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: mistralai/Mistral-7B-v0.3
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: mistral-7b-orpo-noisy-6e-5
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for mistral-7b-orpo-noisy-6e-5
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3).
|
| 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/mistral-7b-orpo-noisy-6e-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/5sqhouro)
|
| 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": 31.135096304757255,
|
| 5 |
+
"train_runtime": 6745.6063,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 3.002,
|
| 8 |
+
"train_steps_per_second": 0.047
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.986666666666667,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 31.135096304757255,
|
| 5 |
+
"train_runtime": 6745.6063,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 3.002,
|
| 8 |
+
"train_steps_per_second": 0.047
|
| 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": 270.0,
|
| 14 |
+
"learning_rate": 9.375000000000001e-06,
|
| 15 |
+
"log_odds_chosen": 0.41771596670150757,
|
| 16 |
+
"log_odds_ratio": -0.7694265246391296,
|
| 17 |
+
"logits/chosen": -2.967926502227783,
|
| 18 |
+
"logits/rejected": -2.8778510093688965,
|
| 19 |
+
"logps/chosen": -1.2910274267196655,
|
| 20 |
+
"logps/rejected": -1.6328433752059937,
|
| 21 |
+
"loss": 51.9175,
|
| 22 |
+
"nll_loss": 1.511154294013977,
|
| 23 |
+
"rewards/accuracies": 0.574999988079071,
|
| 24 |
+
"rewards/chosen": -0.06455137580633163,
|
| 25 |
+
"rewards/margins": 0.017090797424316406,
|
| 26 |
+
"rewards/rejected": -0.08164217323064804,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 61.75,
|
| 32 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 33 |
+
"log_odds_chosen": 0.2715613842010498,
|
| 34 |
+
"log_odds_ratio": -0.7063366174697876,
|
| 35 |
+
"logits/chosen": -2.903649091720581,
|
| 36 |
+
"logits/rejected": -2.737760066986084,
|
| 37 |
+
"logps/chosen": -1.0549781322479248,
|
| 38 |
+
"logps/rejected": -1.2600512504577637,
|
| 39 |
+
"loss": 47.142,
|
| 40 |
+
"nll_loss": 1.388285517692566,
|
| 41 |
+
"rewards/accuracies": 0.59375,
|
| 42 |
+
"rewards/chosen": -0.0527489073574543,
|
| 43 |
+
"rewards/margins": 0.010253657586872578,
|
| 44 |
+
"rewards/rejected": -0.0630025640130043,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 71.0,
|
| 50 |
+
"learning_rate": 2.8125e-05,
|
| 51 |
+
"log_odds_chosen": 0.26208820939064026,
|
| 52 |
+
"log_odds_ratio": -0.6782652139663696,
|
| 53 |
+
"logits/chosen": -2.5858330726623535,
|
| 54 |
+
"logits/rejected": -2.4748170375823975,
|
| 55 |
+
"logps/chosen": -0.921225368976593,
|
| 56 |
+
"logps/rejected": -1.081853985786438,
|
| 57 |
+
"loss": 46.5682,
|
| 58 |
+
"nll_loss": 1.4536203145980835,
|
| 59 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 60 |
+
"rewards/chosen": -0.04606126993894577,
|
| 61 |
+
"rewards/margins": 0.008031422272324562,
|
| 62 |
+
"rewards/rejected": -0.05409269407391548,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 92.0,
|
| 68 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 69 |
+
"log_odds_chosen": 0.18918053805828094,
|
| 70 |
+
"log_odds_ratio": -0.7011532783508301,
|
| 71 |
+
"logits/chosen": -2.4796886444091797,
|
| 72 |
+
"logits/rejected": -2.3676044940948486,
|
| 73 |
+
"logps/chosen": -0.8980884552001953,
|
| 74 |
+
"logps/rejected": -1.0366885662078857,
|
| 75 |
+
"loss": 43.8254,
|
| 76 |
+
"nll_loss": 1.3132244348526,
|
| 77 |
+
"rewards/accuracies": 0.606249988079071,
|
| 78 |
+
"rewards/chosen": -0.044904422014951706,
|
| 79 |
+
"rewards/margins": 0.00693000853061676,
|
| 80 |
+
"rewards/rejected": -0.05183442682027817,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 39.75,
|
| 86 |
+
"learning_rate": 4.6875e-05,
|
| 87 |
+
"log_odds_chosen": 0.20047792792320251,
|
| 88 |
+
"log_odds_ratio": -0.704607367515564,
|
| 89 |
+
"logits/chosen": -2.486077070236206,
|
| 90 |
+
"logits/rejected": -2.380354881286621,
|
| 91 |
+
"logps/chosen": -0.9169360995292664,
|
| 92 |
+
"logps/rejected": -1.0517938137054443,
|
| 93 |
+
"loss": 41.7393,
|
| 94 |
+
"nll_loss": 1.3125925064086914,
|
| 95 |
+
"rewards/accuracies": 0.5625,
|
| 96 |
+
"rewards/chosen": -0.04584680497646332,
|
| 97 |
+
"rewards/margins": 0.006742885801941156,
|
| 98 |
+
"rewards/rejected": -0.052589692175388336,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 48.25,
|
| 104 |
+
"learning_rate": 5.625e-05,
|
| 105 |
+
"log_odds_chosen": 0.13125675916671753,
|
| 106 |
+
"log_odds_ratio": -0.714677631855011,
|
| 107 |
+
"logits/chosen": -2.498034954071045,
|
| 108 |
+
"logits/rejected": -2.1313350200653076,
|
| 109 |
+
"logps/chosen": -0.8735687136650085,
|
| 110 |
+
"logps/rejected": -0.9592132568359375,
|
| 111 |
+
"loss": 41.493,
|
| 112 |
+
"nll_loss": 1.2559503316879272,
|
| 113 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 114 |
+
"rewards/chosen": -0.04367843270301819,
|
| 115 |
+
"rewards/margins": 0.004282232839614153,
|
| 116 |
+
"rewards/rejected": -0.047960661351680756,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 39.0,
|
| 122 |
+
"learning_rate": 5.998336508818541e-05,
|
| 123 |
+
"log_odds_chosen": 0.06586463749408722,
|
| 124 |
+
"log_odds_ratio": -0.7540255188941956,
|
| 125 |
+
"logits/chosen": -2.3031373023986816,
|
| 126 |
+
"logits/rejected": -2.4485721588134766,
|
| 127 |
+
"logps/chosen": -0.8969869613647461,
|
| 128 |
+
"logps/rejected": -0.9382694363594055,
|
| 129 |
+
"loss": 40.5282,
|
| 130 |
+
"nll_loss": 1.2521088123321533,
|
| 131 |
+
"rewards/accuracies": 0.53125,
|
| 132 |
+
"rewards/chosen": -0.04484934359788895,
|
| 133 |
+
"rewards/margins": 0.0020641214214265347,
|
| 134 |
+
"rewards/rejected": -0.04691346734762192,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 35.25,
|
| 140 |
+
"learning_rate": 5.988177409372154e-05,
|
| 141 |
+
"log_odds_chosen": 0.21956849098205566,
|
| 142 |
+
"log_odds_ratio": -0.6834356188774109,
|
| 143 |
+
"logits/chosen": -2.2723240852355957,
|
| 144 |
+
"logits/rejected": -2.065520763397217,
|
| 145 |
+
"logps/chosen": -0.8550432324409485,
|
| 146 |
+
"logps/rejected": -0.9936239123344421,
|
| 147 |
+
"loss": 40.1999,
|
| 148 |
+
"nll_loss": 1.2111032009124756,
|
| 149 |
+
"rewards/accuracies": 0.550000011920929,
|
| 150 |
+
"rewards/chosen": -0.04275216534733772,
|
| 151 |
+
"rewards/margins": 0.006929035298526287,
|
| 152 |
+
"rewards/rejected": -0.049681201577186584,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 33.0,
|
| 158 |
+
"learning_rate": 5.968814624645376e-05,
|
| 159 |
+
"log_odds_chosen": 0.1635906845331192,
|
| 160 |
+
"log_odds_ratio": -0.7335126996040344,
|
| 161 |
+
"logits/chosen": -1.9793834686279297,
|
| 162 |
+
"logits/rejected": -2.0152127742767334,
|
| 163 |
+
"logps/chosen": -0.8692102432250977,
|
| 164 |
+
"logps/rejected": -0.975513756275177,
|
| 165 |
+
"loss": 40.1299,
|
| 166 |
+
"nll_loss": 1.2306907176971436,
|
| 167 |
+
"rewards/accuracies": 0.543749988079071,
|
| 168 |
+
"rewards/chosen": -0.04346051067113876,
|
| 169 |
+
"rewards/margins": 0.005315178073942661,
|
| 170 |
+
"rewards/rejected": -0.04877568781375885,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 39.75,
|
| 176 |
+
"learning_rate": 5.9403077926557534e-05,
|
| 177 |
+
"log_odds_chosen": 0.16285523772239685,
|
| 178 |
+
"log_odds_ratio": -0.7225431203842163,
|
| 179 |
+
"logits/chosen": -1.9700477123260498,
|
| 180 |
+
"logits/rejected": -1.93939208984375,
|
| 181 |
+
"logps/chosen": -0.9150403738021851,
|
| 182 |
+
"logps/rejected": -1.005976676940918,
|
| 183 |
+
"loss": 42.3638,
|
| 184 |
+
"nll_loss": 1.3180006742477417,
|
| 185 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 186 |
+
"rewards/chosen": -0.04575202241539955,
|
| 187 |
+
"rewards/margins": 0.004546813666820526,
|
| 188 |
+
"rewards/rejected": -0.05029883235692978,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 32.5,
|
| 194 |
+
"learning_rate": 5.9027447153889215e-05,
|
| 195 |
+
"log_odds_chosen": 0.074959896504879,
|
| 196 |
+
"log_odds_ratio": -0.7347756624221802,
|
| 197 |
+
"logits/chosen": -1.8091471195220947,
|
| 198 |
+
"logits/rejected": -1.627111792564392,
|
| 199 |
+
"logps/chosen": -0.8783036470413208,
|
| 200 |
+
"logps/rejected": -0.9295312166213989,
|
| 201 |
+
"loss": 39.4889,
|
| 202 |
+
"nll_loss": 1.2200506925582886,
|
| 203 |
+
"rewards/accuracies": 0.543749988079071,
|
| 204 |
+
"rewards/chosen": -0.04391518235206604,
|
| 205 |
+
"rewards/margins": 0.002561377827078104,
|
| 206 |
+
"rewards/rejected": -0.046476561576128006,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 31.875,
|
| 212 |
+
"learning_rate": 5.856241088365584e-05,
|
| 213 |
+
"log_odds_chosen": 0.21836993098258972,
|
| 214 |
+
"log_odds_ratio": -0.6648741960525513,
|
| 215 |
+
"logits/chosen": -2.2059853076934814,
|
| 216 |
+
"logits/rejected": -1.8393570184707642,
|
| 217 |
+
"logps/chosen": -0.8266533613204956,
|
| 218 |
+
"logps/rejected": -0.944961428642273,
|
| 219 |
+
"loss": 38.4828,
|
| 220 |
+
"nll_loss": 1.1561448574066162,
|
| 221 |
+
"rewards/accuracies": 0.5625,
|
| 222 |
+
"rewards/chosen": -0.0413326658308506,
|
| 223 |
+
"rewards/margins": 0.005915405694395304,
|
| 224 |
+
"rewards/rejected": -0.04724807292222977,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 31.75,
|
| 230 |
+
"learning_rate": 5.800940144295476e-05,
|
| 231 |
+
"log_odds_chosen": 0.14650335907936096,
|
| 232 |
+
"log_odds_ratio": -0.7161463499069214,
|
| 233 |
+
"logits/chosen": -1.9598195552825928,
|
| 234 |
+
"logits/rejected": -1.8855581283569336,
|
| 235 |
+
"logps/chosen": -0.890237033367157,
|
| 236 |
+
"logps/rejected": -0.9888992309570312,
|
| 237 |
+
"loss": 38.4474,
|
| 238 |
+
"nll_loss": 1.1701605319976807,
|
| 239 |
+
"rewards/accuracies": 0.543749988079071,
|
| 240 |
+
"rewards/chosen": -0.04451185464859009,
|
| 241 |
+
"rewards/margins": 0.004933114163577557,
|
| 242 |
+
"rewards/rejected": -0.04944496601819992,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 31.25,
|
| 248 |
+
"learning_rate": 5.7370122119158855e-05,
|
| 249 |
+
"log_odds_chosen": 0.2176527976989746,
|
| 250 |
+
"log_odds_ratio": -0.6846314072608948,
|
| 251 |
+
"logits/chosen": -2.401538610458374,
|
| 252 |
+
"logits/rejected": -1.8277839422225952,
|
| 253 |
+
"logps/chosen": -0.8448864817619324,
|
| 254 |
+
"logps/rejected": -1.0054863691329956,
|
| 255 |
+
"loss": 38.0337,
|
| 256 |
+
"nll_loss": 1.1650002002716064,
|
| 257 |
+
"rewards/accuracies": 0.59375,
|
| 258 |
+
"rewards/chosen": -0.0422443225979805,
|
| 259 |
+
"rewards/margins": 0.00802999921143055,
|
| 260 |
+
"rewards/rejected": -0.0502743236720562,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 34.5,
|
| 266 |
+
"learning_rate": 5.6646541913735056e-05,
|
| 267 |
+
"log_odds_chosen": 0.34958410263061523,
|
| 268 |
+
"log_odds_ratio": -0.6144381761550903,
|
| 269 |
+
"logits/chosen": -1.9800045490264893,
|
| 270 |
+
"logits/rejected": -2.1104648113250732,
|
| 271 |
+
"logps/chosen": -0.7909008264541626,
|
| 272 |
+
"logps/rejected": -1.0040924549102783,
|
| 273 |
+
"loss": 38.1687,
|
| 274 |
+
"nll_loss": 1.1795135736465454,
|
| 275 |
+
"rewards/accuracies": 0.675000011920929,
|
| 276 |
+
"rewards/chosen": -0.03954503685235977,
|
| 277 |
+
"rewards/margins": 0.01065958570688963,
|
| 278 |
+
"rewards/rejected": -0.050204623490571976,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 30.75,
|
| 284 |
+
"learning_rate": 5.5840889477654665e-05,
|
| 285 |
+
"log_odds_chosen": 0.25651440024375916,
|
| 286 |
+
"log_odds_ratio": -0.6959986686706543,
|
| 287 |
+
"logits/chosen": -2.3197991847991943,
|
| 288 |
+
"logits/rejected": -1.9464877843856812,
|
| 289 |
+
"logps/chosen": -0.8813208341598511,
|
| 290 |
+
"logps/rejected": -1.0397270917892456,
|
| 291 |
+
"loss": 37.8718,
|
| 292 |
+
"nll_loss": 1.1892088651657104,
|
| 293 |
+
"rewards/accuracies": 0.612500011920929,
|
| 294 |
+
"rewards/chosen": -0.04406604543328285,
|
| 295 |
+
"rewards/margins": 0.007920312695205212,
|
| 296 |
+
"rewards/rejected": -0.05198635905981064,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 26.75,
|
| 302 |
+
"learning_rate": 5.495564624707466e-05,
|
| 303 |
+
"log_odds_chosen": 0.24601168930530548,
|
| 304 |
+
"log_odds_ratio": -0.6635450720787048,
|
| 305 |
+
"logits/chosen": -2.3646152019500732,
|
| 306 |
+
"logits/rejected": -1.6053569316864014,
|
| 307 |
+
"logps/chosen": -0.825157642364502,
|
| 308 |
+
"logps/rejected": -0.9831274151802063,
|
| 309 |
+
"loss": 37.7402,
|
| 310 |
+
"nll_loss": 1.1433099508285522,
|
| 311 |
+
"rewards/accuracies": 0.625,
|
| 312 |
+
"rewards/chosen": -0.04125788062810898,
|
| 313 |
+
"rewards/margins": 0.007898489013314247,
|
| 314 |
+
"rewards/rejected": -0.04915637522935867,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 32.25,
|
| 320 |
+
"learning_rate": 5.399353880043222e-05,
|
| 321 |
+
"log_odds_chosen": 0.2743232548236847,
|
| 322 |
+
"log_odds_ratio": -0.6547017097473145,
|
| 323 |
+
"logits/chosen": -2.2998695373535156,
|
| 324 |
+
"logits/rejected": -1.9147183895111084,
|
| 325 |
+
"logps/chosen": -0.7987005710601807,
|
| 326 |
+
"logps/rejected": -0.9570469856262207,
|
| 327 |
+
"loss": 39.0598,
|
| 328 |
+
"nll_loss": 1.165475606918335,
|
| 329 |
+
"rewards/accuracies": 0.643750011920929,
|
| 330 |
+
"rewards/chosen": -0.03993503004312515,
|
| 331 |
+
"rewards/margins": 0.00791732408106327,
|
| 332 |
+
"rewards/rejected": -0.047852352261543274,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 27.875,
|
| 338 |
+
"learning_rate": 5.295753046049293e-05,
|
| 339 |
+
"log_odds_chosen": 0.3104208707809448,
|
| 340 |
+
"log_odds_ratio": -0.6332544088363647,
|
| 341 |
+
"logits/chosen": -2.3582139015197754,
|
| 342 |
+
"logits/rejected": -1.8759187459945679,
|
| 343 |
+
"logps/chosen": -0.7584289908409119,
|
| 344 |
+
"logps/rejected": -0.9375408887863159,
|
| 345 |
+
"loss": 38.0891,
|
| 346 |
+
"nll_loss": 1.0971782207489014,
|
| 347 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 348 |
+
"rewards/chosen": -0.037921447306871414,
|
| 349 |
+
"rewards/margins": 0.00895559974014759,
|
| 350 |
+
"rewards/rejected": -0.046877048909664154,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 28.0,
|
| 356 |
+
"learning_rate": 5.1850812167218644e-05,
|
| 357 |
+
"log_odds_chosen": 0.14483532309532166,
|
| 358 |
+
"log_odds_ratio": -0.725937008857727,
|
| 359 |
+
"logits/chosen": -2.2758800983428955,
|
| 360 |
+
"logits/rejected": -1.807739019393921,
|
| 361 |
+
"logps/chosen": -0.8715543746948242,
|
| 362 |
+
"logps/rejected": -0.967276394367218,
|
| 363 |
+
"loss": 38.3176,
|
| 364 |
+
"nll_loss": 1.192031979560852,
|
| 365 |
+
"rewards/accuracies": 0.550000011920929,
|
| 366 |
+
"rewards/chosen": -0.04357772320508957,
|
| 367 |
+
"rewards/margins": 0.004786101635545492,
|
| 368 |
+
"rewards/rejected": -0.0483638234436512,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 27.5,
|
| 374 |
+
"learning_rate": 5.067679264956681e-05,
|
| 375 |
+
"log_odds_chosen": 0.2537747621536255,
|
| 376 |
+
"log_odds_ratio": -0.6505337953567505,
|
| 377 |
+
"logits/chosen": -2.356247901916504,
|
| 378 |
+
"logits/rejected": -1.834238052368164,
|
| 379 |
+
"logps/chosen": -0.8050671815872192,
|
| 380 |
+
"logps/rejected": -0.9783474206924438,
|
| 381 |
+
"loss": 37.2872,
|
| 382 |
+
"nll_loss": 1.1223804950714111,
|
| 383 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 384 |
+
"rewards/chosen": -0.04025335982441902,
|
| 385 |
+
"rewards/margins": 0.008664008229970932,
|
| 386 |
+
"rewards/rejected": -0.04891737177968025,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.0429629629629629,
|
| 391 |
+
"grad_norm": 26.625,
|
| 392 |
+
"learning_rate": 4.943908792649255e-05,
|
| 393 |
+
"log_odds_chosen": 0.7013475298881531,
|
| 394 |
+
"log_odds_ratio": -0.5124364495277405,
|
| 395 |
+
"logits/chosen": -2.1879847049713135,
|
| 396 |
+
"logits/rejected": -1.7397973537445068,
|
| 397 |
+
"logps/chosen": -0.6286161541938782,
|
| 398 |
+
"logps/rejected": -0.9900426864624023,
|
| 399 |
+
"loss": 31.0738,
|
| 400 |
+
"nll_loss": 0.9219255447387695,
|
| 401 |
+
"rewards/accuracies": 0.75,
|
| 402 |
+
"rewards/chosen": -0.03143080696463585,
|
| 403 |
+
"rewards/margins": 0.018071329221129417,
|
| 404 |
+
"rewards/rejected": -0.049502138048410416,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0903703703703704,
|
| 409 |
+
"grad_norm": 30.25,
|
| 410 |
+
"learning_rate": 4.814151016949061e-05,
|
| 411 |
+
"log_odds_chosen": 0.9226773977279663,
|
| 412 |
+
"log_odds_ratio": -0.42190057039260864,
|
| 413 |
+
"logits/chosen": -2.087578058242798,
|
| 414 |
+
"logits/rejected": -1.6346263885498047,
|
| 415 |
+
"logps/chosen": -0.5833232998847961,
|
| 416 |
+
"logps/rejected": -1.0628697872161865,
|
| 417 |
+
"loss": 30.0304,
|
| 418 |
+
"nll_loss": 0.8877116441726685,
|
| 419 |
+
"rewards/accuracies": 0.800000011920929,
|
| 420 |
+
"rewards/chosen": -0.02916616201400757,
|
| 421 |
+
"rewards/margins": 0.02397732436656952,
|
| 422 |
+
"rewards/rejected": -0.053143490105867386,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1377777777777778,
|
| 427 |
+
"grad_norm": 28.375,
|
| 428 |
+
"learning_rate": 4.6788055960981e-05,
|
| 429 |
+
"log_odds_chosen": 0.9922162294387817,
|
| 430 |
+
"log_odds_ratio": -0.40503960847854614,
|
| 431 |
+
"logits/chosen": -2.222867488861084,
|
| 432 |
+
"logits/rejected": -1.8629436492919922,
|
| 433 |
+
"logps/chosen": -0.5679124593734741,
|
| 434 |
+
"logps/rejected": -1.079012155532837,
|
| 435 |
+
"loss": 30.4758,
|
| 436 |
+
"nll_loss": 0.9042154550552368,
|
| 437 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 438 |
+
"rewards/chosen": -0.028395622968673706,
|
| 439 |
+
"rewards/margins": 0.02555498108267784,
|
| 440 |
+
"rewards/rejected": -0.053950607776641846,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1851851851851851,
|
| 445 |
+
"grad_norm": 27.625,
|
| 446 |
+
"learning_rate": 4.538289398470304e-05,
|
| 447 |
+
"log_odds_chosen": 0.9345417022705078,
|
| 448 |
+
"log_odds_ratio": -0.4391079545021057,
|
| 449 |
+
"logits/chosen": -2.159498453140259,
|
| 450 |
+
"logits/rejected": -1.9323101043701172,
|
| 451 |
+
"logps/chosen": -0.6106966733932495,
|
| 452 |
+
"logps/rejected": -1.0727020502090454,
|
| 453 |
+
"loss": 28.3606,
|
| 454 |
+
"nll_loss": 0.874626636505127,
|
| 455 |
+
"rewards/accuracies": 0.8125,
|
| 456 |
+
"rewards/chosen": -0.030534833669662476,
|
| 457 |
+
"rewards/margins": 0.023100275546312332,
|
| 458 |
+
"rewards/rejected": -0.05363510921597481,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2325925925925927,
|
| 463 |
+
"grad_norm": 25.75,
|
| 464 |
+
"learning_rate": 4.393035218603139e-05,
|
| 465 |
+
"log_odds_chosen": 0.7891913652420044,
|
| 466 |
+
"log_odds_ratio": -0.4749962389469147,
|
| 467 |
+
"logits/chosen": -2.1488184928894043,
|
| 468 |
+
"logits/rejected": -1.8161399364471436,
|
| 469 |
+
"logps/chosen": -0.6154332160949707,
|
| 470 |
+
"logps/rejected": -1.0038516521453857,
|
| 471 |
+
"loss": 29.5153,
|
| 472 |
+
"nll_loss": 0.9258828163146973,
|
| 473 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 474 |
+
"rewards/chosen": -0.030771661549806595,
|
| 475 |
+
"rewards/margins": 0.019420918077230453,
|
| 476 |
+
"rewards/rejected": -0.05019258335232735,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.28,
|
| 481 |
+
"grad_norm": 26.125,
|
| 482 |
+
"learning_rate": 4.243490444176123e-05,
|
| 483 |
+
"log_odds_chosen": 0.912939190864563,
|
| 484 |
+
"log_odds_ratio": -0.4465225338935852,
|
| 485 |
+
"logits/chosen": -2.023993492126465,
|
| 486 |
+
"logits/rejected": -1.8624738454818726,
|
| 487 |
+
"logps/chosen": -0.5600326061248779,
|
| 488 |
+
"logps/rejected": -1.0113087892532349,
|
| 489 |
+
"loss": 29.4182,
|
| 490 |
+
"nll_loss": 0.8695603609085083,
|
| 491 |
+
"rewards/accuracies": 0.793749988079071,
|
| 492 |
+
"rewards/chosen": -0.028001630678772926,
|
| 493 |
+
"rewards/margins": 0.022563805803656578,
|
| 494 |
+
"rewards/rejected": -0.0505654402077198,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3274074074074074,
|
| 499 |
+
"grad_norm": 24.875,
|
| 500 |
+
"learning_rate": 4.090115678041962e-05,
|
| 501 |
+
"log_odds_chosen": 0.8396116495132446,
|
| 502 |
+
"log_odds_ratio": -0.47167715430259705,
|
| 503 |
+
"logits/chosen": -1.9779258966445923,
|
| 504 |
+
"logits/rejected": -1.8630996942520142,
|
| 505 |
+
"logps/chosen": -0.6393855214118958,
|
| 506 |
+
"logps/rejected": -1.073813557624817,
|
| 507 |
+
"loss": 30.4644,
|
| 508 |
+
"nll_loss": 0.9436683654785156,
|
| 509 |
+
"rewards/accuracies": 0.793749988079071,
|
| 510 |
+
"rewards/chosen": -0.031969279050827026,
|
| 511 |
+
"rewards/margins": 0.021721404045820236,
|
| 512 |
+
"rewards/rejected": -0.053690679371356964,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.374814814814815,
|
| 517 |
+
"grad_norm": 27.0,
|
| 518 |
+
"learning_rate": 3.9333833195545325e-05,
|
| 519 |
+
"log_odds_chosen": 0.8570321798324585,
|
| 520 |
+
"log_odds_ratio": -0.4445961117744446,
|
| 521 |
+
"logits/chosen": -2.178088903427124,
|
| 522 |
+
"logits/rejected": -1.742640495300293,
|
| 523 |
+
"logps/chosen": -0.6361523270606995,
|
| 524 |
+
"logps/rejected": -1.0977928638458252,
|
| 525 |
+
"loss": 29.9436,
|
| 526 |
+
"nll_loss": 0.9138515591621399,
|
| 527 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 528 |
+
"rewards/chosen": -0.03180761635303497,
|
| 529 |
+
"rewards/margins": 0.023082025349140167,
|
| 530 |
+
"rewards/rejected": -0.05488964170217514,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.4222222222222223,
|
| 535 |
+
"grad_norm": 28.75,
|
| 536 |
+
"learning_rate": 3.7737761095632374e-05,
|
| 537 |
+
"log_odds_chosen": 0.8333398699760437,
|
| 538 |
+
"log_odds_ratio": -0.4819715619087219,
|
| 539 |
+
"logits/chosen": -2.0540931224823,
|
| 540 |
+
"logits/rejected": -2.0431177616119385,
|
| 541 |
+
"logps/chosen": -0.5949512124061584,
|
| 542 |
+
"logps/rejected": -0.9885191917419434,
|
| 543 |
+
"loss": 29.3897,
|
| 544 |
+
"nll_loss": 0.8951548337936401,
|
| 545 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 546 |
+
"rewards/chosen": -0.029747556895017624,
|
| 547 |
+
"rewards/margins": 0.019678404554724693,
|
| 548 |
+
"rewards/rejected": -0.049425967037677765,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4696296296296296,
|
| 553 |
+
"grad_norm": 28.5,
|
| 554 |
+
"learning_rate": 3.611785643555225e-05,
|
| 555 |
+
"log_odds_chosen": 0.8982599377632141,
|
| 556 |
+
"log_odds_ratio": -0.44926947355270386,
|
| 557 |
+
"logits/chosen": -2.2695367336273193,
|
| 558 |
+
"logits/rejected": -1.7630071640014648,
|
| 559 |
+
"logps/chosen": -0.6012392044067383,
|
| 560 |
+
"logps/rejected": -1.050581693649292,
|
| 561 |
+
"loss": 29.8918,
|
| 562 |
+
"nll_loss": 0.9108700752258301,
|
| 563 |
+
"rewards/accuracies": 0.78125,
|
| 564 |
+
"rewards/chosen": -0.030061960220336914,
|
| 565 |
+
"rewards/margins": 0.022467125207185745,
|
| 566 |
+
"rewards/rejected": -0.05252908915281296,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.5170370370370372,
|
| 571 |
+
"grad_norm": 30.625,
|
| 572 |
+
"learning_rate": 3.44791085752502e-05,
|
| 573 |
+
"log_odds_chosen": 0.8571161031723022,
|
| 574 |
+
"log_odds_ratio": -0.4356165826320648,
|
| 575 |
+
"logits/chosen": -2.093273639678955,
|
| 576 |
+
"logits/rejected": -2.1345930099487305,
|
| 577 |
+
"logps/chosen": -0.6451684236526489,
|
| 578 |
+
"logps/rejected": -1.1003749370574951,
|
| 579 |
+
"loss": 30.6067,
|
| 580 |
+
"nll_loss": 0.9544156193733215,
|
| 581 |
+
"rewards/accuracies": 0.84375,
|
| 582 |
+
"rewards/chosen": -0.03225841745734215,
|
| 583 |
+
"rewards/margins": 0.02276032790541649,
|
| 584 |
+
"rewards/rejected": -0.055018745362758636,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5644444444444443,
|
| 589 |
+
"grad_norm": 25.0,
|
| 590 |
+
"learning_rate": 3.2826564912351544e-05,
|
| 591 |
+
"log_odds_chosen": 0.9319137334823608,
|
| 592 |
+
"log_odds_ratio": -0.43036922812461853,
|
| 593 |
+
"logits/chosen": -1.9680538177490234,
|
| 594 |
+
"logits/rejected": -2.1300835609436035,
|
| 595 |
+
"logps/chosen": -0.6165143251419067,
|
| 596 |
+
"logps/rejected": -1.106245994567871,
|
| 597 |
+
"loss": 29.5351,
|
| 598 |
+
"nll_loss": 0.8909838795661926,
|
| 599 |
+
"rewards/accuracies": 0.8125,
|
| 600 |
+
"rewards/chosen": -0.030825715512037277,
|
| 601 |
+
"rewards/margins": 0.024486582726240158,
|
| 602 |
+
"rewards/rejected": -0.055312298238277435,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6118518518518519,
|
| 607 |
+
"grad_norm": 27.0,
|
| 608 |
+
"learning_rate": 3.116531533601003e-05,
|
| 609 |
+
"log_odds_chosen": 1.0328500270843506,
|
| 610 |
+
"log_odds_ratio": -0.4052696228027344,
|
| 611 |
+
"logits/chosen": -2.1744275093078613,
|
| 612 |
+
"logits/rejected": -1.9558875560760498,
|
| 613 |
+
"logps/chosen": -0.587931752204895,
|
| 614 |
+
"logps/rejected": -1.127718448638916,
|
| 615 |
+
"loss": 29.1635,
|
| 616 |
+
"nll_loss": 0.9016565084457397,
|
| 617 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 618 |
+
"rewards/chosen": -0.02939658798277378,
|
| 619 |
+
"rewards/margins": 0.02698933705687523,
|
| 620 |
+
"rewards/rejected": -0.05638592690229416,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6592592592592592,
|
| 625 |
+
"grad_norm": 29.875,
|
| 626 |
+
"learning_rate": 2.9500476549880848e-05,
|
| 627 |
+
"log_odds_chosen": 0.8978468179702759,
|
| 628 |
+
"log_odds_ratio": -0.45074257254600525,
|
| 629 |
+
"logits/chosen": -1.9579814672470093,
|
| 630 |
+
"logits/rejected": -1.6362476348876953,
|
| 631 |
+
"logps/chosen": -0.5840794444084167,
|
| 632 |
+
"logps/rejected": -1.0381691455841064,
|
| 633 |
+
"loss": 28.8906,
|
| 634 |
+
"nll_loss": 0.8974016308784485,
|
| 635 |
+
"rewards/accuracies": 0.800000011920929,
|
| 636 |
+
"rewards/chosen": -0.029203975573182106,
|
| 637 |
+
"rewards/margins": 0.022704491391777992,
|
| 638 |
+
"rewards/rejected": -0.0519084632396698,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7066666666666666,
|
| 643 |
+
"grad_norm": 31.75,
|
| 644 |
+
"learning_rate": 2.7837176312504037e-05,
|
| 645 |
+
"log_odds_chosen": 0.804090678691864,
|
| 646 |
+
"log_odds_ratio": -0.46772366762161255,
|
| 647 |
+
"logits/chosen": -1.7048003673553467,
|
| 648 |
+
"logits/rejected": -1.6802536249160767,
|
| 649 |
+
"logps/chosen": -0.6104881167411804,
|
| 650 |
+
"logps/rejected": -1.00138521194458,
|
| 651 |
+
"loss": 29.964,
|
| 652 |
+
"nll_loss": 0.9319046139717102,
|
| 653 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 654 |
+
"rewards/chosen": -0.03052440844476223,
|
| 655 |
+
"rewards/margins": 0.019544851034879684,
|
| 656 |
+
"rewards/rejected": -0.050069261342287064,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.7540740740740741,
|
| 661 |
+
"grad_norm": 24.25,
|
| 662 |
+
"learning_rate": 2.618053764363861e-05,
|
| 663 |
+
"log_odds_chosen": 0.9093812108039856,
|
| 664 |
+
"log_odds_ratio": -0.4248902201652527,
|
| 665 |
+
"logits/chosen": -2.1142642498016357,
|
| 666 |
+
"logits/rejected": -1.8935844898223877,
|
| 667 |
+
"logps/chosen": -0.5908008813858032,
|
| 668 |
+
"logps/rejected": -1.051133632659912,
|
| 669 |
+
"loss": 29.4169,
|
| 670 |
+
"nll_loss": 0.8831952810287476,
|
| 671 |
+
"rewards/accuracies": 0.831250011920929,
|
| 672 |
+
"rewards/chosen": -0.029540037736296654,
|
| 673 |
+
"rewards/margins": 0.023016640916466713,
|
| 674 |
+
"rewards/rejected": -0.052556682378053665,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.8014814814814815,
|
| 679 |
+
"grad_norm": 26.5,
|
| 680 |
+
"learning_rate": 2.453566304519216e-05,
|
| 681 |
+
"log_odds_chosen": 0.9450467228889465,
|
| 682 |
+
"log_odds_ratio": -0.4345301687717438,
|
| 683 |
+
"logits/chosen": -2.1107406616210938,
|
| 684 |
+
"logits/rejected": -1.6903254985809326,
|
| 685 |
+
"logps/chosen": -0.6346092820167542,
|
| 686 |
+
"logps/rejected": -1.1103118658065796,
|
| 687 |
+
"loss": 30.3828,
|
| 688 |
+
"nll_loss": 0.9147791862487793,
|
| 689 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 690 |
+
"rewards/chosen": -0.03173046559095383,
|
| 691 |
+
"rewards/margins": 0.02378513291478157,
|
| 692 |
+
"rewards/rejected": -0.0555155873298645,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8488888888888888,
|
| 697 |
+
"grad_norm": 30.5,
|
| 698 |
+
"learning_rate": 2.29076187853462e-05,
|
| 699 |
+
"log_odds_chosen": 0.9741055369377136,
|
| 700 |
+
"log_odds_ratio": -0.4274386465549469,
|
| 701 |
+
"logits/chosen": -2.0201268196105957,
|
| 702 |
+
"logits/rejected": -1.357716679573059,
|
| 703 |
+
"logps/chosen": -0.5966172218322754,
|
| 704 |
+
"logps/rejected": -1.1088060140609741,
|
| 705 |
+
"loss": 29.2191,
|
| 706 |
+
"nll_loss": 0.882198691368103,
|
| 707 |
+
"rewards/accuracies": 0.831250011920929,
|
| 708 |
+
"rewards/chosen": -0.02983086369931698,
|
| 709 |
+
"rewards/margins": 0.025609437376260757,
|
| 710 |
+
"rewards/rejected": -0.05544029921293259,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8962962962962964,
|
| 715 |
+
"grad_norm": 27.625,
|
| 716 |
+
"learning_rate": 2.130141929428254e-05,
|
| 717 |
+
"log_odds_chosen": 0.8030783534049988,
|
| 718 |
+
"log_odds_ratio": -0.4836719036102295,
|
| 719 |
+
"logits/chosen": -2.072028875350952,
|
| 720 |
+
"logits/rejected": -1.7954127788543701,
|
| 721 |
+
"logps/chosen": -0.6211769580841064,
|
| 722 |
+
"logps/rejected": -1.0174511671066284,
|
| 723 |
+
"loss": 31.2309,
|
| 724 |
+
"nll_loss": 0.9159282445907593,
|
| 725 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 726 |
+
"rewards/chosen": -0.031058847904205322,
|
| 727 |
+
"rewards/margins": 0.01981370709836483,
|
| 728 |
+
"rewards/rejected": -0.050872553139925,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9437037037037037,
|
| 733 |
+
"grad_norm": 25.5,
|
| 734 |
+
"learning_rate": 1.9722011719572444e-05,
|
| 735 |
+
"log_odds_chosen": 0.8332887887954712,
|
| 736 |
+
"log_odds_ratio": -0.46110400557518005,
|
| 737 |
+
"logits/chosen": -2.209178924560547,
|
| 738 |
+
"logits/rejected": -1.4569910764694214,
|
| 739 |
+
"logps/chosen": -0.614986777305603,
|
| 740 |
+
"logps/rejected": -1.0477509498596191,
|
| 741 |
+
"loss": 28.0719,
|
| 742 |
+
"nll_loss": 0.866260826587677,
|
| 743 |
+
"rewards/accuracies": 0.800000011920929,
|
| 744 |
+
"rewards/chosen": -0.03074934147298336,
|
| 745 |
+
"rewards/margins": 0.021638209000229836,
|
| 746 |
+
"rewards/rejected": -0.0523875467479229,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.991111111111111,
|
| 751 |
+
"grad_norm": 27.375,
|
| 752 |
+
"learning_rate": 1.8174260688798445e-05,
|
| 753 |
+
"log_odds_chosen": 0.7869575619697571,
|
| 754 |
+
"log_odds_ratio": -0.4784061312675476,
|
| 755 |
+
"logits/chosen": -1.8811867237091064,
|
| 756 |
+
"logits/rejected": -2.0677828788757324,
|
| 757 |
+
"logps/chosen": -0.5915592908859253,
|
| 758 |
+
"logps/rejected": -0.9474382400512695,
|
| 759 |
+
"loss": 28.0122,
|
| 760 |
+
"nll_loss": 0.8690091967582703,
|
| 761 |
+
"rewards/accuracies": 0.800000011920929,
|
| 762 |
+
"rewards/chosen": -0.029577964916825294,
|
| 763 |
+
"rewards/margins": 0.017793944105505943,
|
| 764 |
+
"rewards/rejected": -0.04737190902233124,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0385185185185186,
|
| 769 |
+
"grad_norm": 23.75,
|
| 770 |
+
"learning_rate": 1.666293332634042e-05,
|
| 771 |
+
"log_odds_chosen": 1.401928186416626,
|
| 772 |
+
"log_odds_ratio": -0.33335039019584656,
|
| 773 |
+
"logits/chosen": -1.884316086769104,
|
| 774 |
+
"logits/rejected": -1.4572067260742188,
|
| 775 |
+
"logps/chosen": -0.4979814887046814,
|
| 776 |
+
"logps/rejected": -1.1203409433364868,
|
| 777 |
+
"loss": 24.9695,
|
| 778 |
+
"nll_loss": 0.7534885406494141,
|
| 779 |
+
"rewards/accuracies": 0.887499988079071,
|
| 780 |
+
"rewards/chosen": -0.02489907667040825,
|
| 781 |
+
"rewards/margins": 0.03111797571182251,
|
| 782 |
+
"rewards/rejected": -0.05601705238223076,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.0859259259259257,
|
| 787 |
+
"grad_norm": 38.5,
|
| 788 |
+
"learning_rate": 1.519268457047482e-05,
|
| 789 |
+
"log_odds_chosen": 1.6839864253997803,
|
| 790 |
+
"log_odds_ratio": -0.2904171049594879,
|
| 791 |
+
"logits/chosen": -1.8016027212142944,
|
| 792 |
+
"logits/rejected": -1.8718116283416748,
|
| 793 |
+
"logps/chosen": -0.4482901096343994,
|
| 794 |
+
"logps/rejected": -1.188301682472229,
|
| 795 |
+
"loss": 23.232,
|
| 796 |
+
"nll_loss": 0.7302739024162292,
|
| 797 |
+
"rewards/accuracies": 0.887499988079071,
|
| 798 |
+
"rewards/chosen": -0.02241450548171997,
|
| 799 |
+
"rewards/margins": 0.03700058162212372,
|
| 800 |
+
"rewards/rejected": -0.05941509082913399,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1333333333333333,
|
| 805 |
+
"grad_norm": 27.5,
|
| 806 |
+
"learning_rate": 1.3768042836010768e-05,
|
| 807 |
+
"log_odds_chosen": 1.6373440027236938,
|
| 808 |
+
"log_odds_ratio": -0.2984515130519867,
|
| 809 |
+
"logits/chosen": -1.8258718252182007,
|
| 810 |
+
"logits/rejected": -1.6223289966583252,
|
| 811 |
+
"logps/chosen": -0.44031524658203125,
|
| 812 |
+
"logps/rejected": -1.1581284999847412,
|
| 813 |
+
"loss": 24.139,
|
| 814 |
+
"nll_loss": 0.7237830758094788,
|
| 815 |
+
"rewards/accuracies": 0.90625,
|
| 816 |
+
"rewards/chosen": -0.022015761584043503,
|
| 817 |
+
"rewards/margins": 0.035890672355890274,
|
| 818 |
+
"rewards/rejected": -0.05790643021464348,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.180740740740741,
|
| 823 |
+
"grad_norm": 27.5,
|
| 824 |
+
"learning_rate": 1.239339606662261e-05,
|
| 825 |
+
"log_odds_chosen": 1.801990270614624,
|
| 826 |
+
"log_odds_ratio": -0.25359493494033813,
|
| 827 |
+
"logits/chosen": -1.93035089969635,
|
| 828 |
+
"logits/rejected": -1.6016725301742554,
|
| 829 |
+
"logps/chosen": -0.4278429448604584,
|
| 830 |
+
"logps/rejected": -1.2159802913665771,
|
| 831 |
+
"loss": 22.8267,
|
| 832 |
+
"nll_loss": 0.7034687995910645,
|
| 833 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 834 |
+
"rewards/chosen": -0.02139214798808098,
|
| 835 |
+
"rewards/margins": 0.039406873285770416,
|
| 836 |
+
"rewards/rejected": -0.060799021273851395,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.228148148148148,
|
| 841 |
+
"grad_norm": 34.5,
|
| 842 |
+
"learning_rate": 1.1072978219838283e-05,
|
| 843 |
+
"log_odds_chosen": 1.565932035446167,
|
| 844 |
+
"log_odds_ratio": -0.3256310820579529,
|
| 845 |
+
"logits/chosen": -1.9141308069229126,
|
| 846 |
+
"logits/rejected": -1.976017951965332,
|
| 847 |
+
"logps/chosen": -0.4726741313934326,
|
| 848 |
+
"logps/rejected": -1.1302521228790283,
|
| 849 |
+
"loss": 23.1599,
|
| 850 |
+
"nll_loss": 0.7224346399307251,
|
| 851 |
+
"rewards/accuracies": 0.862500011920929,
|
| 852 |
+
"rewards/chosen": -0.02363370731472969,
|
| 853 |
+
"rewards/margins": 0.03287890553474426,
|
| 854 |
+
"rewards/rejected": -0.05651261284947395,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2755555555555556,
|
| 859 |
+
"grad_norm": 32.25,
|
| 860 |
+
"learning_rate": 9.810856226309972e-06,
|
| 861 |
+
"log_odds_chosen": 1.7595332860946655,
|
| 862 |
+
"log_odds_ratio": -0.2692697048187256,
|
| 863 |
+
"logits/chosen": -1.8822906017303467,
|
| 864 |
+
"logits/rejected": -1.698127031326294,
|
| 865 |
+
"logps/chosen": -0.430245578289032,
|
| 866 |
+
"logps/rejected": -1.2083370685577393,
|
| 867 |
+
"loss": 23.1595,
|
| 868 |
+
"nll_loss": 0.7202972173690796,
|
| 869 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 870 |
+
"rewards/chosen": -0.02151227928698063,
|
| 871 |
+
"rewards/margins": 0.0389045774936676,
|
| 872 |
+
"rewards/rejected": -0.06041685491800308,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.322962962962963,
|
| 877 |
+
"grad_norm": 26.5,
|
| 878 |
+
"learning_rate": 8.61091746353324e-06,
|
| 879 |
+
"log_odds_chosen": 1.702959418296814,
|
| 880 |
+
"log_odds_ratio": -0.2750469446182251,
|
| 881 |
+
"logits/chosen": -2.1500630378723145,
|
| 882 |
+
"logits/rejected": -1.591073751449585,
|
| 883 |
+
"logps/chosen": -0.4450320601463318,
|
| 884 |
+
"logps/rejected": -1.1653035879135132,
|
| 885 |
+
"loss": 23.0947,
|
| 886 |
+
"nll_loss": 0.7247873544692993,
|
| 887 |
+
"rewards/accuracies": 0.949999988079071,
|
| 888 |
+
"rewards/chosen": -0.022251605987548828,
|
| 889 |
+
"rewards/margins": 0.03601358085870743,
|
| 890 |
+
"rewards/rejected": -0.058265186846256256,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3703703703703702,
|
| 895 |
+
"grad_norm": 29.125,
|
| 896 |
+
"learning_rate": 7.47685778259568e-06,
|
| 897 |
+
"log_odds_chosen": 1.729418158531189,
|
| 898 |
+
"log_odds_ratio": -0.25526946783065796,
|
| 899 |
+
"logits/chosen": -1.865269660949707,
|
| 900 |
+
"logits/rejected": -1.8307578563690186,
|
| 901 |
+
"logps/chosen": -0.43531733751296997,
|
| 902 |
+
"logps/rejected": -1.197790503501892,
|
| 903 |
+
"loss": 22.4449,
|
| 904 |
+
"nll_loss": 0.6787145733833313,
|
| 905 |
+
"rewards/accuracies": 0.9375,
|
| 906 |
+
"rewards/chosen": -0.02176586538553238,
|
| 907 |
+
"rewards/margins": 0.038123659789562225,
|
| 908 |
+
"rewards/rejected": -0.059889525175094604,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.417777777777778,
|
| 913 |
+
"grad_norm": 27.125,
|
| 914 |
+
"learning_rate": 6.4121701248332905e-06,
|
| 915 |
+
"log_odds_chosen": 1.894997000694275,
|
| 916 |
+
"log_odds_ratio": -0.2565176784992218,
|
| 917 |
+
"logits/chosen": -1.9798578023910522,
|
| 918 |
+
"logits/rejected": -1.3841679096221924,
|
| 919 |
+
"logps/chosen": -0.3930845260620117,
|
| 920 |
+
"logps/rejected": -1.2068579196929932,
|
| 921 |
+
"loss": 22.3353,
|
| 922 |
+
"nll_loss": 0.6793208122253418,
|
| 923 |
+
"rewards/accuracies": 0.925000011920929,
|
| 924 |
+
"rewards/chosen": -0.019654225558042526,
|
| 925 |
+
"rewards/margins": 0.04068866744637489,
|
| 926 |
+
"rewards/rejected": -0.06034289672970772,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.4651851851851854,
|
| 931 |
+
"grad_norm": 29.625,
|
| 932 |
+
"learning_rate": 5.420133763455645e-06,
|
| 933 |
+
"log_odds_chosen": 1.909266710281372,
|
| 934 |
+
"log_odds_ratio": -0.25379735231399536,
|
| 935 |
+
"logits/chosen": -1.9765899181365967,
|
| 936 |
+
"logits/rejected": -1.7865279912948608,
|
| 937 |
+
"logps/chosen": -0.4143601059913635,
|
| 938 |
+
"logps/rejected": -1.225185751914978,
|
| 939 |
+
"loss": 22.3829,
|
| 940 |
+
"nll_loss": 0.6900944709777832,
|
| 941 |
+
"rewards/accuracies": 0.90625,
|
| 942 |
+
"rewards/chosen": -0.020718006417155266,
|
| 943 |
+
"rewards/margins": 0.04054127633571625,
|
| 944 |
+
"rewards/rejected": -0.06125928834080696,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5125925925925925,
|
| 949 |
+
"grad_norm": 32.25,
|
| 950 |
+
"learning_rate": 4.503804203275866e-06,
|
| 951 |
+
"log_odds_chosen": 1.7796869277954102,
|
| 952 |
+
"log_odds_ratio": -0.30406466126441956,
|
| 953 |
+
"logits/chosen": -1.8215770721435547,
|
| 954 |
+
"logits/rejected": -1.862717866897583,
|
| 955 |
+
"logps/chosen": -0.4358927607536316,
|
| 956 |
+
"logps/rejected": -1.197788953781128,
|
| 957 |
+
"loss": 22.2978,
|
| 958 |
+
"nll_loss": 0.6913371086120605,
|
| 959 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 960 |
+
"rewards/chosen": -0.0217946395277977,
|
| 961 |
+
"rewards/margins": 0.03809480741620064,
|
| 962 |
+
"rewards/rejected": -0.05988944694399834,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.56,
|
| 967 |
+
"grad_norm": 29.0,
|
| 968 |
+
"learning_rate": 3.6660037696547376e-06,
|
| 969 |
+
"log_odds_chosen": 1.7562096118927002,
|
| 970 |
+
"log_odds_ratio": -0.25910764932632446,
|
| 971 |
+
"logits/chosen": -2.1091978549957275,
|
| 972 |
+
"logits/rejected": -1.8953710794448853,
|
| 973 |
+
"logps/chosen": -0.4530642628669739,
|
| 974 |
+
"logps/rejected": -1.2265712022781372,
|
| 975 |
+
"loss": 23.2665,
|
| 976 |
+
"nll_loss": 0.7341790199279785,
|
| 977 |
+
"rewards/accuracies": 0.9375,
|
| 978 |
+
"rewards/chosen": -0.022653216496109962,
|
| 979 |
+
"rewards/margins": 0.03867534175515175,
|
| 980 |
+
"rewards/rejected": -0.06132856011390686,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.6074074074074076,
|
| 985 |
+
"grad_norm": 30.875,
|
| 986 |
+
"learning_rate": 2.909312915645238e-06,
|
| 987 |
+
"log_odds_chosen": 1.7647335529327393,
|
| 988 |
+
"log_odds_ratio": -0.28405773639678955,
|
| 989 |
+
"logits/chosen": -2.033613681793213,
|
| 990 |
+
"logits/rejected": -1.289603590965271,
|
| 991 |
+
"logps/chosen": -0.4545009732246399,
|
| 992 |
+
"logps/rejected": -1.2115771770477295,
|
| 993 |
+
"loss": 23.1922,
|
| 994 |
+
"nll_loss": 0.7150126695632935,
|
| 995 |
+
"rewards/accuracies": 0.918749988079071,
|
| 996 |
+
"rewards/chosen": -0.022725049406290054,
|
| 997 |
+
"rewards/margins": 0.03785381466150284,
|
| 998 |
+
"rewards/rejected": -0.060578860342502594,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.6548148148148147,
|
| 1003 |
+
"grad_norm": 34.75,
|
| 1004 |
+
"learning_rate": 2.236062274111741e-06,
|
| 1005 |
+
"log_odds_chosen": 1.6408923864364624,
|
| 1006 |
+
"log_odds_ratio": -0.2776089906692505,
|
| 1007 |
+
"logits/chosen": -1.8170640468597412,
|
| 1008 |
+
"logits/rejected": -1.9727897644042969,
|
| 1009 |
+
"logps/chosen": -0.4261111319065094,
|
| 1010 |
+
"logps/rejected": -1.1319156885147095,
|
| 1011 |
+
"loss": 22.1786,
|
| 1012 |
+
"nll_loss": 0.6713584661483765,
|
| 1013 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1014 |
+
"rewards/chosen": -0.02130555734038353,
|
| 1015 |
+
"rewards/margins": 0.035290226340293884,
|
| 1016 |
+
"rewards/rejected": -0.056595779955387115,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.7022222222222223,
|
| 1021 |
+
"grad_norm": 30.75,
|
| 1022 |
+
"learning_rate": 1.648325479303684e-06,
|
| 1023 |
+
"log_odds_chosen": 1.6113086938858032,
|
| 1024 |
+
"log_odds_ratio": -0.2935238778591156,
|
| 1025 |
+
"logits/chosen": -2.0707173347473145,
|
| 1026 |
+
"logits/rejected": -1.4977672100067139,
|
| 1027 |
+
"logps/chosen": -0.4346179962158203,
|
| 1028 |
+
"logps/rejected": -1.1482003927230835,
|
| 1029 |
+
"loss": 23.1607,
|
| 1030 |
+
"nll_loss": 0.6985403895378113,
|
| 1031 |
+
"rewards/accuracies": 0.918749988079071,
|
| 1032 |
+
"rewards/chosen": -0.021730897948145866,
|
| 1033 |
+
"rewards/margins": 0.03567912429571152,
|
| 1034 |
+
"rewards/rejected": -0.05741002410650253,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.74962962962963,
|
| 1039 |
+
"grad_norm": 29.75,
|
| 1040 |
+
"learning_rate": 1.1479127799935029e-06,
|
| 1041 |
+
"log_odds_chosen": 1.8265297412872314,
|
| 1042 |
+
"log_odds_ratio": -0.2631281614303589,
|
| 1043 |
+
"logits/chosen": -1.841491937637329,
|
| 1044 |
+
"logits/rejected": -1.922586441040039,
|
| 1045 |
+
"logps/chosen": -0.4327624440193176,
|
| 1046 |
+
"logps/rejected": -1.2341258525848389,
|
| 1047 |
+
"loss": 22.9795,
|
| 1048 |
+
"nll_loss": 0.7204877734184265,
|
| 1049 |
+
"rewards/accuracies": 0.9375,
|
| 1050 |
+
"rewards/chosen": -0.021638119593262672,
|
| 1051 |
+
"rewards/margins": 0.04006817191839218,
|
| 1052 |
+
"rewards/rejected": -0.0617062933743,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.797037037037037,
|
| 1057 |
+
"grad_norm": 31.625,
|
| 1058 |
+
"learning_rate": 7.363654638505046e-07,
|
| 1059 |
+
"log_odds_chosen": 1.7081098556518555,
|
| 1060 |
+
"log_odds_ratio": -0.29199516773223877,
|
| 1061 |
+
"logits/chosen": -1.7930389642715454,
|
| 1062 |
+
"logits/rejected": -1.7181438207626343,
|
| 1063 |
+
"logps/chosen": -0.449666827917099,
|
| 1064 |
+
"logps/rejected": -1.2206642627716064,
|
| 1065 |
+
"loss": 22.9709,
|
| 1066 |
+
"nll_loss": 0.7106753587722778,
|
| 1067 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1068 |
+
"rewards/chosen": -0.0224833432585001,
|
| 1069 |
+
"rewards/margins": 0.03854987770318985,
|
| 1070 |
+
"rewards/rejected": -0.0610332190990448,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.8444444444444446,
|
| 1075 |
+
"grad_norm": 31.25,
|
| 1076 |
+
"learning_rate": 4.149511102238568e-07,
|
| 1077 |
+
"log_odds_chosen": 1.5754259824752808,
|
| 1078 |
+
"log_odds_ratio": -0.3034347891807556,
|
| 1079 |
+
"logits/chosen": -2.2559409141540527,
|
| 1080 |
+
"logits/rejected": -1.71217942237854,
|
| 1081 |
+
"logps/chosen": -0.46836423873901367,
|
| 1082 |
+
"logps/rejected": -1.2223981618881226,
|
| 1083 |
+
"loss": 22.8601,
|
| 1084 |
+
"nll_loss": 0.7257949113845825,
|
| 1085 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1086 |
+
"rewards/chosen": -0.023418214172124863,
|
| 1087 |
+
"rewards/margins": 0.037701696157455444,
|
| 1088 |
+
"rewards/rejected": -0.06111990660429001,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.891851851851852,
|
| 1093 |
+
"grad_norm": 28.625,
|
| 1094 |
+
"learning_rate": 1.8465968595625105e-07,
|
| 1095 |
+
"log_odds_chosen": 1.6639511585235596,
|
| 1096 |
+
"log_odds_ratio": -0.2808656096458435,
|
| 1097 |
+
"logits/chosen": -2.1249499320983887,
|
| 1098 |
+
"logits/rejected": -1.6745857000350952,
|
| 1099 |
+
"logps/chosen": -0.475193589925766,
|
| 1100 |
+
"logps/rejected": -1.1988952159881592,
|
| 1101 |
+
"loss": 21.8942,
|
| 1102 |
+
"nll_loss": 0.6743995547294617,
|
| 1103 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1104 |
+
"rewards/chosen": -0.02375968173146248,
|
| 1105 |
+
"rewards/margins": 0.03618507459759712,
|
| 1106 |
+
"rewards/rejected": -0.0599447600543499,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9392592592592592,
|
| 1111 |
+
"grad_norm": 30.75,
|
| 1112 |
+
"learning_rate": 4.620049625329803e-08,
|
| 1113 |
+
"log_odds_chosen": 1.7966537475585938,
|
| 1114 |
+
"log_odds_ratio": -0.25377795100212097,
|
| 1115 |
+
"logits/chosen": -1.9389528036117554,
|
| 1116 |
+
"logits/rejected": -1.399864912033081,
|
| 1117 |
+
"logps/chosen": -0.4379648268222809,
|
| 1118 |
+
"logps/rejected": -1.1957590579986572,
|
| 1119 |
+
"loss": 22.7814,
|
| 1120 |
+
"nll_loss": 0.6846402883529663,
|
| 1121 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 1122 |
+
"rewards/chosen": -0.021898243576288223,
|
| 1123 |
+
"rewards/margins": 0.03788971155881882,
|
| 1124 |
+
"rewards/rejected": -0.05978795886039734,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.986666666666667,
|
| 1129 |
+
"grad_norm": 32.75,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 1.8841956853866577,
|
| 1132 |
+
"log_odds_ratio": -0.24858447909355164,
|
| 1133 |
+
"logits/chosen": -1.8983337879180908,
|
| 1134 |
+
"logits/rejected": -1.5074989795684814,
|
| 1135 |
+
"logps/chosen": -0.40289902687072754,
|
| 1136 |
+
"logps/rejected": -1.209084391593933,
|
| 1137 |
+
"loss": 22.233,
|
| 1138 |
+
"nll_loss": 0.6980301737785339,
|
| 1139 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1140 |
+
"rewards/chosen": -0.020144950598478317,
|
| 1141 |
+
"rewards/margins": 0.040309272706508636,
|
| 1142 |
+
"rewards/rejected": -0.06045422703027725,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.986666666666667,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 31.135096304757255,
|
| 1150 |
+
"train_runtime": 6745.6063,
|
| 1151 |
+
"train_samples_per_second": 3.002,
|
| 1152 |
+
"train_steps_per_second": 0.047
|
| 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 |
+
}
|