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-borpo-noisy-5e-5-02-norm2
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for mistral-7b-borpo-noisy-5e-5-02-norm2
|
| 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-borpo-noisy-5e-5-02-norm2", 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/bz4a6nik)
|
| 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: 4.48.1
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.2.0
|
| 41 |
+
- Tokenizers: 0.21.0
|
| 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": 2.9765925925925925,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 26.480164833674355,
|
| 5 |
+
"train_runtime": 6938.2073,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.919,
|
| 8 |
+
"train_steps_per_second": 0.045
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.9765925925925925,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 26.480164833674355,
|
| 5 |
+
"train_runtime": 6938.2073,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.919,
|
| 8 |
+
"train_steps_per_second": 0.045
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.9765925925925925,
|
| 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": 108.05298614501953,
|
| 14 |
+
"learning_rate": 7.8125e-06,
|
| 15 |
+
"log_odds_chosen": 0.27033674716949463,
|
| 16 |
+
"log_odds_ratio": -0.6787606477737427,
|
| 17 |
+
"logits/chosen": -2.960299253463745,
|
| 18 |
+
"logits/rejected": -2.877185106277466,
|
| 19 |
+
"logps/chosen": -1.1196008920669556,
|
| 20 |
+
"logps/rejected": -1.324358582496643,
|
| 21 |
+
"loss": 47.1436,
|
| 22 |
+
"nll_loss": 1.4213258028030396,
|
| 23 |
+
"rewards/accuracies": 0.574999988079071,
|
| 24 |
+
"rewards/chosen": -0.22392019629478455,
|
| 25 |
+
"rewards/margins": 0.04095155745744705,
|
| 26 |
+
"rewards/rejected": -0.2648717164993286,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 60.31058883666992,
|
| 32 |
+
"learning_rate": 1.5625e-05,
|
| 33 |
+
"log_odds_chosen": 0.1787528693675995,
|
| 34 |
+
"log_odds_ratio": -0.6780639886856079,
|
| 35 |
+
"logits/chosen": -2.965952157974243,
|
| 36 |
+
"logits/rejected": -2.846004009246826,
|
| 37 |
+
"logps/chosen": -0.9941836595535278,
|
| 38 |
+
"logps/rejected": -1.1146419048309326,
|
| 39 |
+
"loss": 41.0236,
|
| 40 |
+
"nll_loss": 1.2662255764007568,
|
| 41 |
+
"rewards/accuracies": 0.59375,
|
| 42 |
+
"rewards/chosen": -0.19883672893047333,
|
| 43 |
+
"rewards/margins": 0.02409166470170021,
|
| 44 |
+
"rewards/rejected": -0.22292840480804443,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 64.69369506835938,
|
| 50 |
+
"learning_rate": 2.34375e-05,
|
| 51 |
+
"log_odds_chosen": 0.1927437037229538,
|
| 52 |
+
"log_odds_ratio": -0.6706465482711792,
|
| 53 |
+
"logits/chosen": -2.8047168254852295,
|
| 54 |
+
"logits/rejected": -2.7224316596984863,
|
| 55 |
+
"logps/chosen": -0.7822418212890625,
|
| 56 |
+
"logps/rejected": -0.890445351600647,
|
| 57 |
+
"loss": 38.9336,
|
| 58 |
+
"nll_loss": 1.2353577613830566,
|
| 59 |
+
"rewards/accuracies": 0.581250011920929,
|
| 60 |
+
"rewards/chosen": -0.1564483642578125,
|
| 61 |
+
"rewards/margins": 0.021640703082084656,
|
| 62 |
+
"rewards/rejected": -0.17808905243873596,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 47.254459381103516,
|
| 68 |
+
"learning_rate": 3.125e-05,
|
| 69 |
+
"log_odds_chosen": 0.1654689460992813,
|
| 70 |
+
"log_odds_ratio": -0.6847632527351379,
|
| 71 |
+
"logits/chosen": -2.812252998352051,
|
| 72 |
+
"logits/rejected": -2.759857654571533,
|
| 73 |
+
"logps/chosen": -0.806632399559021,
|
| 74 |
+
"logps/rejected": -0.9170172810554504,
|
| 75 |
+
"loss": 36.5821,
|
| 76 |
+
"nll_loss": 1.121074914932251,
|
| 77 |
+
"rewards/accuracies": 0.581250011920929,
|
| 78 |
+
"rewards/chosen": -0.16132649779319763,
|
| 79 |
+
"rewards/margins": 0.02207697369158268,
|
| 80 |
+
"rewards/rejected": -0.18340346217155457,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 32.87034606933594,
|
| 86 |
+
"learning_rate": 3.90625e-05,
|
| 87 |
+
"log_odds_chosen": 0.11796001344919205,
|
| 88 |
+
"log_odds_ratio": -0.7380820512771606,
|
| 89 |
+
"logits/chosen": -2.8849737644195557,
|
| 90 |
+
"logits/rejected": -2.837324857711792,
|
| 91 |
+
"logps/chosen": -0.796675443649292,
|
| 92 |
+
"logps/rejected": -0.8797528147697449,
|
| 93 |
+
"loss": 35.6029,
|
| 94 |
+
"nll_loss": 1.143587350845337,
|
| 95 |
+
"rewards/accuracies": 0.518750011920929,
|
| 96 |
+
"rewards/chosen": -0.15933510661125183,
|
| 97 |
+
"rewards/margins": 0.0166154894977808,
|
| 98 |
+
"rewards/rejected": -0.17595058679580688,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 44.035709381103516,
|
| 104 |
+
"learning_rate": 4.6875e-05,
|
| 105 |
+
"log_odds_chosen": 0.11726967990398407,
|
| 106 |
+
"log_odds_ratio": -0.7224973440170288,
|
| 107 |
+
"logits/chosen": -2.8627395629882812,
|
| 108 |
+
"logits/rejected": -2.7379047870635986,
|
| 109 |
+
"logps/chosen": -0.7604724764823914,
|
| 110 |
+
"logps/rejected": -0.8269100189208984,
|
| 111 |
+
"loss": 34.6175,
|
| 112 |
+
"nll_loss": 1.0858324766159058,
|
| 113 |
+
"rewards/accuracies": 0.59375,
|
| 114 |
+
"rewards/chosen": -0.1520944982767105,
|
| 115 |
+
"rewards/margins": 0.013287502340972424,
|
| 116 |
+
"rewards/rejected": -0.1653819978237152,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 35.993125915527344,
|
| 122 |
+
"learning_rate": 4.998613757348784e-05,
|
| 123 |
+
"log_odds_chosen": 0.11861761659383774,
|
| 124 |
+
"log_odds_ratio": -0.7235144972801208,
|
| 125 |
+
"logits/chosen": -2.8734326362609863,
|
| 126 |
+
"logits/rejected": -2.924297571182251,
|
| 127 |
+
"logps/chosen": -0.7806745171546936,
|
| 128 |
+
"logps/rejected": -0.8397527933120728,
|
| 129 |
+
"loss": 33.9132,
|
| 130 |
+
"nll_loss": 1.096482515335083,
|
| 131 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 132 |
+
"rewards/chosen": -0.15613490343093872,
|
| 133 |
+
"rewards/margins": 0.01181565411388874,
|
| 134 |
+
"rewards/rejected": -0.1679505705833435,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 31.309890747070312,
|
| 140 |
+
"learning_rate": 4.990147841143462e-05,
|
| 141 |
+
"log_odds_chosen": 0.1591743528842926,
|
| 142 |
+
"log_odds_ratio": -0.6983242034912109,
|
| 143 |
+
"logits/chosen": -2.8734216690063477,
|
| 144 |
+
"logits/rejected": -2.8256828784942627,
|
| 145 |
+
"logps/chosen": -0.7766333818435669,
|
| 146 |
+
"logps/rejected": -0.8613529205322266,
|
| 147 |
+
"loss": 34.0145,
|
| 148 |
+
"nll_loss": 1.0602055788040161,
|
| 149 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 150 |
+
"rewards/chosen": -0.1553266942501068,
|
| 151 |
+
"rewards/margins": 0.016943905502557755,
|
| 152 |
+
"rewards/rejected": -0.17227059602737427,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 26.746047973632812,
|
| 158 |
+
"learning_rate": 4.97401218720448e-05,
|
| 159 |
+
"log_odds_chosen": 0.147725909948349,
|
| 160 |
+
"log_odds_ratio": -0.7261999249458313,
|
| 161 |
+
"logits/chosen": -2.816972017288208,
|
| 162 |
+
"logits/rejected": -2.8251900672912598,
|
| 163 |
+
"logps/chosen": -0.7309020161628723,
|
| 164 |
+
"logps/rejected": -0.8109971880912781,
|
| 165 |
+
"loss": 33.7617,
|
| 166 |
+
"nll_loss": 1.0613343715667725,
|
| 167 |
+
"rewards/accuracies": 0.5,
|
| 168 |
+
"rewards/chosen": -0.1461803913116455,
|
| 169 |
+
"rewards/margins": 0.016019051894545555,
|
| 170 |
+
"rewards/rejected": -0.1621994525194168,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 31.442655563354492,
|
| 176 |
+
"learning_rate": 4.9502564938797946e-05,
|
| 177 |
+
"log_odds_chosen": 0.11614716053009033,
|
| 178 |
+
"log_odds_ratio": -0.7367695569992065,
|
| 179 |
+
"logits/chosen": -2.8157620429992676,
|
| 180 |
+
"logits/rejected": -2.8260910511016846,
|
| 181 |
+
"logps/chosen": -0.7965345978736877,
|
| 182 |
+
"logps/rejected": -0.8452186584472656,
|
| 183 |
+
"loss": 35.2613,
|
| 184 |
+
"nll_loss": 1.1184347867965698,
|
| 185 |
+
"rewards/accuracies": 0.53125,
|
| 186 |
+
"rewards/chosen": -0.15930691361427307,
|
| 187 |
+
"rewards/margins": 0.009736819192767143,
|
| 188 |
+
"rewards/rejected": -0.16904373466968536,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 25.172645568847656,
|
| 194 |
+
"learning_rate": 4.918953929490768e-05,
|
| 195 |
+
"log_odds_chosen": 0.09427499771118164,
|
| 196 |
+
"log_odds_ratio": -0.7385081052780151,
|
| 197 |
+
"logits/chosen": -2.794956684112549,
|
| 198 |
+
"logits/rejected": -2.7535877227783203,
|
| 199 |
+
"logps/chosen": -0.7553600072860718,
|
| 200 |
+
"logps/rejected": -0.8219752311706543,
|
| 201 |
+
"loss": 32.9776,
|
| 202 |
+
"nll_loss": 1.0361286401748657,
|
| 203 |
+
"rewards/accuracies": 0.53125,
|
| 204 |
+
"rewards/chosen": -0.15107201039791107,
|
| 205 |
+
"rewards/margins": 0.013323022052645683,
|
| 206 |
+
"rewards/rejected": -0.1643950492143631,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 27.215576171875,
|
| 212 |
+
"learning_rate": 4.88020090697132e-05,
|
| 213 |
+
"log_odds_chosen": 0.17814065515995026,
|
| 214 |
+
"log_odds_ratio": -0.6895138025283813,
|
| 215 |
+
"logits/chosen": -2.9259557723999023,
|
| 216 |
+
"logits/rejected": -2.827220916748047,
|
| 217 |
+
"logps/chosen": -0.7460590600967407,
|
| 218 |
+
"logps/rejected": -0.828277587890625,
|
| 219 |
+
"loss": 32.1892,
|
| 220 |
+
"nll_loss": 1.0066068172454834,
|
| 221 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 222 |
+
"rewards/chosen": -0.1492118090391159,
|
| 223 |
+
"rewards/margins": 0.016443701460957527,
|
| 224 |
+
"rewards/rejected": -0.16565552353858948,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 25.96778106689453,
|
| 230 |
+
"learning_rate": 4.834116786912897e-05,
|
| 231 |
+
"log_odds_chosen": 0.14030879735946655,
|
| 232 |
+
"log_odds_ratio": -0.7043845057487488,
|
| 233 |
+
"logits/chosen": -2.836535692214966,
|
| 234 |
+
"logits/rejected": -2.8294970989227295,
|
| 235 |
+
"logps/chosen": -0.7776893377304077,
|
| 236 |
+
"logps/rejected": -0.8747210502624512,
|
| 237 |
+
"loss": 32.3069,
|
| 238 |
+
"nll_loss": 1.0067697763442993,
|
| 239 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 240 |
+
"rewards/chosen": -0.15553787350654602,
|
| 241 |
+
"rewards/margins": 0.019406333565711975,
|
| 242 |
+
"rewards/rejected": -0.174944207072258,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 25.73870849609375,
|
| 248 |
+
"learning_rate": 4.7808435099299045e-05,
|
| 249 |
+
"log_odds_chosen": 0.10110124200582504,
|
| 250 |
+
"log_odds_ratio": -0.7182499170303345,
|
| 251 |
+
"logits/chosen": -2.9134488105773926,
|
| 252 |
+
"logits/rejected": -2.7823774814605713,
|
| 253 |
+
"logps/chosen": -0.7712094187736511,
|
| 254 |
+
"logps/rejected": -0.8390757441520691,
|
| 255 |
+
"loss": 31.6312,
|
| 256 |
+
"nll_loss": 1.0089954137802124,
|
| 257 |
+
"rewards/accuracies": 0.581250011920929,
|
| 258 |
+
"rewards/chosen": -0.1542418897151947,
|
| 259 |
+
"rewards/margins": 0.013573260977864265,
|
| 260 |
+
"rewards/rejected": -0.167815163731575,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 27.71648597717285,
|
| 266 |
+
"learning_rate": 4.720545159477922e-05,
|
| 267 |
+
"log_odds_chosen": 0.32593604922294617,
|
| 268 |
+
"log_odds_ratio": -0.6149824857711792,
|
| 269 |
+
"logits/chosen": -2.844346284866333,
|
| 270 |
+
"logits/rejected": -2.8723151683807373,
|
| 271 |
+
"logps/chosen": -0.6918070316314697,
|
| 272 |
+
"logps/rejected": -0.8593597412109375,
|
| 273 |
+
"loss": 32.1098,
|
| 274 |
+
"nll_loss": 1.009124755859375,
|
| 275 |
+
"rewards/accuracies": 0.6875,
|
| 276 |
+
"rewards/chosen": -0.13836140930652618,
|
| 277 |
+
"rewards/margins": 0.03351055458188057,
|
| 278 |
+
"rewards/rejected": -0.17187196016311646,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 25.093658447265625,
|
| 284 |
+
"learning_rate": 4.653407456471222e-05,
|
| 285 |
+
"log_odds_chosen": 0.1939249038696289,
|
| 286 |
+
"log_odds_ratio": -0.686955451965332,
|
| 287 |
+
"logits/chosen": -2.961365222930908,
|
| 288 |
+
"logits/rejected": -2.869180917739868,
|
| 289 |
+
"logps/chosen": -0.7829879522323608,
|
| 290 |
+
"logps/rejected": -0.8784705400466919,
|
| 291 |
+
"loss": 31.996,
|
| 292 |
+
"nll_loss": 1.0403110980987549,
|
| 293 |
+
"rewards/accuracies": 0.637499988079071,
|
| 294 |
+
"rewards/chosen": -0.1565975844860077,
|
| 295 |
+
"rewards/margins": 0.019096530973911285,
|
| 296 |
+
"rewards/rejected": -0.17569413781166077,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 21.968870162963867,
|
| 302 |
+
"learning_rate": 4.579637187256222e-05,
|
| 303 |
+
"log_odds_chosen": 0.15405571460723877,
|
| 304 |
+
"log_odds_ratio": -0.691379189491272,
|
| 305 |
+
"logits/chosen": -2.946697473526001,
|
| 306 |
+
"logits/rejected": -2.7557473182678223,
|
| 307 |
+
"logps/chosen": -0.7400744557380676,
|
| 308 |
+
"logps/rejected": -0.8276380300521851,
|
| 309 |
+
"loss": 31.9056,
|
| 310 |
+
"nll_loss": 1.0012738704681396,
|
| 311 |
+
"rewards/accuracies": 0.581250011920929,
|
| 312 |
+
"rewards/chosen": -0.14801490306854248,
|
| 313 |
+
"rewards/margins": 0.017512720078229904,
|
| 314 |
+
"rewards/rejected": -0.16552762687206268,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 24.323413848876953,
|
| 320 |
+
"learning_rate": 4.499461566702685e-05,
|
| 321 |
+
"log_odds_chosen": 0.2325698584318161,
|
| 322 |
+
"log_odds_ratio": -0.6755145788192749,
|
| 323 |
+
"logits/chosen": -2.9155423641204834,
|
| 324 |
+
"logits/rejected": -2.8289742469787598,
|
| 325 |
+
"logps/chosen": -0.7051252126693726,
|
| 326 |
+
"logps/rejected": -0.8203322291374207,
|
| 327 |
+
"loss": 32.5934,
|
| 328 |
+
"nll_loss": 0.9984562993049622,
|
| 329 |
+
"rewards/accuracies": 0.612500011920929,
|
| 330 |
+
"rewards/chosen": -0.14102503657341003,
|
| 331 |
+
"rewards/margins": 0.023041393607854843,
|
| 332 |
+
"rewards/rejected": -0.16406643390655518,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 23.40336036682129,
|
| 338 |
+
"learning_rate": 4.413127538374411e-05,
|
| 339 |
+
"log_odds_chosen": 0.25212088227272034,
|
| 340 |
+
"log_odds_ratio": -0.6699780225753784,
|
| 341 |
+
"logits/chosen": -2.8937487602233887,
|
| 342 |
+
"logits/rejected": -2.7795820236206055,
|
| 343 |
+
"logps/chosen": -0.6867400407791138,
|
| 344 |
+
"logps/rejected": -0.8178392648696899,
|
| 345 |
+
"loss": 32.0021,
|
| 346 |
+
"nll_loss": 0.9535757899284363,
|
| 347 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 348 |
+
"rewards/chosen": -0.1373480260372162,
|
| 349 |
+
"rewards/margins": 0.026219844818115234,
|
| 350 |
+
"rewards/rejected": -0.16356787085533142,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 23.989849090576172,
|
| 356 |
+
"learning_rate": 4.320901013934887e-05,
|
| 357 |
+
"log_odds_chosen": 0.05919067934155464,
|
| 358 |
+
"log_odds_ratio": -0.7570820450782776,
|
| 359 |
+
"logits/chosen": -2.867305040359497,
|
| 360 |
+
"logits/rejected": -2.7521088123321533,
|
| 361 |
+
"logps/chosen": -0.7711822986602783,
|
| 362 |
+
"logps/rejected": -0.8085872530937195,
|
| 363 |
+
"loss": 32.0116,
|
| 364 |
+
"nll_loss": 1.0364195108413696,
|
| 365 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 366 |
+
"rewards/chosen": -0.15423646569252014,
|
| 367 |
+
"rewards/margins": 0.007480976637452841,
|
| 368 |
+
"rewards/rejected": -0.16171741485595703,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 23.476768493652344,
|
| 374 |
+
"learning_rate": 4.223066054130568e-05,
|
| 375 |
+
"log_odds_chosen": 0.15499885380268097,
|
| 376 |
+
"log_odds_ratio": -0.7011421918869019,
|
| 377 |
+
"logits/chosen": -2.934399127960205,
|
| 378 |
+
"logits/rejected": -2.7926974296569824,
|
| 379 |
+
"logps/chosen": -0.7334684729576111,
|
| 380 |
+
"logps/rejected": -0.8502451181411743,
|
| 381 |
+
"loss": 31.8454,
|
| 382 |
+
"nll_loss": 0.992477297782898,
|
| 383 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 384 |
+
"rewards/chosen": -0.14669370651245117,
|
| 385 |
+
"rewards/margins": 0.02335532382130623,
|
| 386 |
+
"rewards/rejected": -0.1700490266084671,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.037925925925926,
|
| 391 |
+
"grad_norm": 24.518178939819336,
|
| 392 |
+
"learning_rate": 4.1199239938743797e-05,
|
| 393 |
+
"log_odds_chosen": 0.48953691124916077,
|
| 394 |
+
"log_odds_ratio": -0.5804011225700378,
|
| 395 |
+
"logits/chosen": -2.850722074508667,
|
| 396 |
+
"logits/rejected": -2.777690887451172,
|
| 397 |
+
"logps/chosen": -0.5838567614555359,
|
| 398 |
+
"logps/rejected": -0.8051884174346924,
|
| 399 |
+
"loss": 24.0268,
|
| 400 |
+
"nll_loss": 0.8107532262802124,
|
| 401 |
+
"rewards/accuracies": 0.7062937021255493,
|
| 402 |
+
"rewards/chosen": -0.11677134782075882,
|
| 403 |
+
"rewards/margins": 0.04426633194088936,
|
| 404 |
+
"rewards/rejected": -0.16103769838809967,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0853333333333333,
|
| 409 |
+
"grad_norm": 27.153846740722656,
|
| 410 |
+
"learning_rate": 4.0117925141242174e-05,
|
| 411 |
+
"log_odds_chosen": 0.6579824686050415,
|
| 412 |
+
"log_odds_ratio": -0.5037034749984741,
|
| 413 |
+
"logits/chosen": -2.88224458694458,
|
| 414 |
+
"logits/rejected": -2.683182954788208,
|
| 415 |
+
"logps/chosen": -0.5709269046783447,
|
| 416 |
+
"logps/rejected": -0.8794952630996704,
|
| 417 |
+
"loss": 25.9524,
|
| 418 |
+
"nll_loss": 0.7917619347572327,
|
| 419 |
+
"rewards/accuracies": 0.793749988079071,
|
| 420 |
+
"rewards/chosen": -0.1141853779554367,
|
| 421 |
+
"rewards/margins": 0.06171366572380066,
|
| 422 |
+
"rewards/rejected": -0.17589905858039856,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1327407407407408,
|
| 427 |
+
"grad_norm": 24.396657943725586,
|
| 428 |
+
"learning_rate": 3.899004663415084e-05,
|
| 429 |
+
"log_odds_chosen": 0.7225304841995239,
|
| 430 |
+
"log_odds_ratio": -0.49574628472328186,
|
| 431 |
+
"logits/chosen": -2.8413453102111816,
|
| 432 |
+
"logits/rejected": -2.783343553543091,
|
| 433 |
+
"logps/chosen": -0.5242325067520142,
|
| 434 |
+
"logps/rejected": -0.8295241594314575,
|
| 435 |
+
"loss": 25.8745,
|
| 436 |
+
"nll_loss": 0.786653995513916,
|
| 437 |
+
"rewards/accuracies": 0.762499988079071,
|
| 438 |
+
"rewards/chosen": -0.1048465147614479,
|
| 439 |
+
"rewards/margins": 0.06105831265449524,
|
| 440 |
+
"rewards/rejected": -0.16590481996536255,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1801481481481482,
|
| 445 |
+
"grad_norm": 25.048795700073242,
|
| 446 |
+
"learning_rate": 3.781907832058587e-05,
|
| 447 |
+
"log_odds_chosen": 0.7397602796554565,
|
| 448 |
+
"log_odds_ratio": -0.4922989308834076,
|
| 449 |
+
"logits/chosen": -2.8375182151794434,
|
| 450 |
+
"logits/rejected": -2.7994544506073,
|
| 451 |
+
"logps/chosen": -0.5756106972694397,
|
| 452 |
+
"logps/rejected": -0.9090244174003601,
|
| 453 |
+
"loss": 24.2883,
|
| 454 |
+
"nll_loss": 0.7707120180130005,
|
| 455 |
+
"rewards/accuracies": 0.793749988079071,
|
| 456 |
+
"rewards/chosen": -0.11512213945388794,
|
| 457 |
+
"rewards/margins": 0.06668277084827423,
|
| 458 |
+
"rewards/rejected": -0.18180491030216217,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2275555555555555,
|
| 463 |
+
"grad_norm": 24.017641067504883,
|
| 464 |
+
"learning_rate": 3.660862682169282e-05,
|
| 465 |
+
"log_odds_chosen": 0.619869589805603,
|
| 466 |
+
"log_odds_ratio": -0.5279988050460815,
|
| 467 |
+
"logits/chosen": -2.8795347213745117,
|
| 468 |
+
"logits/rejected": -2.7905712127685547,
|
| 469 |
+
"logps/chosen": -0.5735269784927368,
|
| 470 |
+
"logps/rejected": -0.8608913421630859,
|
| 471 |
+
"loss": 24.6652,
|
| 472 |
+
"nll_loss": 0.7880049347877502,
|
| 473 |
+
"rewards/accuracies": 0.737500011920929,
|
| 474 |
+
"rewards/chosen": -0.11470538377761841,
|
| 475 |
+
"rewards/margins": 0.057472895830869675,
|
| 476 |
+
"rewards/rejected": -0.17217828333377838,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.274962962962963,
|
| 481 |
+
"grad_norm": 23.851089477539062,
|
| 482 |
+
"learning_rate": 3.5362420368134356e-05,
|
| 483 |
+
"log_odds_chosen": 0.685693085193634,
|
| 484 |
+
"log_odds_ratio": -0.5290096998214722,
|
| 485 |
+
"logits/chosen": -2.858466625213623,
|
| 486 |
+
"logits/rejected": -2.7995998859405518,
|
| 487 |
+
"logps/chosen": -0.5094814896583557,
|
| 488 |
+
"logps/rejected": -0.7905184030532837,
|
| 489 |
+
"loss": 24.8932,
|
| 490 |
+
"nll_loss": 0.7591756582260132,
|
| 491 |
+
"rewards/accuracies": 0.71875,
|
| 492 |
+
"rewards/chosen": -0.10189628601074219,
|
| 493 |
+
"rewards/margins": 0.05620738863945007,
|
| 494 |
+
"rewards/rejected": -0.15810367465019226,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3223703703703704,
|
| 499 |
+
"grad_norm": 25.953433990478516,
|
| 500 |
+
"learning_rate": 3.408429731701635e-05,
|
| 501 |
+
"log_odds_chosen": 0.6586907505989075,
|
| 502 |
+
"log_odds_ratio": -0.5338261723518372,
|
| 503 |
+
"logits/chosen": -2.802279233932495,
|
| 504 |
+
"logits/rejected": -2.8112754821777344,
|
| 505 |
+
"logps/chosen": -0.6032203435897827,
|
| 506 |
+
"logps/rejected": -0.907819926738739,
|
| 507 |
+
"loss": 26.2333,
|
| 508 |
+
"nll_loss": 0.8216356039047241,
|
| 509 |
+
"rewards/accuracies": 0.731249988079071,
|
| 510 |
+
"rewards/chosen": -0.12064407765865326,
|
| 511 |
+
"rewards/margins": 0.06091991066932678,
|
| 512 |
+
"rewards/rejected": -0.18156400322914124,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.3697777777777778,
|
| 517 |
+
"grad_norm": 25.108062744140625,
|
| 518 |
+
"learning_rate": 3.2778194329621104e-05,
|
| 519 |
+
"log_odds_chosen": 0.5383874177932739,
|
| 520 |
+
"log_odds_ratio": -0.5537405610084534,
|
| 521 |
+
"logits/chosen": -2.90958833694458,
|
| 522 |
+
"logits/rejected": -2.7673959732055664,
|
| 523 |
+
"logps/chosen": -0.6065815687179565,
|
| 524 |
+
"logps/rejected": -0.8782438039779663,
|
| 525 |
+
"loss": 26.0014,
|
| 526 |
+
"nll_loss": 0.8171162605285645,
|
| 527 |
+
"rewards/accuracies": 0.668749988079071,
|
| 528 |
+
"rewards/chosen": -0.1213163286447525,
|
| 529 |
+
"rewards/margins": 0.054332435131073,
|
| 530 |
+
"rewards/rejected": -0.1756487637758255,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.417185185185185,
|
| 535 |
+
"grad_norm": 26.676942825317383,
|
| 536 |
+
"learning_rate": 3.144813424636031e-05,
|
| 537 |
+
"log_odds_chosen": 0.5983611345291138,
|
| 538 |
+
"log_odds_ratio": -0.547215461730957,
|
| 539 |
+
"logits/chosen": -2.847853422164917,
|
| 540 |
+
"logits/rejected": -2.8667774200439453,
|
| 541 |
+
"logps/chosen": -0.563664436340332,
|
| 542 |
+
"logps/rejected": -0.8132143020629883,
|
| 543 |
+
"loss": 25.1239,
|
| 544 |
+
"nll_loss": 0.7761403918266296,
|
| 545 |
+
"rewards/accuracies": 0.699999988079071,
|
| 546 |
+
"rewards/chosen": -0.11273287236690521,
|
| 547 |
+
"rewards/margins": 0.049909986555576324,
|
| 548 |
+
"rewards/rejected": -0.16264286637306213,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4645925925925927,
|
| 553 |
+
"grad_norm": 25.11577606201172,
|
| 554 |
+
"learning_rate": 3.0098213696293542e-05,
|
| 555 |
+
"log_odds_chosen": 0.674574077129364,
|
| 556 |
+
"log_odds_ratio": -0.5198480486869812,
|
| 557 |
+
"logits/chosen": -2.9048805236816406,
|
| 558 |
+
"logits/rejected": -2.7696218490600586,
|
| 559 |
+
"logps/chosen": -0.5536966919898987,
|
| 560 |
+
"logps/rejected": -0.8423013687133789,
|
| 561 |
+
"loss": 25.521,
|
| 562 |
+
"nll_loss": 0.8120133280754089,
|
| 563 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 564 |
+
"rewards/chosen": -0.11073935031890869,
|
| 565 |
+
"rewards/margins": 0.05772091820836067,
|
| 566 |
+
"rewards/rejected": -0.16846027970314026,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.512,
|
| 571 |
+
"grad_norm": 25.947433471679688,
|
| 572 |
+
"learning_rate": 2.8732590479375165e-05,
|
| 573 |
+
"log_odds_chosen": 0.5704504251480103,
|
| 574 |
+
"log_odds_ratio": -0.5273481607437134,
|
| 575 |
+
"logits/chosen": -2.8649063110351562,
|
| 576 |
+
"logits/rejected": -2.8651492595672607,
|
| 577 |
+
"logps/chosen": -0.6272570490837097,
|
| 578 |
+
"logps/rejected": -0.9066923260688782,
|
| 579 |
+
"loss": 26.2994,
|
| 580 |
+
"nll_loss": 0.8410785794258118,
|
| 581 |
+
"rewards/accuracies": 0.731249988079071,
|
| 582 |
+
"rewards/chosen": -0.12545141577720642,
|
| 583 |
+
"rewards/margins": 0.05588705092668533,
|
| 584 |
+
"rewards/rejected": -0.18133845925331116,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5594074074074074,
|
| 589 |
+
"grad_norm": 24.40997886657715,
|
| 590 |
+
"learning_rate": 2.7355470760292956e-05,
|
| 591 |
+
"log_odds_chosen": 0.8175321817398071,
|
| 592 |
+
"log_odds_ratio": -0.46456480026245117,
|
| 593 |
+
"logits/chosen": -2.8088977336883545,
|
| 594 |
+
"logits/rejected": -2.8634910583496094,
|
| 595 |
+
"logps/chosen": -0.5574014782905579,
|
| 596 |
+
"logps/rejected": -0.9002832174301147,
|
| 597 |
+
"loss": 25.9286,
|
| 598 |
+
"nll_loss": 0.7942954301834106,
|
| 599 |
+
"rewards/accuracies": 0.78125,
|
| 600 |
+
"rewards/chosen": -0.11148029565811157,
|
| 601 |
+
"rewards/margins": 0.06857635080814362,
|
| 602 |
+
"rewards/rejected": -0.1800566464662552,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6068148148148147,
|
| 607 |
+
"grad_norm": 26.33037757873535,
|
| 608 |
+
"learning_rate": 2.597109611334169e-05,
|
| 609 |
+
"log_odds_chosen": 0.7566910982131958,
|
| 610 |
+
"log_odds_ratio": -0.477250874042511,
|
| 611 |
+
"logits/chosen": -2.8108813762664795,
|
| 612 |
+
"logits/rejected": -2.784144878387451,
|
| 613 |
+
"logps/chosen": -0.5413329601287842,
|
| 614 |
+
"logps/rejected": -0.874762237071991,
|
| 615 |
+
"loss": 25.1232,
|
| 616 |
+
"nll_loss": 0.7829151153564453,
|
| 617 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 618 |
+
"rewards/chosen": -0.10826659202575684,
|
| 619 |
+
"rewards/margins": 0.06668586283922195,
|
| 620 |
+
"rewards/rejected": -0.1749524623155594,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6542222222222223,
|
| 625 |
+
"grad_norm": 25.30082130432129,
|
| 626 |
+
"learning_rate": 2.458373045823404e-05,
|
| 627 |
+
"log_odds_chosen": 0.6260709762573242,
|
| 628 |
+
"log_odds_ratio": -0.5326700210571289,
|
| 629 |
+
"logits/chosen": -2.8064045906066895,
|
| 630 |
+
"logits/rejected": -2.7094810009002686,
|
| 631 |
+
"logps/chosen": -0.5634613633155823,
|
| 632 |
+
"logps/rejected": -0.8632895350456238,
|
| 633 |
+
"loss": 25.126,
|
| 634 |
+
"nll_loss": 0.8014172315597534,
|
| 635 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 636 |
+
"rewards/chosen": -0.11269225925207138,
|
| 637 |
+
"rewards/margins": 0.05996562913060188,
|
| 638 |
+
"rewards/rejected": -0.17265789210796356,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7016296296296296,
|
| 643 |
+
"grad_norm": 25.368301391601562,
|
| 644 |
+
"learning_rate": 2.3197646927086697e-05,
|
| 645 |
+
"log_odds_chosen": 0.49957117438316345,
|
| 646 |
+
"log_odds_ratio": -0.5909265279769897,
|
| 647 |
+
"logits/chosen": -2.728269338607788,
|
| 648 |
+
"logits/rejected": -2.769505739212036,
|
| 649 |
+
"logps/chosen": -0.5491189360618591,
|
| 650 |
+
"logps/rejected": -0.7592207193374634,
|
| 651 |
+
"loss": 25.3579,
|
| 652 |
+
"nll_loss": 0.7878440618515015,
|
| 653 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 654 |
+
"rewards/chosen": -0.1098237857222557,
|
| 655 |
+
"rewards/margins": 0.04202036187052727,
|
| 656 |
+
"rewards/rejected": -0.15184415876865387,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.749037037037037,
|
| 661 |
+
"grad_norm": 24.203075408935547,
|
| 662 |
+
"learning_rate": 2.1817114703032176e-05,
|
| 663 |
+
"log_odds_chosen": 0.6133986115455627,
|
| 664 |
+
"log_odds_ratio": -0.5178900957107544,
|
| 665 |
+
"logits/chosen": -2.849193572998047,
|
| 666 |
+
"logits/rejected": -2.8242928981781006,
|
| 667 |
+
"logps/chosen": -0.5669192671775818,
|
| 668 |
+
"logps/rejected": -0.837879478931427,
|
| 669 |
+
"loss": 25.8406,
|
| 670 |
+
"nll_loss": 0.8041340112686157,
|
| 671 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 672 |
+
"rewards/chosen": -0.11338386684656143,
|
| 673 |
+
"rewards/margins": 0.054192036390304565,
|
| 674 |
+
"rewards/rejected": -0.1675758957862854,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.7964444444444445,
|
| 679 |
+
"grad_norm": 23.888410568237305,
|
| 680 |
+
"learning_rate": 2.0446385870993467e-05,
|
| 681 |
+
"log_odds_chosen": 0.7135507464408875,
|
| 682 |
+
"log_odds_ratio": -0.49830254912376404,
|
| 683 |
+
"logits/chosen": -2.871323585510254,
|
| 684 |
+
"logits/rejected": -2.716801404953003,
|
| 685 |
+
"logps/chosen": -0.5923060774803162,
|
| 686 |
+
"logps/rejected": -0.9126560091972351,
|
| 687 |
+
"loss": 25.5896,
|
| 688 |
+
"nll_loss": 0.7930334806442261,
|
| 689 |
+
"rewards/accuracies": 0.793749988079071,
|
| 690 |
+
"rewards/chosen": -0.1184612289071083,
|
| 691 |
+
"rewards/margins": 0.0640699714422226,
|
| 692 |
+
"rewards/rejected": -0.1825312077999115,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8438518518518519,
|
| 697 |
+
"grad_norm": 26.604984283447266,
|
| 698 |
+
"learning_rate": 1.9089682321121834e-05,
|
| 699 |
+
"log_odds_chosen": 0.7206512689590454,
|
| 700 |
+
"log_odds_ratio": -0.5070890784263611,
|
| 701 |
+
"logits/chosen": -2.8227410316467285,
|
| 702 |
+
"logits/rejected": -2.683023452758789,
|
| 703 |
+
"logps/chosen": -0.5759368538856506,
|
| 704 |
+
"logps/rejected": -0.8958457112312317,
|
| 705 |
+
"loss": 25.1795,
|
| 706 |
+
"nll_loss": 0.7880135774612427,
|
| 707 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 708 |
+
"rewards/chosen": -0.11518736928701401,
|
| 709 |
+
"rewards/margins": 0.06398177146911621,
|
| 710 |
+
"rewards/rejected": -0.17916914820671082,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8912592592592592,
|
| 715 |
+
"grad_norm": 24.46125602722168,
|
| 716 |
+
"learning_rate": 1.775118274523545e-05,
|
| 717 |
+
"log_odds_chosen": 0.6902771592140198,
|
| 718 |
+
"log_odds_ratio": -0.5180584788322449,
|
| 719 |
+
"logits/chosen": -2.8943402767181396,
|
| 720 |
+
"logits/rejected": -2.8360090255737305,
|
| 721 |
+
"logps/chosen": -0.574259877204895,
|
| 722 |
+
"logps/rejected": -0.8920720219612122,
|
| 723 |
+
"loss": 26.2124,
|
| 724 |
+
"nll_loss": 0.7853151559829712,
|
| 725 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 726 |
+
"rewards/chosen": -0.11485197395086288,
|
| 727 |
+
"rewards/margins": 0.06356243789196014,
|
| 728 |
+
"rewards/rejected": -0.17841440439224243,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9386666666666668,
|
| 733 |
+
"grad_norm": 22.397045135498047,
|
| 734 |
+
"learning_rate": 1.643500976631037e-05,
|
| 735 |
+
"log_odds_chosen": 0.5525709986686707,
|
| 736 |
+
"log_odds_ratio": -0.5616833567619324,
|
| 737 |
+
"logits/chosen": -2.8779659271240234,
|
| 738 |
+
"logits/rejected": -2.7355799674987793,
|
| 739 |
+
"logps/chosen": -0.5917361378669739,
|
| 740 |
+
"logps/rejected": -0.8651636242866516,
|
| 741 |
+
"loss": 25.1435,
|
| 742 |
+
"nll_loss": 0.7885797619819641,
|
| 743 |
+
"rewards/accuracies": 0.706250011920929,
|
| 744 |
+
"rewards/chosen": -0.11834722757339478,
|
| 745 |
+
"rewards/margins": 0.054685525596141815,
|
| 746 |
+
"rewards/rejected": -0.173032745718956,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.986074074074074,
|
| 751 |
+
"grad_norm": 23.24612808227539,
|
| 752 |
+
"learning_rate": 1.514521724066537e-05,
|
| 753 |
+
"log_odds_chosen": 0.6066923141479492,
|
| 754 |
+
"log_odds_ratio": -0.5472615361213684,
|
| 755 |
+
"logits/chosen": -2.8647236824035645,
|
| 756 |
+
"logits/rejected": -2.9078686237335205,
|
| 757 |
+
"logps/chosen": -0.5761645436286926,
|
| 758 |
+
"logps/rejected": -0.8358944058418274,
|
| 759 |
+
"loss": 23.8296,
|
| 760 |
+
"nll_loss": 0.7535713911056519,
|
| 761 |
+
"rewards/accuracies": 0.731249988079071,
|
| 762 |
+
"rewards/chosen": -0.1152329072356224,
|
| 763 |
+
"rewards/margins": 0.051945965737104416,
|
| 764 |
+
"rewards/rejected": -0.16717885434627533,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0284444444444443,
|
| 769 |
+
"grad_norm": 25.06052017211914,
|
| 770 |
+
"learning_rate": 1.3885777771950348e-05,
|
| 771 |
+
"log_odds_chosen": 1.0616474151611328,
|
| 772 |
+
"log_odds_ratio": -0.4191078841686249,
|
| 773 |
+
"logits/chosen": -2.7932209968566895,
|
| 774 |
+
"logits/rejected": -2.7296035289764404,
|
| 775 |
+
"logps/chosen": -0.471683144569397,
|
| 776 |
+
"logps/rejected": -0.8782158493995667,
|
| 777 |
+
"loss": 19.6408,
|
| 778 |
+
"nll_loss": 0.6726168990135193,
|
| 779 |
+
"rewards/accuracies": 0.811188817024231,
|
| 780 |
+
"rewards/chosen": -0.0943366289138794,
|
| 781 |
+
"rewards/margins": 0.08130654692649841,
|
| 782 |
+
"rewards/rejected": -0.1756431758403778,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.075851851851852,
|
| 787 |
+
"grad_norm": 31.775066375732422,
|
| 788 |
+
"learning_rate": 1.2660570475395683e-05,
|
| 789 |
+
"log_odds_chosen": 1.2414283752441406,
|
| 790 |
+
"log_odds_ratio": -0.37072741985321045,
|
| 791 |
+
"logits/chosen": -2.8022265434265137,
|
| 792 |
+
"logits/rejected": -2.7584967613220215,
|
| 793 |
+
"logps/chosen": -0.4369787573814392,
|
| 794 |
+
"logps/rejected": -0.9211339950561523,
|
| 795 |
+
"loss": 19.9908,
|
| 796 |
+
"nll_loss": 0.647934079170227,
|
| 797 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 798 |
+
"rewards/chosen": -0.08739574998617172,
|
| 799 |
+
"rewards/margins": 0.09683104604482651,
|
| 800 |
+
"rewards/rejected": -0.18422682583332062,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1232592592592594,
|
| 805 |
+
"grad_norm": 26.847965240478516,
|
| 806 |
+
"learning_rate": 1.1473369030008974e-05,
|
| 807 |
+
"log_odds_chosen": 1.1905401945114136,
|
| 808 |
+
"log_odds_ratio": -0.3879564702510834,
|
| 809 |
+
"logits/chosen": -2.7660067081451416,
|
| 810 |
+
"logits/rejected": -2.8042025566101074,
|
| 811 |
+
"logps/chosen": -0.4038045406341553,
|
| 812 |
+
"logps/rejected": -0.8510030508041382,
|
| 813 |
+
"loss": 20.6864,
|
| 814 |
+
"nll_loss": 0.6349369287490845,
|
| 815 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 816 |
+
"rewards/chosen": -0.0807609036564827,
|
| 817 |
+
"rewards/margins": 0.08943970501422882,
|
| 818 |
+
"rewards/rejected": -0.17020061612129211,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.1706666666666665,
|
| 823 |
+
"grad_norm": 26.880268096923828,
|
| 824 |
+
"learning_rate": 1.0327830055518842e-05,
|
| 825 |
+
"log_odds_chosen": 1.3126904964447021,
|
| 826 |
+
"log_odds_ratio": -0.3657984137535095,
|
| 827 |
+
"logits/chosen": -2.776597023010254,
|
| 828 |
+
"logits/rejected": -2.7055163383483887,
|
| 829 |
+
"logps/chosen": -0.41880035400390625,
|
| 830 |
+
"logps/rejected": -0.8985490798950195,
|
| 831 |
+
"loss": 19.3404,
|
| 832 |
+
"nll_loss": 0.5868525505065918,
|
| 833 |
+
"rewards/accuracies": 0.856249988079071,
|
| 834 |
+
"rewards/chosen": -0.08376006782054901,
|
| 835 |
+
"rewards/margins": 0.09594975411891937,
|
| 836 |
+
"rewards/rejected": -0.17970982193946838,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.218074074074074,
|
| 841 |
+
"grad_norm": 31.306821823120117,
|
| 842 |
+
"learning_rate": 9.227481849865235e-06,
|
| 843 |
+
"log_odds_chosen": 1.1681408882141113,
|
| 844 |
+
"log_odds_ratio": -0.4105006754398346,
|
| 845 |
+
"logits/chosen": -2.784541368484497,
|
| 846 |
+
"logits/rejected": -2.7821319103240967,
|
| 847 |
+
"logps/chosen": -0.44898858666419983,
|
| 848 |
+
"logps/rejected": -0.870564341545105,
|
| 849 |
+
"loss": 20.274,
|
| 850 |
+
"nll_loss": 0.6389156579971313,
|
| 851 |
+
"rewards/accuracies": 0.831250011920929,
|
| 852 |
+
"rewards/chosen": -0.08979771286249161,
|
| 853 |
+
"rewards/margins": 0.08431516587734222,
|
| 854 |
+
"rewards/rejected": -0.17411288619041443,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2654814814814817,
|
| 859 |
+
"grad_norm": 26.963720321655273,
|
| 860 |
+
"learning_rate": 8.175713521924978e-06,
|
| 861 |
+
"log_odds_chosen": 1.2784669399261475,
|
| 862 |
+
"log_odds_ratio": -0.3686579167842865,
|
| 863 |
+
"logits/chosen": -2.7864890098571777,
|
| 864 |
+
"logits/rejected": -2.8318161964416504,
|
| 865 |
+
"logps/chosen": -0.43692564964294434,
|
| 866 |
+
"logps/rejected": -0.9061982035636902,
|
| 867 |
+
"loss": 20.5384,
|
| 868 |
+
"nll_loss": 0.659681499004364,
|
| 869 |
+
"rewards/accuracies": 0.862500011920929,
|
| 870 |
+
"rewards/chosen": -0.0873851329088211,
|
| 871 |
+
"rewards/margins": 0.09385450184345245,
|
| 872 |
+
"rewards/rejected": -0.18123963475227356,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.3128888888888888,
|
| 877 |
+
"grad_norm": 27.23887062072754,
|
| 878 |
+
"learning_rate": 7.1757645529443665e-06,
|
| 879 |
+
"log_odds_chosen": 1.3585851192474365,
|
| 880 |
+
"log_odds_ratio": -0.3239150047302246,
|
| 881 |
+
"logits/chosen": -2.794093370437622,
|
| 882 |
+
"logits/rejected": -2.7042746543884277,
|
| 883 |
+
"logps/chosen": -0.41034674644470215,
|
| 884 |
+
"logps/rejected": -0.9184103012084961,
|
| 885 |
+
"loss": 19.5157,
|
| 886 |
+
"nll_loss": 0.6223703026771545,
|
| 887 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 888 |
+
"rewards/chosen": -0.08206935226917267,
|
| 889 |
+
"rewards/margins": 0.10161270946264267,
|
| 890 |
+
"rewards/rejected": -0.18368205428123474,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3602962962962963,
|
| 895 |
+
"grad_norm": 27.25604248046875,
|
| 896 |
+
"learning_rate": 6.230714818829733e-06,
|
| 897 |
+
"log_odds_chosen": 1.3011986017227173,
|
| 898 |
+
"log_odds_ratio": -0.3395061194896698,
|
| 899 |
+
"logits/chosen": -2.805485248565674,
|
| 900 |
+
"logits/rejected": -2.7763752937316895,
|
| 901 |
+
"logps/chosen": -0.44493547081947327,
|
| 902 |
+
"logps/rejected": -0.9599026441574097,
|
| 903 |
+
"loss": 19.1928,
|
| 904 |
+
"nll_loss": 0.6162556409835815,
|
| 905 |
+
"rewards/accuracies": 0.887499988079071,
|
| 906 |
+
"rewards/chosen": -0.08898709714412689,
|
| 907 |
+
"rewards/margins": 0.102993443608284,
|
| 908 |
+
"rewards/rejected": -0.1919805258512497,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.407703703703704,
|
| 913 |
+
"grad_norm": 27.4134578704834,
|
| 914 |
+
"learning_rate": 5.343475104027743e-06,
|
| 915 |
+
"log_odds_chosen": 1.4432458877563477,
|
| 916 |
+
"log_odds_ratio": -0.32654649019241333,
|
| 917 |
+
"logits/chosen": -2.748802661895752,
|
| 918 |
+
"logits/rejected": -2.7334983348846436,
|
| 919 |
+
"logps/chosen": -0.3842257261276245,
|
| 920 |
+
"logps/rejected": -0.9072483777999878,
|
| 921 |
+
"loss": 19.1784,
|
| 922 |
+
"nll_loss": 0.5993472337722778,
|
| 923 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 924 |
+
"rewards/chosen": -0.07684514671564102,
|
| 925 |
+
"rewards/margins": 0.1046045571565628,
|
| 926 |
+
"rewards/rejected": -0.18144968152046204,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.455111111111111,
|
| 931 |
+
"grad_norm": 28.401473999023438,
|
| 932 |
+
"learning_rate": 4.516778136213037e-06,
|
| 933 |
+
"log_odds_chosen": 1.488121747970581,
|
| 934 |
+
"log_odds_ratio": -0.3242756724357605,
|
| 935 |
+
"logits/chosen": -2.7776598930358887,
|
| 936 |
+
"logits/rejected": -2.6771626472473145,
|
| 937 |
+
"logps/chosen": -0.392120897769928,
|
| 938 |
+
"logps/rejected": -0.9110913276672363,
|
| 939 |
+
"loss": 19.3257,
|
| 940 |
+
"nll_loss": 0.5952539443969727,
|
| 941 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 942 |
+
"rewards/chosen": -0.07842419296503067,
|
| 943 |
+
"rewards/margins": 0.10379407554864883,
|
| 944 |
+
"rewards/rejected": -0.1822182834148407,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5025185185185186,
|
| 949 |
+
"grad_norm": 28.88060760498047,
|
| 950 |
+
"learning_rate": 3.7531701693965554e-06,
|
| 951 |
+
"log_odds_chosen": 1.3083126544952393,
|
| 952 |
+
"log_odds_ratio": -0.3742141127586365,
|
| 953 |
+
"logits/chosen": -2.752208709716797,
|
| 954 |
+
"logits/rejected": -2.7710514068603516,
|
| 955 |
+
"logps/chosen": -0.4226323068141937,
|
| 956 |
+
"logps/rejected": -0.9085744619369507,
|
| 957 |
+
"loss": 19.018,
|
| 958 |
+
"nll_loss": 0.5947185158729553,
|
| 959 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 960 |
+
"rewards/chosen": -0.08452647179365158,
|
| 961 |
+
"rewards/margins": 0.09718842804431915,
|
| 962 |
+
"rewards/rejected": -0.18171489238739014,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.549925925925926,
|
| 967 |
+
"grad_norm": 26.92013168334961,
|
| 968 |
+
"learning_rate": 3.055003141378948e-06,
|
| 969 |
+
"log_odds_chosen": 1.292988657951355,
|
| 970 |
+
"log_odds_ratio": -0.3492991328239441,
|
| 971 |
+
"logits/chosen": -2.8405444622039795,
|
| 972 |
+
"logits/rejected": -2.795259475708008,
|
| 973 |
+
"logps/chosen": -0.4322505593299866,
|
| 974 |
+
"logps/rejected": -0.9027539491653442,
|
| 975 |
+
"loss": 19.1197,
|
| 976 |
+
"nll_loss": 0.6248734593391418,
|
| 977 |
+
"rewards/accuracies": 0.875,
|
| 978 |
+
"rewards/chosen": -0.08645011484622955,
|
| 979 |
+
"rewards/margins": 0.09410068392753601,
|
| 980 |
+
"rewards/rejected": -0.18055078387260437,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.5973333333333333,
|
| 985 |
+
"grad_norm": 25.733835220336914,
|
| 986 |
+
"learning_rate": 2.424427429704365e-06,
|
| 987 |
+
"log_odds_chosen": 1.3847897052764893,
|
| 988 |
+
"log_odds_ratio": -0.33749812841415405,
|
| 989 |
+
"logits/chosen": -2.8116893768310547,
|
| 990 |
+
"logits/rejected": -2.695439100265503,
|
| 991 |
+
"logps/chosen": -0.4366413950920105,
|
| 992 |
+
"logps/rejected": -0.971570611000061,
|
| 993 |
+
"loss": 20.1018,
|
| 994 |
+
"nll_loss": 0.6324763894081116,
|
| 995 |
+
"rewards/accuracies": 0.875,
|
| 996 |
+
"rewards/chosen": -0.08732827007770538,
|
| 997 |
+
"rewards/margins": 0.10698586702346802,
|
| 998 |
+
"rewards/rejected": -0.1943141520023346,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.644740740740741,
|
| 1003 |
+
"grad_norm": 26.444095611572266,
|
| 1004 |
+
"learning_rate": 1.8633852284264508e-06,
|
| 1005 |
+
"log_odds_chosen": 1.2408794164657593,
|
| 1006 |
+
"log_odds_ratio": -0.3706284165382385,
|
| 1007 |
+
"logits/chosen": -2.783261775970459,
|
| 1008 |
+
"logits/rejected": -2.787781238555908,
|
| 1009 |
+
"logps/chosen": -0.4150269031524658,
|
| 1010 |
+
"logps/rejected": -0.8832289576530457,
|
| 1011 |
+
"loss": 19.2227,
|
| 1012 |
+
"nll_loss": 0.5912734270095825,
|
| 1013 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1014 |
+
"rewards/chosen": -0.0830053836107254,
|
| 1015 |
+
"rewards/margins": 0.09364042431116104,
|
| 1016 |
+
"rewards/rejected": -0.17664580047130585,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.6921481481481484,
|
| 1021 |
+
"grad_norm": 29.028366088867188,
|
| 1022 |
+
"learning_rate": 1.3736045660864034e-06,
|
| 1023 |
+
"log_odds_chosen": 1.2308132648468018,
|
| 1024 |
+
"log_odds_ratio": -0.3709443509578705,
|
| 1025 |
+
"logits/chosen": -2.829824924468994,
|
| 1026 |
+
"logits/rejected": -2.750471830368042,
|
| 1027 |
+
"logps/chosen": -0.4345945417881012,
|
| 1028 |
+
"logps/rejected": -0.8928596377372742,
|
| 1029 |
+
"loss": 19.7763,
|
| 1030 |
+
"nll_loss": 0.6206212639808655,
|
| 1031 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1032 |
+
"rewards/chosen": -0.086918905377388,
|
| 1033 |
+
"rewards/margins": 0.09165303409099579,
|
| 1034 |
+
"rewards/rejected": -0.17857195436954498,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.7395555555555555,
|
| 1039 |
+
"grad_norm": 27.85979652404785,
|
| 1040 |
+
"learning_rate": 9.565939833279192e-07,
|
| 1041 |
+
"log_odds_chosen": 1.3631649017333984,
|
| 1042 |
+
"log_odds_ratio": -0.34059271216392517,
|
| 1043 |
+
"logits/chosen": -2.741471529006958,
|
| 1044 |
+
"logits/rejected": -2.7575411796569824,
|
| 1045 |
+
"logps/chosen": -0.3912351727485657,
|
| 1046 |
+
"logps/rejected": -0.8874975442886353,
|
| 1047 |
+
"loss": 19.7779,
|
| 1048 |
+
"nll_loss": 0.6171984672546387,
|
| 1049 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1050 |
+
"rewards/chosen": -0.07824703305959702,
|
| 1051 |
+
"rewards/margins": 0.09925246983766556,
|
| 1052 |
+
"rewards/rejected": -0.17749951779842377,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.786962962962963,
|
| 1057 |
+
"grad_norm": 28.014738082885742,
|
| 1058 |
+
"learning_rate": 6.136378865420872e-07,
|
| 1059 |
+
"log_odds_chosen": 1.1799755096435547,
|
| 1060 |
+
"log_odds_ratio": -0.39853397011756897,
|
| 1061 |
+
"logits/chosen": -2.691950559616089,
|
| 1062 |
+
"logits/rejected": -2.7646536827087402,
|
| 1063 |
+
"logps/chosen": -0.45684152841567993,
|
| 1064 |
+
"logps/rejected": -0.9104090929031372,
|
| 1065 |
+
"loss": 19.6944,
|
| 1066 |
+
"nll_loss": 0.6349056959152222,
|
| 1067 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1068 |
+
"rewards/chosen": -0.09136830270290375,
|
| 1069 |
+
"rewards/margins": 0.0907135158777237,
|
| 1070 |
+
"rewards/rejected": -0.18208181858062744,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.83437037037037,
|
| 1075 |
+
"grad_norm": 29.324554443359375,
|
| 1076 |
+
"learning_rate": 3.45792591853214e-07,
|
| 1077 |
+
"log_odds_chosen": 1.1676690578460693,
|
| 1078 |
+
"log_odds_ratio": -0.3855035603046417,
|
| 1079 |
+
"logits/chosen": -2.804230213165283,
|
| 1080 |
+
"logits/rejected": -2.6715846061706543,
|
| 1081 |
+
"logps/chosen": -0.44394516944885254,
|
| 1082 |
+
"logps/rejected": -0.9196891784667969,
|
| 1083 |
+
"loss": 19.6876,
|
| 1084 |
+
"nll_loss": 0.6331617832183838,
|
| 1085 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1086 |
+
"rewards/chosen": -0.08878904581069946,
|
| 1087 |
+
"rewards/margins": 0.09514881670475006,
|
| 1088 |
+
"rewards/rejected": -0.18393786251544952,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.8817777777777778,
|
| 1093 |
+
"grad_norm": 26.675434112548828,
|
| 1094 |
+
"learning_rate": 1.538830716302092e-07,
|
| 1095 |
+
"log_odds_chosen": 1.2555936574935913,
|
| 1096 |
+
"log_odds_ratio": -0.3468943238258362,
|
| 1097 |
+
"logits/chosen": -2.866672992706299,
|
| 1098 |
+
"logits/rejected": -2.79132342338562,
|
| 1099 |
+
"logps/chosen": -0.4585191607475281,
|
| 1100 |
+
"logps/rejected": -0.9410701990127563,
|
| 1101 |
+
"loss": 19.4548,
|
| 1102 |
+
"nll_loss": 0.6043152809143066,
|
| 1103 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1104 |
+
"rewards/chosen": -0.09170383214950562,
|
| 1105 |
+
"rewards/margins": 0.09651019424200058,
|
| 1106 |
+
"rewards/rejected": -0.1882140338420868,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9291851851851853,
|
| 1111 |
+
"grad_norm": 30.77156639099121,
|
| 1112 |
+
"learning_rate": 3.8500413544415025e-08,
|
| 1113 |
+
"log_odds_chosen": 1.2845065593719482,
|
| 1114 |
+
"log_odds_ratio": -0.3690781891345978,
|
| 1115 |
+
"logits/chosen": -2.767889976501465,
|
| 1116 |
+
"logits/rejected": -2.665135383605957,
|
| 1117 |
+
"logps/chosen": -0.41893452405929565,
|
| 1118 |
+
"logps/rejected": -0.8851953744888306,
|
| 1119 |
+
"loss": 19.3972,
|
| 1120 |
+
"nll_loss": 0.6142546534538269,
|
| 1121 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1122 |
+
"rewards/chosen": -0.08378690481185913,
|
| 1123 |
+
"rewards/margins": 0.09325218200683594,
|
| 1124 |
+
"rewards/rejected": -0.17703908681869507,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.9765925925925925,
|
| 1129 |
+
"grad_norm": 27.507280349731445,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 1.394679069519043,
|
| 1132 |
+
"log_odds_ratio": -0.32423505187034607,
|
| 1133 |
+
"logits/chosen": -2.744528293609619,
|
| 1134 |
+
"logits/rejected": -2.668449640274048,
|
| 1135 |
+
"logps/chosen": -0.4062735140323639,
|
| 1136 |
+
"logps/rejected": -0.9089689254760742,
|
| 1137 |
+
"loss": 18.6834,
|
| 1138 |
+
"nll_loss": 0.5965942144393921,
|
| 1139 |
+
"rewards/accuracies": 0.90625,
|
| 1140 |
+
"rewards/chosen": -0.08125469833612442,
|
| 1141 |
+
"rewards/margins": 0.10053907334804535,
|
| 1142 |
+
"rewards/rejected": -0.18179377913475037,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.9765925925925925,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 26.480164833674355,
|
| 1150 |
+
"train_runtime": 6938.2073,
|
| 1151 |
+
"train_samples_per_second": 2.919,
|
| 1152 |
+
"train_steps_per_second": 0.045
|
| 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 |
+
}
|