Model save
Browse files- README.md +67 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1176 -0
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-7b
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: gemma-7b-orpo-noisy-1e-5-norm
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-orpo-noisy-1e-5-norm
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="silviasapora/gemma-7b-orpo-noisy-1e-5-norm", 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/ytq75ba2)
|
| 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": 95.27591378348214,
|
| 5 |
+
"train_runtime": 8112.4006,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.496,
|
| 8 |
+
"train_steps_per_second": 0.039
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.9765925925925925,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 95.27591378348214,
|
| 5 |
+
"train_runtime": 8112.4006,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.496,
|
| 8 |
+
"train_steps_per_second": 0.039
|
| 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": 2580.320556640625,
|
| 14 |
+
"learning_rate": 1.5625e-06,
|
| 15 |
+
"log_odds_chosen": 5.033224582672119,
|
| 16 |
+
"log_odds_ratio": -9.833906173706055,
|
| 17 |
+
"logits/chosen": 138.36488342285156,
|
| 18 |
+
"logits/rejected": 152.9020233154297,
|
| 19 |
+
"logps/chosen": -20.877660751342773,
|
| 20 |
+
"logps/rejected": -25.9111385345459,
|
| 21 |
+
"loss": 459.1401,
|
| 22 |
+
"nll_loss": 8.45106315612793,
|
| 23 |
+
"rewards/accuracies": 0.53125,
|
| 24 |
+
"rewards/chosen": -10.438830375671387,
|
| 25 |
+
"rewards/margins": 2.5167386531829834,
|
| 26 |
+
"rewards/rejected": -12.95556926727295,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 956.760009765625,
|
| 32 |
+
"learning_rate": 3.125e-06,
|
| 33 |
+
"log_odds_chosen": 2.400622606277466,
|
| 34 |
+
"log_odds_ratio": -7.376997470855713,
|
| 35 |
+
"logits/chosen": 131.07968139648438,
|
| 36 |
+
"logits/rejected": 155.00018310546875,
|
| 37 |
+
"logps/chosen": -16.098163604736328,
|
| 38 |
+
"logps/rejected": -18.498001098632812,
|
| 39 |
+
"loss": 398.9935,
|
| 40 |
+
"nll_loss": 7.736745357513428,
|
| 41 |
+
"rewards/accuracies": 0.543749988079071,
|
| 42 |
+
"rewards/chosen": -8.049081802368164,
|
| 43 |
+
"rewards/margins": 1.1999186277389526,
|
| 44 |
+
"rewards/rejected": -9.249000549316406,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 571.1576538085938,
|
| 50 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 51 |
+
"log_odds_chosen": 4.6121039390563965,
|
| 52 |
+
"log_odds_ratio": -8.182807922363281,
|
| 53 |
+
"logits/chosen": 105.5162124633789,
|
| 54 |
+
"logits/rejected": 132.4467315673828,
|
| 55 |
+
"logps/chosen": -21.264659881591797,
|
| 56 |
+
"logps/rejected": -25.8754940032959,
|
| 57 |
+
"loss": 389.0222,
|
| 58 |
+
"nll_loss": 9.0397367477417,
|
| 59 |
+
"rewards/accuracies": 0.581250011920929,
|
| 60 |
+
"rewards/chosen": -10.632329940795898,
|
| 61 |
+
"rewards/margins": 2.30541729927063,
|
| 62 |
+
"rewards/rejected": -12.93774700164795,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 905.5711669921875,
|
| 68 |
+
"learning_rate": 6.25e-06,
|
| 69 |
+
"log_odds_chosen": -0.20019039511680603,
|
| 70 |
+
"log_odds_ratio": -7.8728742599487305,
|
| 71 |
+
"logits/chosen": 133.976318359375,
|
| 72 |
+
"logits/rejected": 139.43692016601562,
|
| 73 |
+
"logps/chosen": -18.147478103637695,
|
| 74 |
+
"logps/rejected": -17.94868278503418,
|
| 75 |
+
"loss": 419.4126,
|
| 76 |
+
"nll_loss": 8.399177551269531,
|
| 77 |
+
"rewards/accuracies": 0.53125,
|
| 78 |
+
"rewards/chosen": -9.073739051818848,
|
| 79 |
+
"rewards/margins": -0.09939660876989365,
|
| 80 |
+
"rewards/rejected": -8.97434139251709,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 316.11334228515625,
|
| 86 |
+
"learning_rate": 7.8125e-06,
|
| 87 |
+
"log_odds_chosen": 4.466104030609131,
|
| 88 |
+
"log_odds_ratio": -7.779648780822754,
|
| 89 |
+
"logits/chosen": 137.90159606933594,
|
| 90 |
+
"logits/rejected": 146.57969665527344,
|
| 91 |
+
"logps/chosen": -18.408245086669922,
|
| 92 |
+
"logps/rejected": -22.874719619750977,
|
| 93 |
+
"loss": 347.403,
|
| 94 |
+
"nll_loss": 7.184973239898682,
|
| 95 |
+
"rewards/accuracies": 0.518750011920929,
|
| 96 |
+
"rewards/chosen": -9.204122543334961,
|
| 97 |
+
"rewards/margins": 2.2332377433776855,
|
| 98 |
+
"rewards/rejected": -11.437359809875488,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 597.9056396484375,
|
| 104 |
+
"learning_rate": 9.375000000000001e-06,
|
| 105 |
+
"log_odds_chosen": 1.4436252117156982,
|
| 106 |
+
"log_odds_ratio": -7.299798011779785,
|
| 107 |
+
"logits/chosen": 127.5577621459961,
|
| 108 |
+
"logits/rejected": 165.3824005126953,
|
| 109 |
+
"logps/chosen": -18.25235366821289,
|
| 110 |
+
"logps/rejected": -19.692398071289062,
|
| 111 |
+
"loss": 367.166,
|
| 112 |
+
"nll_loss": 7.642889976501465,
|
| 113 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 114 |
+
"rewards/chosen": -9.126176834106445,
|
| 115 |
+
"rewards/margins": 0.7200227379798889,
|
| 116 |
+
"rewards/rejected": -9.846199035644531,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 1674.3773193359375,
|
| 122 |
+
"learning_rate": 9.997227514697568e-06,
|
| 123 |
+
"log_odds_chosen": -2.2156496047973633,
|
| 124 |
+
"log_odds_ratio": -7.339943885803223,
|
| 125 |
+
"logits/chosen": 163.3466033935547,
|
| 126 |
+
"logits/rejected": 147.68124389648438,
|
| 127 |
+
"logps/chosen": -15.995862007141113,
|
| 128 |
+
"logps/rejected": -13.778966903686523,
|
| 129 |
+
"loss": 308.5459,
|
| 130 |
+
"nll_loss": 6.2163190841674805,
|
| 131 |
+
"rewards/accuracies": 0.5,
|
| 132 |
+
"rewards/chosen": -7.997931003570557,
|
| 133 |
+
"rewards/margins": -1.1084463596343994,
|
| 134 |
+
"rewards/rejected": -6.889483451843262,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 1445.621337890625,
|
| 140 |
+
"learning_rate": 9.980295682286924e-06,
|
| 141 |
+
"log_odds_chosen": 0.39161425828933716,
|
| 142 |
+
"log_odds_ratio": -2.787511110305786,
|
| 143 |
+
"logits/chosen": 178.052978515625,
|
| 144 |
+
"logits/rejected": 191.52259826660156,
|
| 145 |
+
"logps/chosen": -8.379082679748535,
|
| 146 |
+
"logps/rejected": -8.763626098632812,
|
| 147 |
+
"loss": 205.9666,
|
| 148 |
+
"nll_loss": 4.583432197570801,
|
| 149 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 150 |
+
"rewards/chosen": -4.189541339874268,
|
| 151 |
+
"rewards/margins": 0.19227202236652374,
|
| 152 |
+
"rewards/rejected": -4.381813049316406,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 437.58502197265625,
|
| 158 |
+
"learning_rate": 9.94802437440896e-06,
|
| 159 |
+
"log_odds_chosen": 0.5884231925010681,
|
| 160 |
+
"log_odds_ratio": -1.4498884677886963,
|
| 161 |
+
"logits/chosen": 213.7032928466797,
|
| 162 |
+
"logits/rejected": 203.2352752685547,
|
| 163 |
+
"logps/chosen": -3.589550733566284,
|
| 164 |
+
"logps/rejected": -4.178198337554932,
|
| 165 |
+
"loss": 102.45,
|
| 166 |
+
"nll_loss": 2.611146926879883,
|
| 167 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 168 |
+
"rewards/chosen": -1.794775366783142,
|
| 169 |
+
"rewards/margins": 0.29432395100593567,
|
| 170 |
+
"rewards/rejected": -2.089099168777466,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 180.1491241455078,
|
| 176 |
+
"learning_rate": 9.90051298775959e-06,
|
| 177 |
+
"log_odds_chosen": 0.2308872491121292,
|
| 178 |
+
"log_odds_ratio": -0.8247494697570801,
|
| 179 |
+
"logits/chosen": 221.44058227539062,
|
| 180 |
+
"logits/rejected": 221.6063995361328,
|
| 181 |
+
"logps/chosen": -1.8605806827545166,
|
| 182 |
+
"logps/rejected": -2.0761098861694336,
|
| 183 |
+
"loss": 83.7704,
|
| 184 |
+
"nll_loss": 2.2189698219299316,
|
| 185 |
+
"rewards/accuracies": 0.53125,
|
| 186 |
+
"rewards/chosen": -0.9302903413772583,
|
| 187 |
+
"rewards/margins": 0.10776450484991074,
|
| 188 |
+
"rewards/rejected": -1.0380549430847168,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 131.46673583984375,
|
| 194 |
+
"learning_rate": 9.837907858981536e-06,
|
| 195 |
+
"log_odds_chosen": 0.12095841020345688,
|
| 196 |
+
"log_odds_ratio": -0.8109160661697388,
|
| 197 |
+
"logits/chosen": 231.9627685546875,
|
| 198 |
+
"logits/rejected": 238.06460571289062,
|
| 199 |
+
"logps/chosen": -1.6934187412261963,
|
| 200 |
+
"logps/rejected": -1.7964763641357422,
|
| 201 |
+
"loss": 74.7448,
|
| 202 |
+
"nll_loss": 2.007000207901001,
|
| 203 |
+
"rewards/accuracies": 0.5625,
|
| 204 |
+
"rewards/chosen": -0.8467093706130981,
|
| 205 |
+
"rewards/margins": 0.051528822630643845,
|
| 206 |
+
"rewards/rejected": -0.8982381820678711,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 149.99351501464844,
|
| 212 |
+
"learning_rate": 9.760401813942641e-06,
|
| 213 |
+
"log_odds_chosen": 0.2649911046028137,
|
| 214 |
+
"log_odds_ratio": -0.7256797552108765,
|
| 215 |
+
"logits/chosen": 218.29061889648438,
|
| 216 |
+
"logits/rejected": 230.5416259765625,
|
| 217 |
+
"logps/chosen": -1.5244457721710205,
|
| 218 |
+
"logps/rejected": -1.7421019077301025,
|
| 219 |
+
"loss": 70.0635,
|
| 220 |
+
"nll_loss": 1.8057339191436768,
|
| 221 |
+
"rewards/accuracies": 0.5625,
|
| 222 |
+
"rewards/chosen": -0.7622228860855103,
|
| 223 |
+
"rewards/margins": 0.10882814228534698,
|
| 224 |
+
"rewards/rejected": -0.8710509538650513,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 172.64825439453125,
|
| 230 |
+
"learning_rate": 9.668233573825794e-06,
|
| 231 |
+
"log_odds_chosen": 0.12981751561164856,
|
| 232 |
+
"log_odds_ratio": -0.8062798380851746,
|
| 233 |
+
"logits/chosen": 232.5802764892578,
|
| 234 |
+
"logits/rejected": 233.1844482421875,
|
| 235 |
+
"logps/chosen": -1.5933538675308228,
|
| 236 |
+
"logps/rejected": -1.6922533512115479,
|
| 237 |
+
"loss": 68.5854,
|
| 238 |
+
"nll_loss": 1.815799355506897,
|
| 239 |
+
"rewards/accuracies": 0.53125,
|
| 240 |
+
"rewards/chosen": -0.7966769337654114,
|
| 241 |
+
"rewards/margins": 0.049449823796749115,
|
| 242 |
+
"rewards/rejected": -0.8461266756057739,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 136.158447265625,
|
| 248 |
+
"learning_rate": 9.56168701985981e-06,
|
| 249 |
+
"log_odds_chosen": 0.42648229002952576,
|
| 250 |
+
"log_odds_ratio": -0.6542950868606567,
|
| 251 |
+
"logits/chosen": 224.38955688476562,
|
| 252 |
+
"logits/rejected": 242.11837768554688,
|
| 253 |
+
"logps/chosen": -1.4570573568344116,
|
| 254 |
+
"logps/rejected": -1.8463979959487915,
|
| 255 |
+
"loss": 67.3288,
|
| 256 |
+
"nll_loss": 1.753805160522461,
|
| 257 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 258 |
+
"rewards/chosen": -0.7285286784172058,
|
| 259 |
+
"rewards/margins": 0.19467028975486755,
|
| 260 |
+
"rewards/rejected": -0.9231989979743958,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 115.01741790771484,
|
| 266 |
+
"learning_rate": 9.441090318955843e-06,
|
| 267 |
+
"log_odds_chosen": 0.24410974979400635,
|
| 268 |
+
"log_odds_ratio": -0.7884246706962585,
|
| 269 |
+
"logits/chosen": 249.4243621826172,
|
| 270 |
+
"logits/rejected": 242.49984741210938,
|
| 271 |
+
"logps/chosen": -1.5666649341583252,
|
| 272 |
+
"logps/rejected": -1.7553819417953491,
|
| 273 |
+
"loss": 66.3816,
|
| 274 |
+
"nll_loss": 1.7346446514129639,
|
| 275 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 276 |
+
"rewards/chosen": -0.7833324670791626,
|
| 277 |
+
"rewards/margins": 0.09435858577489853,
|
| 278 |
+
"rewards/rejected": -0.8776909708976746,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 117.21932983398438,
|
| 284 |
+
"learning_rate": 9.306814912942445e-06,
|
| 285 |
+
"log_odds_chosen": 0.33950307965278625,
|
| 286 |
+
"log_odds_ratio": -0.7339600920677185,
|
| 287 |
+
"logits/chosen": 235.14981079101562,
|
| 288 |
+
"logits/rejected": 246.0205078125,
|
| 289 |
+
"logps/chosen": -1.488161325454712,
|
| 290 |
+
"logps/rejected": -1.7720199823379517,
|
| 291 |
+
"loss": 64.8768,
|
| 292 |
+
"nll_loss": 1.7132374048233032,
|
| 293 |
+
"rewards/accuracies": 0.606249988079071,
|
| 294 |
+
"rewards/chosen": -0.744080662727356,
|
| 295 |
+
"rewards/margins": 0.1419292837381363,
|
| 296 |
+
"rewards/rejected": -0.8860099911689758,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 94.97506713867188,
|
| 302 |
+
"learning_rate": 9.159274374512444e-06,
|
| 303 |
+
"log_odds_chosen": 0.3309640884399414,
|
| 304 |
+
"log_odds_ratio": -0.7209373116493225,
|
| 305 |
+
"logits/chosen": 235.80746459960938,
|
| 306 |
+
"logits/rejected": 265.1528015136719,
|
| 307 |
+
"logps/chosen": -1.376119613647461,
|
| 308 |
+
"logps/rejected": -1.6490930318832397,
|
| 309 |
+
"loss": 63.5063,
|
| 310 |
+
"nll_loss": 1.6010291576385498,
|
| 311 |
+
"rewards/accuracies": 0.643750011920929,
|
| 312 |
+
"rewards/chosen": -0.6880598068237305,
|
| 313 |
+
"rewards/margins": 0.13648679852485657,
|
| 314 |
+
"rewards/rejected": -0.8245465159416199,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 124.18742370605469,
|
| 320 |
+
"learning_rate": 8.99892313340537e-06,
|
| 321 |
+
"log_odds_chosen": 0.28358036279678345,
|
| 322 |
+
"log_odds_ratio": -0.7167706489562988,
|
| 323 |
+
"logits/chosen": 244.89981079101562,
|
| 324 |
+
"logits/rejected": 260.676025390625,
|
| 325 |
+
"logps/chosen": -1.414159893989563,
|
| 326 |
+
"logps/rejected": -1.6283620595932007,
|
| 327 |
+
"loss": 65.2253,
|
| 328 |
+
"nll_loss": 1.647252082824707,
|
| 329 |
+
"rewards/accuracies": 0.612500011920929,
|
| 330 |
+
"rewards/chosen": -0.7070799469947815,
|
| 331 |
+
"rewards/margins": 0.10710110515356064,
|
| 332 |
+
"rewards/rejected": -0.8141810297966003,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 100.92960357666016,
|
| 338 |
+
"learning_rate": 8.826255076748823e-06,
|
| 339 |
+
"log_odds_chosen": 0.3998146057128906,
|
| 340 |
+
"log_odds_ratio": -0.6268864870071411,
|
| 341 |
+
"logits/chosen": 235.48410034179688,
|
| 342 |
+
"logits/rejected": 251.33712768554688,
|
| 343 |
+
"logps/chosen": -1.2698018550872803,
|
| 344 |
+
"logps/rejected": -1.5803816318511963,
|
| 345 |
+
"loss": 63.0803,
|
| 346 |
+
"nll_loss": 1.5454211235046387,
|
| 347 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 348 |
+
"rewards/chosen": -0.6349009275436401,
|
| 349 |
+
"rewards/margins": 0.15528982877731323,
|
| 350 |
+
"rewards/rejected": -0.7901908159255981,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 141.9030303955078,
|
| 356 |
+
"learning_rate": 8.641802027869774e-06,
|
| 357 |
+
"log_odds_chosen": 0.21526363492012024,
|
| 358 |
+
"log_odds_ratio": -0.7336837649345398,
|
| 359 |
+
"logits/chosen": 244.89962768554688,
|
| 360 |
+
"logits/rejected": 258.1126403808594,
|
| 361 |
+
"logps/chosen": -1.3898332118988037,
|
| 362 |
+
"logps/rejected": -1.5845530033111572,
|
| 363 |
+
"loss": 63.6891,
|
| 364 |
+
"nll_loss": 1.6614280939102173,
|
| 365 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 366 |
+
"rewards/chosen": -0.6949166059494019,
|
| 367 |
+
"rewards/margins": 0.09735982120037079,
|
| 368 |
+
"rewards/rejected": -0.7922765016555786,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 100.77186584472656,
|
| 374 |
+
"learning_rate": 8.446132108261136e-06,
|
| 375 |
+
"log_odds_chosen": 0.35006898641586304,
|
| 376 |
+
"log_odds_ratio": -0.6226455569267273,
|
| 377 |
+
"logits/chosen": 238.6287841796875,
|
| 378 |
+
"logits/rejected": 256.33172607421875,
|
| 379 |
+
"logps/chosen": -1.2620939016342163,
|
| 380 |
+
"logps/rejected": -1.5345309972763062,
|
| 381 |
+
"loss": 59.8393,
|
| 382 |
+
"nll_loss": 1.5451446771621704,
|
| 383 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 384 |
+
"rewards/chosen": -0.6310469508171082,
|
| 385 |
+
"rewards/margins": 0.13621851801872253,
|
| 386 |
+
"rewards/rejected": -0.7672654986381531,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.037925925925926,
|
| 391 |
+
"grad_norm": 96.23641967773438,
|
| 392 |
+
"learning_rate": 8.23984798774876e-06,
|
| 393 |
+
"log_odds_chosen": 0.34819138050079346,
|
| 394 |
+
"log_odds_ratio": -0.654488205909729,
|
| 395 |
+
"logits/chosen": 246.05015563964844,
|
| 396 |
+
"logits/rejected": 255.47967529296875,
|
| 397 |
+
"logps/chosen": -1.2216328382492065,
|
| 398 |
+
"logps/rejected": -1.5190995931625366,
|
| 399 |
+
"loss": 52.3665,
|
| 400 |
+
"nll_loss": 1.4913266897201538,
|
| 401 |
+
"rewards/accuracies": 0.5944055914878845,
|
| 402 |
+
"rewards/chosen": -0.6108164191246033,
|
| 403 |
+
"rewards/margins": 0.148733451962471,
|
| 404 |
+
"rewards/rejected": -0.7595497965812683,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0853333333333333,
|
| 409 |
+
"grad_norm": 102.46453094482422,
|
| 410 |
+
"learning_rate": 8.023585028248435e-06,
|
| 411 |
+
"log_odds_chosen": 0.5755366086959839,
|
| 412 |
+
"log_odds_ratio": -0.5598169565200806,
|
| 413 |
+
"logits/chosen": 231.81973266601562,
|
| 414 |
+
"logits/rejected": 254.8740234375,
|
| 415 |
+
"logps/chosen": -1.1871004104614258,
|
| 416 |
+
"logps/rejected": -1.6207106113433838,
|
| 417 |
+
"loss": 56.5061,
|
| 418 |
+
"nll_loss": 1.442861795425415,
|
| 419 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 420 |
+
"rewards/chosen": -0.5935502052307129,
|
| 421 |
+
"rewards/margins": 0.21680507063865662,
|
| 422 |
+
"rewards/rejected": -0.8103553056716919,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1327407407407408,
|
| 427 |
+
"grad_norm": 95.08788299560547,
|
| 428 |
+
"learning_rate": 7.798009326830167e-06,
|
| 429 |
+
"log_odds_chosen": 0.5298383235931396,
|
| 430 |
+
"log_odds_ratio": -0.5681466460227966,
|
| 431 |
+
"logits/chosen": 227.4845428466797,
|
| 432 |
+
"logits/rejected": 237.4995880126953,
|
| 433 |
+
"logps/chosen": -1.187435507774353,
|
| 434 |
+
"logps/rejected": -1.5832734107971191,
|
| 435 |
+
"loss": 58.2104,
|
| 436 |
+
"nll_loss": 1.5183286666870117,
|
| 437 |
+
"rewards/accuracies": 0.706250011920929,
|
| 438 |
+
"rewards/chosen": -0.5937177538871765,
|
| 439 |
+
"rewards/margins": 0.19791889190673828,
|
| 440 |
+
"rewards/rejected": -0.7916367053985596,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1801481481481482,
|
| 445 |
+
"grad_norm": 131.9794158935547,
|
| 446 |
+
"learning_rate": 7.563815664117173e-06,
|
| 447 |
+
"log_odds_chosen": 0.4175822138786316,
|
| 448 |
+
"log_odds_ratio": -0.6133934259414673,
|
| 449 |
+
"logits/chosen": 232.56655883789062,
|
| 450 |
+
"logits/rejected": 236.88711547851562,
|
| 451 |
+
"logps/chosen": -1.2231204509735107,
|
| 452 |
+
"logps/rejected": -1.5356894731521606,
|
| 453 |
+
"loss": 55.8452,
|
| 454 |
+
"nll_loss": 1.4756473302841187,
|
| 455 |
+
"rewards/accuracies": 0.6875,
|
| 456 |
+
"rewards/chosen": -0.6115602254867554,
|
| 457 |
+
"rewards/margins": 0.1562844067811966,
|
| 458 |
+
"rewards/rejected": -0.7678447365760803,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2275555555555555,
|
| 463 |
+
"grad_norm": 128.4602813720703,
|
| 464 |
+
"learning_rate": 7.321725364338566e-06,
|
| 465 |
+
"log_odds_chosen": 0.3267204165458679,
|
| 466 |
+
"log_odds_ratio": -0.6442457437515259,
|
| 467 |
+
"logits/chosen": 226.52706909179688,
|
| 468 |
+
"logits/rejected": 236.42977905273438,
|
| 469 |
+
"logps/chosen": -1.1632269620895386,
|
| 470 |
+
"logps/rejected": -1.3935999870300293,
|
| 471 |
+
"loss": 55.0227,
|
| 472 |
+
"nll_loss": 1.4348909854888916,
|
| 473 |
+
"rewards/accuracies": 0.637499988079071,
|
| 474 |
+
"rewards/chosen": -0.5816134810447693,
|
| 475 |
+
"rewards/margins": 0.11518646776676178,
|
| 476 |
+
"rewards/rejected": -0.6967999935150146,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.274962962962963,
|
| 481 |
+
"grad_norm": 111.24720764160156,
|
| 482 |
+
"learning_rate": 7.072484073626872e-06,
|
| 483 |
+
"log_odds_chosen": 0.39476487040519714,
|
| 484 |
+
"log_odds_ratio": -0.6248899698257446,
|
| 485 |
+
"logits/chosen": 231.5218505859375,
|
| 486 |
+
"logits/rejected": 237.8566131591797,
|
| 487 |
+
"logps/chosen": -1.1376434564590454,
|
| 488 |
+
"logps/rejected": -1.4273903369903564,
|
| 489 |
+
"loss": 55.291,
|
| 490 |
+
"nll_loss": 1.433115839958191,
|
| 491 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 492 |
+
"rewards/chosen": -0.5688217282295227,
|
| 493 |
+
"rewards/margins": 0.14487341046333313,
|
| 494 |
+
"rewards/rejected": -0.7136951684951782,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3223703703703704,
|
| 499 |
+
"grad_norm": 123.10855102539062,
|
| 500 |
+
"learning_rate": 6.816859463403271e-06,
|
| 501 |
+
"log_odds_chosen": 0.4121381640434265,
|
| 502 |
+
"log_odds_ratio": -0.6339040994644165,
|
| 503 |
+
"logits/chosen": 231.813720703125,
|
| 504 |
+
"logits/rejected": 230.7804412841797,
|
| 505 |
+
"logps/chosen": -1.231404423713684,
|
| 506 |
+
"logps/rejected": -1.5583487749099731,
|
| 507 |
+
"loss": 57.6659,
|
| 508 |
+
"nll_loss": 1.4830906391143799,
|
| 509 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 510 |
+
"rewards/chosen": -0.615702211856842,
|
| 511 |
+
"rewards/margins": 0.16347214579582214,
|
| 512 |
+
"rewards/rejected": -0.7791743874549866,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.3697777777777778,
|
| 517 |
+
"grad_norm": 130.8789825439453,
|
| 518 |
+
"learning_rate": 6.555638865924221e-06,
|
| 519 |
+
"log_odds_chosen": 0.5062509775161743,
|
| 520 |
+
"log_odds_ratio": -0.5808830857276917,
|
| 521 |
+
"logits/chosen": 227.046875,
|
| 522 |
+
"logits/rejected": 243.47970581054688,
|
| 523 |
+
"logps/chosen": -1.1645643711090088,
|
| 524 |
+
"logps/rejected": -1.5737435817718506,
|
| 525 |
+
"loss": 55.8075,
|
| 526 |
+
"nll_loss": 1.456322193145752,
|
| 527 |
+
"rewards/accuracies": 0.668749988079071,
|
| 528 |
+
"rewards/chosen": -0.5822821855545044,
|
| 529 |
+
"rewards/margins": 0.20458965003490448,
|
| 530 |
+
"rewards/rejected": -0.7868717908859253,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.417185185185185,
|
| 535 |
+
"grad_norm": 204.43161010742188,
|
| 536 |
+
"learning_rate": 6.289626849272062e-06,
|
| 537 |
+
"log_odds_chosen": 0.27410560846328735,
|
| 538 |
+
"log_odds_ratio": -0.6676009893417358,
|
| 539 |
+
"logits/chosen": 231.1629180908203,
|
| 540 |
+
"logits/rejected": 226.8252716064453,
|
| 541 |
+
"logps/chosen": -1.1737444400787354,
|
| 542 |
+
"logps/rejected": -1.372044324874878,
|
| 543 |
+
"loss": 55.1833,
|
| 544 |
+
"nll_loss": 1.4221489429473877,
|
| 545 |
+
"rewards/accuracies": 0.606249988079071,
|
| 546 |
+
"rewards/chosen": -0.5868722200393677,
|
| 547 |
+
"rewards/margins": 0.09915001690387726,
|
| 548 |
+
"rewards/rejected": -0.686022162437439,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4645925925925927,
|
| 553 |
+
"grad_norm": 107.84815979003906,
|
| 554 |
+
"learning_rate": 6.0196427392587085e-06,
|
| 555 |
+
"log_odds_chosen": 0.4244801104068756,
|
| 556 |
+
"log_odds_ratio": -0.5939399600028992,
|
| 557 |
+
"logits/chosen": 218.2533721923828,
|
| 558 |
+
"logits/rejected": 238.97201538085938,
|
| 559 |
+
"logps/chosen": -1.1000165939331055,
|
| 560 |
+
"logps/rejected": -1.4028135538101196,
|
| 561 |
+
"loss": 55.1904,
|
| 562 |
+
"nll_loss": 1.4290273189544678,
|
| 563 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 564 |
+
"rewards/chosen": -0.5500082969665527,
|
| 565 |
+
"rewards/margins": 0.15139850974082947,
|
| 566 |
+
"rewards/rejected": -0.7014067769050598,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.512,
|
| 571 |
+
"grad_norm": 124.19522094726562,
|
| 572 |
+
"learning_rate": 5.746518095875033e-06,
|
| 573 |
+
"log_odds_chosen": 0.42281055450439453,
|
| 574 |
+
"log_odds_ratio": -0.5916022062301636,
|
| 575 |
+
"logits/chosen": 226.8450469970703,
|
| 576 |
+
"logits/rejected": 230.33285522460938,
|
| 577 |
+
"logps/chosen": -1.174420952796936,
|
| 578 |
+
"logps/rejected": -1.5028213262557983,
|
| 579 |
+
"loss": 56.0624,
|
| 580 |
+
"nll_loss": 1.4398643970489502,
|
| 581 |
+
"rewards/accuracies": 0.65625,
|
| 582 |
+
"rewards/chosen": -0.587210476398468,
|
| 583 |
+
"rewards/margins": 0.16420014202594757,
|
| 584 |
+
"rewards/rejected": -0.7514106631278992,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5594074074074074,
|
| 589 |
+
"grad_norm": 125.1372299194336,
|
| 590 |
+
"learning_rate": 5.471094152058592e-06,
|
| 591 |
+
"log_odds_chosen": 0.4360503554344177,
|
| 592 |
+
"log_odds_ratio": -0.589511513710022,
|
| 593 |
+
"logits/chosen": 234.1468505859375,
|
| 594 |
+
"logits/rejected": 225.77023315429688,
|
| 595 |
+
"logps/chosen": -1.1270663738250732,
|
| 596 |
+
"logps/rejected": -1.4443070888519287,
|
| 597 |
+
"loss": 55.663,
|
| 598 |
+
"nll_loss": 1.4153871536254883,
|
| 599 |
+
"rewards/accuracies": 0.6875,
|
| 600 |
+
"rewards/chosen": -0.5635331869125366,
|
| 601 |
+
"rewards/margins": 0.15862040221691132,
|
| 602 |
+
"rewards/rejected": -0.7221535444259644,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6068148148148147,
|
| 607 |
+
"grad_norm": 170.6956787109375,
|
| 608 |
+
"learning_rate": 5.1942192226683385e-06,
|
| 609 |
+
"log_odds_chosen": 0.4941047728061676,
|
| 610 |
+
"log_odds_ratio": -0.5777939558029175,
|
| 611 |
+
"logits/chosen": 225.4412078857422,
|
| 612 |
+
"logits/rejected": 230.6029052734375,
|
| 613 |
+
"logps/chosen": -1.1379103660583496,
|
| 614 |
+
"logps/rejected": -1.5056231021881104,
|
| 615 |
+
"loss": 54.5688,
|
| 616 |
+
"nll_loss": 1.4074928760528564,
|
| 617 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 618 |
+
"rewards/chosen": -0.5689551830291748,
|
| 619 |
+
"rewards/margins": 0.18385633826255798,
|
| 620 |
+
"rewards/rejected": -0.7528115510940552,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6542222222222223,
|
| 625 |
+
"grad_norm": 93.71338653564453,
|
| 626 |
+
"learning_rate": 4.916746091646808e-06,
|
| 627 |
+
"log_odds_chosen": 0.4696536958217621,
|
| 628 |
+
"log_odds_ratio": -0.5783597230911255,
|
| 629 |
+
"logits/chosen": 226.40939331054688,
|
| 630 |
+
"logits/rejected": 238.772705078125,
|
| 631 |
+
"logps/chosen": -1.1081968545913696,
|
| 632 |
+
"logps/rejected": -1.4637243747711182,
|
| 633 |
+
"loss": 54.5723,
|
| 634 |
+
"nll_loss": 1.4280409812927246,
|
| 635 |
+
"rewards/accuracies": 0.675000011920929,
|
| 636 |
+
"rewards/chosen": -0.5540984272956848,
|
| 637 |
+
"rewards/margins": 0.1777636557817459,
|
| 638 |
+
"rewards/rejected": -0.7318621873855591,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7016296296296296,
|
| 643 |
+
"grad_norm": 103.26416015625,
|
| 644 |
+
"learning_rate": 4.6395293854173395e-06,
|
| 645 |
+
"log_odds_chosen": 0.2048105001449585,
|
| 646 |
+
"log_odds_ratio": -0.6917891502380371,
|
| 647 |
+
"logits/chosen": 233.7334442138672,
|
| 648 |
+
"logits/rejected": 230.9724578857422,
|
| 649 |
+
"logps/chosen": -1.193915843963623,
|
| 650 |
+
"logps/rejected": -1.3438174724578857,
|
| 651 |
+
"loss": 55.4722,
|
| 652 |
+
"nll_loss": 1.475660800933838,
|
| 653 |
+
"rewards/accuracies": 0.5625,
|
| 654 |
+
"rewards/chosen": -0.5969579219818115,
|
| 655 |
+
"rewards/margins": 0.07495082169771194,
|
| 656 |
+
"rewards/rejected": -0.6719087362289429,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.749037037037037,
|
| 661 |
+
"grad_norm": 87.70002746582031,
|
| 662 |
+
"learning_rate": 4.363422940606435e-06,
|
| 663 |
+
"log_odds_chosen": 0.4551773965358734,
|
| 664 |
+
"log_odds_ratio": -0.5870417356491089,
|
| 665 |
+
"logits/chosen": 229.46359252929688,
|
| 666 |
+
"logits/rejected": 232.06924438476562,
|
| 667 |
+
"logps/chosen": -1.1224743127822876,
|
| 668 |
+
"logps/rejected": -1.4738192558288574,
|
| 669 |
+
"loss": 54.9357,
|
| 670 |
+
"nll_loss": 1.4060946702957153,
|
| 671 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 672 |
+
"rewards/chosen": -0.5612371563911438,
|
| 673 |
+
"rewards/margins": 0.17567242681980133,
|
| 674 |
+
"rewards/rejected": -0.7369096279144287,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.7964444444444445,
|
| 679 |
+
"grad_norm": 95.35106658935547,
|
| 680 |
+
"learning_rate": 4.089277174198694e-06,
|
| 681 |
+
"log_odds_chosen": 0.45670628547668457,
|
| 682 |
+
"log_odds_ratio": -0.5882663726806641,
|
| 683 |
+
"logits/chosen": 221.1391143798828,
|
| 684 |
+
"logits/rejected": 236.39956665039062,
|
| 685 |
+
"logps/chosen": -1.1329078674316406,
|
| 686 |
+
"logps/rejected": -1.476986289024353,
|
| 687 |
+
"loss": 55.4382,
|
| 688 |
+
"nll_loss": 1.4155741930007935,
|
| 689 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 690 |
+
"rewards/chosen": -0.5664539337158203,
|
| 691 |
+
"rewards/margins": 0.1720392405986786,
|
| 692 |
+
"rewards/rejected": -0.7384931445121765,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8438518518518519,
|
| 697 |
+
"grad_norm": 114.8470458984375,
|
| 698 |
+
"learning_rate": 3.817936464224367e-06,
|
| 699 |
+
"log_odds_chosen": 0.5381921529769897,
|
| 700 |
+
"log_odds_ratio": -0.5697619915008545,
|
| 701 |
+
"logits/chosen": 229.50509643554688,
|
| 702 |
+
"logits/rejected": 242.05123901367188,
|
| 703 |
+
"logps/chosen": -1.1255033016204834,
|
| 704 |
+
"logps/rejected": -1.5437520742416382,
|
| 705 |
+
"loss": 54.134,
|
| 706 |
+
"nll_loss": 1.4217894077301025,
|
| 707 |
+
"rewards/accuracies": 0.71875,
|
| 708 |
+
"rewards/chosen": -0.5627516508102417,
|
| 709 |
+
"rewards/margins": 0.209124356508255,
|
| 710 |
+
"rewards/rejected": -0.7718760371208191,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8912592592592592,
|
| 715 |
+
"grad_norm": 118.22327423095703,
|
| 716 |
+
"learning_rate": 3.55023654904709e-06,
|
| 717 |
+
"log_odds_chosen": 0.3448692858219147,
|
| 718 |
+
"log_odds_ratio": -0.6549310088157654,
|
| 719 |
+
"logits/chosen": 225.15869140625,
|
| 720 |
+
"logits/rejected": 235.4357147216797,
|
| 721 |
+
"logps/chosen": -1.1163337230682373,
|
| 722 |
+
"logps/rejected": -1.368422508239746,
|
| 723 |
+
"loss": 55.3987,
|
| 724 |
+
"nll_loss": 1.3597387075424194,
|
| 725 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 726 |
+
"rewards/chosen": -0.5581668615341187,
|
| 727 |
+
"rewards/margins": 0.12604445219039917,
|
| 728 |
+
"rewards/rejected": -0.684211254119873,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9386666666666668,
|
| 733 |
+
"grad_norm": 90.22616577148438,
|
| 734 |
+
"learning_rate": 3.2870019532620744e-06,
|
| 735 |
+
"log_odds_chosen": 0.401312917470932,
|
| 736 |
+
"log_odds_ratio": -0.6004291772842407,
|
| 737 |
+
"logits/chosen": 224.0797119140625,
|
| 738 |
+
"logits/rejected": 241.08871459960938,
|
| 739 |
+
"logps/chosen": -1.0669735670089722,
|
| 740 |
+
"logps/rejected": -1.3696752786636353,
|
| 741 |
+
"loss": 53.038,
|
| 742 |
+
"nll_loss": 1.351078987121582,
|
| 743 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 744 |
+
"rewards/chosen": -0.5334867835044861,
|
| 745 |
+
"rewards/margins": 0.15135079622268677,
|
| 746 |
+
"rewards/rejected": -0.6848376393318176,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.986074074074074,
|
| 751 |
+
"grad_norm": 131.29962158203125,
|
| 752 |
+
"learning_rate": 3.0290434481330746e-06,
|
| 753 |
+
"log_odds_chosen": 0.3439275622367859,
|
| 754 |
+
"log_odds_ratio": -0.6156661510467529,
|
| 755 |
+
"logits/chosen": 233.643798828125,
|
| 756 |
+
"logits/rejected": 226.2475128173828,
|
| 757 |
+
"logps/chosen": -1.0474224090576172,
|
| 758 |
+
"logps/rejected": -1.2926290035247803,
|
| 759 |
+
"loss": 51.5863,
|
| 760 |
+
"nll_loss": 1.3270413875579834,
|
| 761 |
+
"rewards/accuracies": 0.606249988079071,
|
| 762 |
+
"rewards/chosen": -0.5237112045288086,
|
| 763 |
+
"rewards/margins": 0.12260331213474274,
|
| 764 |
+
"rewards/rejected": -0.6463145017623901,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0284444444444443,
|
| 769 |
+
"grad_norm": 96.40355682373047,
|
| 770 |
+
"learning_rate": 2.77715555439007e-06,
|
| 771 |
+
"log_odds_chosen": 0.5393110513687134,
|
| 772 |
+
"log_odds_ratio": -0.546492338180542,
|
| 773 |
+
"logits/chosen": 227.2359619140625,
|
| 774 |
+
"logits/rejected": 236.05026245117188,
|
| 775 |
+
"logps/chosen": -1.061828374862671,
|
| 776 |
+
"logps/rejected": -1.4429984092712402,
|
| 777 |
+
"loss": 46.8863,
|
| 778 |
+
"nll_loss": 1.3469436168670654,
|
| 779 |
+
"rewards/accuracies": 0.7342657446861267,
|
| 780 |
+
"rewards/chosen": -0.5309141874313354,
|
| 781 |
+
"rewards/margins": 0.19058500230312347,
|
| 782 |
+
"rewards/rejected": -0.7214992046356201,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.075851851851852,
|
| 787 |
+
"grad_norm": 105.6058120727539,
|
| 788 |
+
"learning_rate": 2.532114095079137e-06,
|
| 789 |
+
"log_odds_chosen": 0.6614600419998169,
|
| 790 |
+
"log_odds_ratio": -0.5195887684822083,
|
| 791 |
+
"logits/chosen": 222.1413116455078,
|
| 792 |
+
"logits/rejected": 226.68173217773438,
|
| 793 |
+
"logps/chosen": -0.9896618723869324,
|
| 794 |
+
"logps/rejected": -1.4654362201690674,
|
| 795 |
+
"loss": 50.1408,
|
| 796 |
+
"nll_loss": 1.314392328262329,
|
| 797 |
+
"rewards/accuracies": 0.71875,
|
| 798 |
+
"rewards/chosen": -0.4948309361934662,
|
| 799 |
+
"rewards/margins": 0.23788709938526154,
|
| 800 |
+
"rewards/rejected": -0.7327181100845337,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1232592592592594,
|
| 805 |
+
"grad_norm": 86.1953353881836,
|
| 806 |
+
"learning_rate": 2.2946738060017947e-06,
|
| 807 |
+
"log_odds_chosen": 0.6218869686126709,
|
| 808 |
+
"log_odds_ratio": -0.5385203957557678,
|
| 809 |
+
"logits/chosen": 225.3985137939453,
|
| 810 |
+
"logits/rejected": 221.6757049560547,
|
| 811 |
+
"logps/chosen": -0.9887048602104187,
|
| 812 |
+
"logps/rejected": -1.440615177154541,
|
| 813 |
+
"loss": 50.805,
|
| 814 |
+
"nll_loss": 1.342362880706787,
|
| 815 |
+
"rewards/accuracies": 0.75,
|
| 816 |
+
"rewards/chosen": -0.49435243010520935,
|
| 817 |
+
"rewards/margins": 0.22595517337322235,
|
| 818 |
+
"rewards/rejected": -0.7203075885772705,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.1706666666666665,
|
| 823 |
+
"grad_norm": 123.00335693359375,
|
| 824 |
+
"learning_rate": 2.0655660111037685e-06,
|
| 825 |
+
"log_odds_chosen": 0.7268826365470886,
|
| 826 |
+
"log_odds_ratio": -0.4982364773750305,
|
| 827 |
+
"logits/chosen": 222.36849975585938,
|
| 828 |
+
"logits/rejected": 233.1688995361328,
|
| 829 |
+
"logps/chosen": -1.0594282150268555,
|
| 830 |
+
"logps/rejected": -1.601413369178772,
|
| 831 |
+
"loss": 50.4757,
|
| 832 |
+
"nll_loss": 1.3385509252548218,
|
| 833 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 834 |
+
"rewards/chosen": -0.5297141075134277,
|
| 835 |
+
"rewards/margins": 0.2709924578666687,
|
| 836 |
+
"rewards/rejected": -0.800706684589386,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.218074074074074,
|
| 841 |
+
"grad_norm": 119.10016632080078,
|
| 842 |
+
"learning_rate": 1.8454963699730471e-06,
|
| 843 |
+
"log_odds_chosen": 0.6008008718490601,
|
| 844 |
+
"log_odds_ratio": -0.5274027585983276,
|
| 845 |
+
"logits/chosen": 214.6153106689453,
|
| 846 |
+
"logits/rejected": 219.9340057373047,
|
| 847 |
+
"logps/chosen": -1.0045088529586792,
|
| 848 |
+
"logps/rejected": -1.422736406326294,
|
| 849 |
+
"loss": 50.8648,
|
| 850 |
+
"nll_loss": 1.3140861988067627,
|
| 851 |
+
"rewards/accuracies": 0.75,
|
| 852 |
+
"rewards/chosen": -0.5022544264793396,
|
| 853 |
+
"rewards/margins": 0.20911379158496857,
|
| 854 |
+
"rewards/rejected": -0.711368203163147,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2654814814814817,
|
| 859 |
+
"grad_norm": 95.31659698486328,
|
| 860 |
+
"learning_rate": 1.6351427043849955e-06,
|
| 861 |
+
"log_odds_chosen": 0.5878480076789856,
|
| 862 |
+
"log_odds_ratio": -0.5382899045944214,
|
| 863 |
+
"logits/chosen": 223.2783203125,
|
| 864 |
+
"logits/rejected": 223.6912384033203,
|
| 865 |
+
"logps/chosen": -1.025895595550537,
|
| 866 |
+
"logps/rejected": -1.4464019536972046,
|
| 867 |
+
"loss": 50.6412,
|
| 868 |
+
"nll_loss": 1.3646161556243896,
|
| 869 |
+
"rewards/accuracies": 0.706250011920929,
|
| 870 |
+
"rewards/chosen": -0.5129477977752686,
|
| 871 |
+
"rewards/margins": 0.21025320887565613,
|
| 872 |
+
"rewards/rejected": -0.7232009768486023,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.3128888888888888,
|
| 877 |
+
"grad_norm": 114.522705078125,
|
| 878 |
+
"learning_rate": 1.4351529105888735e-06,
|
| 879 |
+
"log_odds_chosen": 0.7168388366699219,
|
| 880 |
+
"log_odds_ratio": -0.4856956899166107,
|
| 881 |
+
"logits/chosen": 215.4445343017578,
|
| 882 |
+
"logits/rejected": 233.8488311767578,
|
| 883 |
+
"logps/chosen": -0.9942634701728821,
|
| 884 |
+
"logps/rejected": -1.5078818798065186,
|
| 885 |
+
"loss": 49.269,
|
| 886 |
+
"nll_loss": 1.3349363803863525,
|
| 887 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 888 |
+
"rewards/chosen": -0.49713173508644104,
|
| 889 |
+
"rewards/margins": 0.25680914521217346,
|
| 890 |
+
"rewards/rejected": -0.7539409399032593,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3602962962962963,
|
| 895 |
+
"grad_norm": 162.42523193359375,
|
| 896 |
+
"learning_rate": 1.2461429637659466e-06,
|
| 897 |
+
"log_odds_chosen": 0.7001165151596069,
|
| 898 |
+
"log_odds_ratio": -0.48200806975364685,
|
| 899 |
+
"logits/chosen": 218.6329803466797,
|
| 900 |
+
"logits/rejected": 225.58749389648438,
|
| 901 |
+
"logps/chosen": -1.0117690563201904,
|
| 902 |
+
"logps/rejected": -1.5104820728302002,
|
| 903 |
+
"loss": 49.1429,
|
| 904 |
+
"nll_loss": 1.3088269233703613,
|
| 905 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 906 |
+
"rewards/chosen": -0.5058845281600952,
|
| 907 |
+
"rewards/margins": 0.24935641884803772,
|
| 908 |
+
"rewards/rejected": -0.7552410364151001,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.407703703703704,
|
| 913 |
+
"grad_norm": 174.11715698242188,
|
| 914 |
+
"learning_rate": 1.0686950208055486e-06,
|
| 915 |
+
"log_odds_chosen": 0.8261364698410034,
|
| 916 |
+
"log_odds_ratio": -0.47225421667099,
|
| 917 |
+
"logits/chosen": 215.776611328125,
|
| 918 |
+
"logits/rejected": 220.90621948242188,
|
| 919 |
+
"logps/chosen": -0.9497050046920776,
|
| 920 |
+
"logps/rejected": -1.5556175708770752,
|
| 921 |
+
"loss": 48.4801,
|
| 922 |
+
"nll_loss": 1.2853540182113647,
|
| 923 |
+
"rewards/accuracies": 0.78125,
|
| 924 |
+
"rewards/chosen": -0.4748525023460388,
|
| 925 |
+
"rewards/margins": 0.302956223487854,
|
| 926 |
+
"rewards/rejected": -0.7778087854385376,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.455111111111111,
|
| 931 |
+
"grad_norm": 110.5379638671875,
|
| 932 |
+
"learning_rate": 9.033556272426075e-07,
|
| 933 |
+
"log_odds_chosen": 0.8216450810432434,
|
| 934 |
+
"log_odds_ratio": -0.4542720913887024,
|
| 935 |
+
"logits/chosen": 208.1492156982422,
|
| 936 |
+
"logits/rejected": 226.76187133789062,
|
| 937 |
+
"logps/chosen": -0.9726914167404175,
|
| 938 |
+
"logps/rejected": -1.557952642440796,
|
| 939 |
+
"loss": 49.1974,
|
| 940 |
+
"nll_loss": 1.2973281145095825,
|
| 941 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 942 |
+
"rewards/chosen": -0.48634570837020874,
|
| 943 |
+
"rewards/margins": 0.292630672454834,
|
| 944 |
+
"rewards/rejected": -0.778976321220398,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5025185185185186,
|
| 949 |
+
"grad_norm": 110.2140884399414,
|
| 950 |
+
"learning_rate": 7.506340338793111e-07,
|
| 951 |
+
"log_odds_chosen": 0.660784125328064,
|
| 952 |
+
"log_odds_ratio": -0.5323556065559387,
|
| 953 |
+
"logits/chosen": 219.0674591064453,
|
| 954 |
+
"logits/rejected": 217.68307495117188,
|
| 955 |
+
"logps/chosen": -0.9799796938896179,
|
| 956 |
+
"logps/rejected": -1.453094720840454,
|
| 957 |
+
"loss": 48.5513,
|
| 958 |
+
"nll_loss": 1.2660871744155884,
|
| 959 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 960 |
+
"rewards/chosen": -0.48998984694480896,
|
| 961 |
+
"rewards/margins": 0.23655752837657928,
|
| 962 |
+
"rewards/rejected": -0.726547360420227,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.549925925925926,
|
| 967 |
+
"grad_norm": 125.52155303955078,
|
| 968 |
+
"learning_rate": 6.110006282757897e-07,
|
| 969 |
+
"log_odds_chosen": 0.838211178779602,
|
| 970 |
+
"log_odds_ratio": -0.4636700749397278,
|
| 971 |
+
"logits/chosen": 210.531982421875,
|
| 972 |
+
"logits/rejected": 217.48464965820312,
|
| 973 |
+
"logps/chosen": -0.9940751791000366,
|
| 974 |
+
"logps/rejected": -1.5986590385437012,
|
| 975 |
+
"loss": 48.4161,
|
| 976 |
+
"nll_loss": 1.2988741397857666,
|
| 977 |
+
"rewards/accuracies": 0.800000011920929,
|
| 978 |
+
"rewards/chosen": -0.4970375895500183,
|
| 979 |
+
"rewards/margins": 0.3022919297218323,
|
| 980 |
+
"rewards/rejected": -0.7993295192718506,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.5973333333333333,
|
| 985 |
+
"grad_norm": 99.75615692138672,
|
| 986 |
+
"learning_rate": 4.848854859408731e-07,
|
| 987 |
+
"log_odds_chosen": 0.8387314677238464,
|
| 988 |
+
"log_odds_ratio": -0.45085686445236206,
|
| 989 |
+
"logits/chosen": 213.62991333007812,
|
| 990 |
+
"logits/rejected": 234.63095092773438,
|
| 991 |
+
"logps/chosen": -0.9795564413070679,
|
| 992 |
+
"logps/rejected": -1.5703213214874268,
|
| 993 |
+
"loss": 50.0056,
|
| 994 |
+
"nll_loss": 1.3050158023834229,
|
| 995 |
+
"rewards/accuracies": 0.8125,
|
| 996 |
+
"rewards/chosen": -0.48977822065353394,
|
| 997 |
+
"rewards/margins": 0.29538241028785706,
|
| 998 |
+
"rewards/rejected": -0.7851606607437134,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.644740740740741,
|
| 1003 |
+
"grad_norm": 120.89396667480469,
|
| 1004 |
+
"learning_rate": 3.7267704568529015e-07,
|
| 1005 |
+
"log_odds_chosen": 0.7018845081329346,
|
| 1006 |
+
"log_odds_ratio": -0.5149911046028137,
|
| 1007 |
+
"logits/chosen": 216.6708526611328,
|
| 1008 |
+
"logits/rejected": 217.61160278320312,
|
| 1009 |
+
"logps/chosen": -0.9656041860580444,
|
| 1010 |
+
"logps/rejected": -1.4782164096832275,
|
| 1011 |
+
"loss": 48.7583,
|
| 1012 |
+
"nll_loss": 1.2335693836212158,
|
| 1013 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1014 |
+
"rewards/chosen": -0.4828020930290222,
|
| 1015 |
+
"rewards/margins": 0.25630611181259155,
|
| 1016 |
+
"rewards/rejected": -0.7391082048416138,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.6921481481481484,
|
| 1021 |
+
"grad_norm": 121.43464660644531,
|
| 1022 |
+
"learning_rate": 2.7472091321728067e-07,
|
| 1023 |
+
"log_odds_chosen": 0.6797521710395813,
|
| 1024 |
+
"log_odds_ratio": -0.5134819746017456,
|
| 1025 |
+
"logits/chosen": 214.6907501220703,
|
| 1026 |
+
"logits/rejected": 224.23226928710938,
|
| 1027 |
+
"logps/chosen": -0.9947472810745239,
|
| 1028 |
+
"logps/rejected": -1.4958022832870483,
|
| 1029 |
+
"loss": 50.1254,
|
| 1030 |
+
"nll_loss": 1.3161695003509521,
|
| 1031 |
+
"rewards/accuracies": 0.75,
|
| 1032 |
+
"rewards/chosen": -0.49737364053726196,
|
| 1033 |
+
"rewards/margins": 0.25052759051322937,
|
| 1034 |
+
"rewards/rejected": -0.7479011416435242,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.7395555555555555,
|
| 1039 |
+
"grad_norm": 128.531005859375,
|
| 1040 |
+
"learning_rate": 1.9131879666558385e-07,
|
| 1041 |
+
"log_odds_chosen": 0.7722429037094116,
|
| 1042 |
+
"log_odds_ratio": -0.4660676121711731,
|
| 1043 |
+
"logits/chosen": 220.88125610351562,
|
| 1044 |
+
"logits/rejected": 225.862548828125,
|
| 1045 |
+
"logps/chosen": -0.9653012156486511,
|
| 1046 |
+
"logps/rejected": -1.5093042850494385,
|
| 1047 |
+
"loss": 49.7205,
|
| 1048 |
+
"nll_loss": 1.3300297260284424,
|
| 1049 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1050 |
+
"rewards/chosen": -0.48265060782432556,
|
| 1051 |
+
"rewards/margins": 0.27200156450271606,
|
| 1052 |
+
"rewards/rejected": -0.7546521425247192,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.786962962962963,
|
| 1057 |
+
"grad_norm": 99.14717102050781,
|
| 1058 |
+
"learning_rate": 1.2272757730841744e-07,
|
| 1059 |
+
"log_odds_chosen": 0.7158989906311035,
|
| 1060 |
+
"log_odds_ratio": -0.5291553735733032,
|
| 1061 |
+
"logits/chosen": 215.935302734375,
|
| 1062 |
+
"logits/rejected": 213.4801788330078,
|
| 1063 |
+
"logps/chosen": -1.0447461605072021,
|
| 1064 |
+
"logps/rejected": -1.5658835172653198,
|
| 1065 |
+
"loss": 49.2845,
|
| 1066 |
+
"nll_loss": 1.304596185684204,
|
| 1067 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1068 |
+
"rewards/chosen": -0.5223730802536011,
|
| 1069 |
+
"rewards/margins": 0.260568767786026,
|
| 1070 |
+
"rewards/rejected": -0.7829417586326599,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.83437037037037,
|
| 1075 |
+
"grad_norm": 123.91796875,
|
| 1076 |
+
"learning_rate": 6.91585183706428e-08,
|
| 1077 |
+
"log_odds_chosen": 0.8705388903617859,
|
| 1078 |
+
"log_odds_ratio": -0.4770691394805908,
|
| 1079 |
+
"logits/chosen": 206.53121948242188,
|
| 1080 |
+
"logits/rejected": 225.6171417236328,
|
| 1081 |
+
"logps/chosen": -0.9915706515312195,
|
| 1082 |
+
"logps/rejected": -1.6319612264633179,
|
| 1083 |
+
"loss": 49.6516,
|
| 1084 |
+
"nll_loss": 1.3192683458328247,
|
| 1085 |
+
"rewards/accuracies": 0.78125,
|
| 1086 |
+
"rewards/chosen": -0.49578532576560974,
|
| 1087 |
+
"rewards/margins": 0.32019534707069397,
|
| 1088 |
+
"rewards/rejected": -0.8159806132316589,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.8817777777777778,
|
| 1093 |
+
"grad_norm": 100.9173812866211,
|
| 1094 |
+
"learning_rate": 3.077661432604184e-08,
|
| 1095 |
+
"log_odds_chosen": 0.7253429293632507,
|
| 1096 |
+
"log_odds_ratio": -0.49313846230506897,
|
| 1097 |
+
"logits/chosen": 211.61221313476562,
|
| 1098 |
+
"logits/rejected": 224.7270050048828,
|
| 1099 |
+
"logps/chosen": -1.0019609928131104,
|
| 1100 |
+
"logps/rejected": -1.5113885402679443,
|
| 1101 |
+
"loss": 48.0438,
|
| 1102 |
+
"nll_loss": 1.2643382549285889,
|
| 1103 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1104 |
+
"rewards/chosen": -0.5009804964065552,
|
| 1105 |
+
"rewards/margins": 0.254713773727417,
|
| 1106 |
+
"rewards/rejected": -0.7556942701339722,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9291851851851853,
|
| 1111 |
+
"grad_norm": 134.68287658691406,
|
| 1112 |
+
"learning_rate": 7.700082708883006e-09,
|
| 1113 |
+
"log_odds_chosen": 0.7188720703125,
|
| 1114 |
+
"log_odds_ratio": -0.4834628999233246,
|
| 1115 |
+
"logits/chosen": 216.98306274414062,
|
| 1116 |
+
"logits/rejected": 232.9964599609375,
|
| 1117 |
+
"logps/chosen": -0.9981603622436523,
|
| 1118 |
+
"logps/rejected": -1.4897390604019165,
|
| 1119 |
+
"loss": 49.1365,
|
| 1120 |
+
"nll_loss": 1.3062077760696411,
|
| 1121 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1122 |
+
"rewards/chosen": -0.49908018112182617,
|
| 1123 |
+
"rewards/margins": 0.24578937888145447,
|
| 1124 |
+
"rewards/rejected": -0.7448695302009583,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.9765925925925925,
|
| 1129 |
+
"grad_norm": 172.74346923828125,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 0.808211624622345,
|
| 1132 |
+
"log_odds_ratio": -0.4583892822265625,
|
| 1133 |
+
"logits/chosen": 211.4867706298828,
|
| 1134 |
+
"logits/rejected": 224.31991577148438,
|
| 1135 |
+
"logps/chosen": -0.9515358209609985,
|
| 1136 |
+
"logps/rejected": -1.5278295278549194,
|
| 1137 |
+
"loss": 47.6357,
|
| 1138 |
+
"nll_loss": 1.26894211769104,
|
| 1139 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1140 |
+
"rewards/chosen": -0.47576791048049927,
|
| 1141 |
+
"rewards/margins": 0.2881469428539276,
|
| 1142 |
+
"rewards/rejected": -0.7639147639274597,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.9765925925925925,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 95.27591378348214,
|
| 1150 |
+
"train_runtime": 8112.4006,
|
| 1151 |
+
"train_samples_per_second": 2.496,
|
| 1152 |
+
"train_steps_per_second": 0.039
|
| 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 |
+
}
|