1141 |
+
"rewards/margins": 0.042625121772289276,
|
| 1142 |
+
"rewards/rejected": -0.07153080403804779,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.986666666666667,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 50.88312741234189,
|
| 1150 |
+
"train_runtime": 6896.118,
|
| 1151 |
+
"train_samples_per_second": 2.936,
|
| 1152 |
+
"train_steps_per_second": 0.046
|
| 1153 |
+
}
|
| 1154 |
+
],
|
| 1155 |
+
"logging_steps": 5,
|
| 1156 |
+
"max_steps": 315,
|
| 1157 |
+
"num_input_tokens_seen": 0,
|
| 1158 |
+
"num_train_epochs": 3,
|
| 1159 |
+
"save_steps": 100000,
|
| 1160 |
+
"stateful_callbacks": {
|
| 1161 |
+
"TrainerControl": {
|
| 1162 |
+
"args": {
|
| 1163 |
+
"should_epoch_stop": false,
|
| 1164 |
+
"should_evaluate": false,
|
| 1165 |
+
"should_log": false,
|
| 1166 |
+
"should_save": true,
|
| 1167 |
+
"should_training_stop": true
|
| 1168 |
+
},
|
| 1169 |
+
"attributes": {}
|
| 1170 |
+
}
|
| 1171 |
+
},
|
| 1172 |
+
"total_flos": 0.0,
|
| 1173 |
+
"train_batch_size": 1,
|
| 1174 |
+
"trial_name": null,
|
| 1175 |
+
"trial_params": null
|
| 1176 |
+
}
|