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-7e-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-7e-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-7e-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/rds7ebnu)
|
| 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": 60.23838462224082,
|
| 5 |
+
"train_runtime": 8115.1309,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.495,
|
| 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": 60.23838462224082,
|
| 5 |
+
"train_runtime": 8115.1309,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.495,
|
| 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": 631.015625,
|
| 14 |
+
"learning_rate": 1.1718749999999999e-05,
|
| 15 |
+
"log_odds_chosen": 4.92927360534668,
|
| 16 |
+
"log_odds_ratio": -9.756586074829102,
|
| 17 |
+
"logits/chosen": 138.57809448242188,
|
| 18 |
+
"logits/rejected": 153.22152709960938,
|
| 19 |
+
"logps/chosen": -20.633487701416016,
|
| 20 |
+
"logps/rejected": -25.562952041625977,
|
| 21 |
+
"loss": 452.9617,
|
| 22 |
+
"nll_loss": 8.288494110107422,
|
| 23 |
+
"rewards/accuracies": 0.53125,
|
| 24 |
+
"rewards/chosen": -10.316743850708008,
|
| 25 |
+
"rewards/margins": 2.4647319316864014,
|
| 26 |
+
"rewards/rejected": -12.781476020812988,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 552.6162719726562,
|
| 32 |
+
"learning_rate": 2.3437499999999997e-05,
|
| 33 |
+
"log_odds_chosen": 2.3541274070739746,
|
| 34 |
+
"log_odds_ratio": -7.061516761779785,
|
| 35 |
+
"logits/chosen": 139.1152801513672,
|
| 36 |
+
"logits/rejected": 163.85743713378906,
|
| 37 |
+
"logps/chosen": -15.125625610351562,
|
| 38 |
+
"logps/rejected": -17.478649139404297,
|
| 39 |
+
"loss": 374.0408,
|
| 40 |
+
"nll_loss": 7.115235805511475,
|
| 41 |
+
"rewards/accuracies": 0.543749988079071,
|
| 42 |
+
"rewards/chosen": -7.562812805175781,
|
| 43 |
+
"rewards/margins": 1.1765111684799194,
|
| 44 |
+
"rewards/rejected": -8.739324569702148,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 2310.692626953125,
|
| 50 |
+
"learning_rate": 3.515625e-05,
|
| 51 |
+
"log_odds_chosen": 4.172502040863037,
|
| 52 |
+
"log_odds_ratio": -6.55011510848999,
|
| 53 |
+
"logits/chosen": 125.50711822509766,
|
| 54 |
+
"logits/rejected": 151.8711395263672,
|
| 55 |
+
"logps/chosen": -17.319055557250977,
|
| 56 |
+
"logps/rejected": -21.48957061767578,
|
| 57 |
+
"loss": 325.2039,
|
| 58 |
+
"nll_loss": 7.520108222961426,
|
| 59 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 60 |
+
"rewards/chosen": -8.659527778625488,
|
| 61 |
+
"rewards/margins": 2.0852572917938232,
|
| 62 |
+
"rewards/rejected": -10.74478530883789,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 438.0211486816406,
|
| 68 |
+
"learning_rate": 4.6874999999999994e-05,
|
| 69 |
+
"log_odds_chosen": -0.5363087058067322,
|
| 70 |
+
"log_odds_ratio": -2.402191638946533,
|
| 71 |
+
"logits/chosen": 162.6514129638672,
|
| 72 |
+
"logits/rejected": 166.47042846679688,
|
| 73 |
+
"logps/chosen": -5.853331565856934,
|
| 74 |
+
"logps/rejected": -5.3130202293396,
|
| 75 |
+
"loss": 147.371,
|
| 76 |
+
"nll_loss": 3.4735121726989746,
|
| 77 |
+
"rewards/accuracies": 0.518750011920929,
|
| 78 |
+
"rewards/chosen": -2.926665782928467,
|
| 79 |
+
"rewards/margins": -0.27015575766563416,
|
| 80 |
+
"rewards/rejected": -2.6565101146698,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 358.36737060546875,
|
| 86 |
+
"learning_rate": 5.859375e-05,
|
| 87 |
+
"log_odds_chosen": 0.3639217019081116,
|
| 88 |
+
"log_odds_ratio": -0.7449954748153687,
|
| 89 |
+
"logits/chosen": 196.79856872558594,
|
| 90 |
+
"logits/rejected": 204.99942016601562,
|
| 91 |
+
"logps/chosen": -1.7752330303192139,
|
| 92 |
+
"logps/rejected": -2.1123547554016113,
|
| 93 |
+
"loss": 77.053,
|
| 94 |
+
"nll_loss": 2.062661647796631,
|
| 95 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 96 |
+
"rewards/chosen": -0.8876165151596069,
|
| 97 |
+
"rewards/margins": 0.16856074333190918,
|
| 98 |
+
"rewards/rejected": -1.0561773777008057,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 316.6930236816406,
|
| 104 |
+
"learning_rate": 7.03125e-05,
|
| 105 |
+
"log_odds_chosen": 0.1705448478460312,
|
| 106 |
+
"log_odds_ratio": -0.7665665745735168,
|
| 107 |
+
"logits/chosen": 221.40567016601562,
|
| 108 |
+
"logits/rejected": 226.0666046142578,
|
| 109 |
+
"logps/chosen": -1.6369438171386719,
|
| 110 |
+
"logps/rejected": -1.7822250127792358,
|
| 111 |
+
"loss": 73.8902,
|
| 112 |
+
"nll_loss": 1.9145492315292358,
|
| 113 |
+
"rewards/accuracies": 0.5625,
|
| 114 |
+
"rewards/chosen": -0.8184719085693359,
|
| 115 |
+
"rewards/margins": 0.07264064252376556,
|
| 116 |
+
"rewards/rejected": -0.8911125063896179,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 253.6866455078125,
|
| 122 |
+
"learning_rate": 7.497920636023175e-05,
|
| 123 |
+
"log_odds_chosen": 0.025636177510023117,
|
| 124 |
+
"log_odds_ratio": -0.821603000164032,
|
| 125 |
+
"logits/chosen": 216.9961700439453,
|
| 126 |
+
"logits/rejected": 218.8159637451172,
|
| 127 |
+
"logps/chosen": -1.591193675994873,
|
| 128 |
+
"logps/rejected": -1.6127433776855469,
|
| 129 |
+
"loss": 69.6548,
|
| 130 |
+
"nll_loss": 1.8054049015045166,
|
| 131 |
+
"rewards/accuracies": 0.518750011920929,
|
| 132 |
+
"rewards/chosen": -0.7955968379974365,
|
| 133 |
+
"rewards/margins": 0.010774785652756691,
|
| 134 |
+
"rewards/rejected": -0.8063716888427734,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 118.463623046875,
|
| 140 |
+
"learning_rate": 7.485221761715193e-05,
|
| 141 |
+
"log_odds_chosen": 0.2294560670852661,
|
| 142 |
+
"log_odds_ratio": -0.6809016466140747,
|
| 143 |
+
"logits/chosen": 203.59107971191406,
|
| 144 |
+
"logits/rejected": 201.36801147460938,
|
| 145 |
+
"logps/chosen": -1.3837480545043945,
|
| 146 |
+
"logps/rejected": -1.5675389766693115,
|
| 147 |
+
"loss": 65.3998,
|
| 148 |
+
"nll_loss": 1.682397484779358,
|
| 149 |
+
"rewards/accuracies": 0.53125,
|
| 150 |
+
"rewards/chosen": -0.6918740272521973,
|
| 151 |
+
"rewards/margins": 0.09189549833536148,
|
| 152 |
+
"rewards/rejected": -0.7837694883346558,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 168.12261962890625,
|
| 158 |
+
"learning_rate": 7.461018280806719e-05,
|
| 159 |
+
"log_odds_chosen": 0.19497309625148773,
|
| 160 |
+
"log_odds_ratio": -0.7434221506118774,
|
| 161 |
+
"logits/chosen": 181.99807739257812,
|
| 162 |
+
"logits/rejected": 182.77682495117188,
|
| 163 |
+
"logps/chosen": -1.381994366645813,
|
| 164 |
+
"logps/rejected": -1.541455626487732,
|
| 165 |
+
"loss": 63.7627,
|
| 166 |
+
"nll_loss": 1.6508877277374268,
|
| 167 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 168 |
+
"rewards/chosen": -0.6909971833229065,
|
| 169 |
+
"rewards/margins": 0.07973069697618484,
|
| 170 |
+
"rewards/rejected": -0.770727813243866,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 278.8425598144531,
|
| 176 |
+
"learning_rate": 7.425384740819691e-05,
|
| 177 |
+
"log_odds_chosen": 0.21113845705986023,
|
| 178 |
+
"log_odds_ratio": -0.6987509727478027,
|
| 179 |
+
"logits/chosen": 176.6704864501953,
|
| 180 |
+
"logits/rejected": 177.52133178710938,
|
| 181 |
+
"logps/chosen": -1.358520269393921,
|
| 182 |
+
"logps/rejected": -1.5214390754699707,
|
| 183 |
+
"loss": 64.9056,
|
| 184 |
+
"nll_loss": 1.7045795917510986,
|
| 185 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 186 |
+
"rewards/chosen": -0.6792601346969604,
|
| 187 |
+
"rewards/margins": 0.08145944774150848,
|
| 188 |
+
"rewards/rejected": -0.7607195377349854,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 85.80127716064453,
|
| 194 |
+
"learning_rate": 7.378430894236152e-05,
|
| 195 |
+
"log_odds_chosen": 0.12195976823568344,
|
| 196 |
+
"log_odds_ratio": -0.7067708373069763,
|
| 197 |
+
"logits/chosen": 183.97714233398438,
|
| 198 |
+
"logits/rejected": 180.34552001953125,
|
| 199 |
+
"logps/chosen": -1.2856115102767944,
|
| 200 |
+
"logps/rejected": -1.3806015253067017,
|
| 201 |
+
"loss": 61.2303,
|
| 202 |
+
"nll_loss": 1.5915721654891968,
|
| 203 |
+
"rewards/accuracies": 0.574999988079071,
|
| 204 |
+
"rewards/chosen": -0.6428057551383972,
|
| 205 |
+
"rewards/margins": 0.04749499261379242,
|
| 206 |
+
"rewards/rejected": -0.6903007626533508,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 93.30135345458984,
|
| 212 |
+
"learning_rate": 7.320301360456979e-05,
|
| 213 |
+
"log_odds_chosen": 0.25577789545059204,
|
| 214 |
+
"log_odds_ratio": -0.6447214484214783,
|
| 215 |
+
"logits/chosen": 187.6861572265625,
|
| 216 |
+
"logits/rejected": 176.8941192626953,
|
| 217 |
+
"logps/chosen": -1.1724777221679688,
|
| 218 |
+
"logps/rejected": -1.3559234142303467,
|
| 219 |
+
"loss": 58.3784,
|
| 220 |
+
"nll_loss": 1.4778730869293213,
|
| 221 |
+
"rewards/accuracies": 0.606249988079071,
|
| 222 |
+
"rewards/chosen": -0.5862388610839844,
|
| 223 |
+
"rewards/margins": 0.09172289073467255,
|
| 224 |
+
"rewards/rejected": -0.6779617071151733,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 94.57402801513672,
|
| 230 |
+
"learning_rate": 7.251175180369343e-05,
|
| 231 |
+
"log_odds_chosen": 0.20048114657402039,
|
| 232 |
+
"log_odds_ratio": -0.6868171095848083,
|
| 233 |
+
"logits/chosen": 186.01736450195312,
|
| 234 |
+
"logits/rejected": 187.6791534423828,
|
| 235 |
+
"logps/chosen": -1.2308342456817627,
|
| 236 |
+
"logps/rejected": -1.3798383474349976,
|
| 237 |
+
"loss": 57.794,
|
| 238 |
+
"nll_loss": 1.496966004371643,
|
| 239 |
+
"rewards/accuracies": 0.5625,
|
| 240 |
+
"rewards/chosen": -0.6154171228408813,
|
| 241 |
+
"rewards/margins": 0.07450205832719803,
|
| 242 |
+
"rewards/rejected": -0.6899191737174988,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 152.06646728515625,
|
| 248 |
+
"learning_rate": 7.171265264894856e-05,
|
| 249 |
+
"log_odds_chosen": 0.27526047825813293,
|
| 250 |
+
"log_odds_ratio": -0.6563318967819214,
|
| 251 |
+
"logits/chosen": 200.08290100097656,
|
| 252 |
+
"logits/rejected": 180.6443328857422,
|
| 253 |
+
"logps/chosen": -1.1693233251571655,
|
| 254 |
+
"logps/rejected": -1.3890639543533325,
|
| 255 |
+
"loss": 57.0621,
|
| 256 |
+
"nll_loss": 1.482000708580017,
|
| 257 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 258 |
+
"rewards/chosen": -0.5846616625785828,
|
| 259 |
+
"rewards/margins": 0.10987037420272827,
|
| 260 |
+
"rewards/rejected": -0.6945319771766663,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 103.90552520751953,
|
| 266 |
+
"learning_rate": 7.080817739216882e-05,
|
| 267 |
+
"log_odds_chosen": 0.3350622057914734,
|
| 268 |
+
"log_odds_ratio": -0.6169534921646118,
|
| 269 |
+
"logits/chosen": 186.0503692626953,
|
| 270 |
+
"logits/rejected": 194.65689086914062,
|
| 271 |
+
"logps/chosen": -1.1045267581939697,
|
| 272 |
+
"logps/rejected": -1.351041555404663,
|
| 273 |
+
"loss": 55.9369,
|
| 274 |
+
"nll_loss": 1.4453600645065308,
|
| 275 |
+
"rewards/accuracies": 0.643750011920929,
|
| 276 |
+
"rewards/chosen": -0.5522633790969849,
|
| 277 |
+
"rewards/margins": 0.12325732409954071,
|
| 278 |
+
"rewards/rejected": -0.6755207777023315,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 90.7190170288086,
|
| 284 |
+
"learning_rate": 6.980111184706833e-05,
|
| 285 |
+
"log_odds_chosen": 0.2888467609882355,
|
| 286 |
+
"log_odds_ratio": -0.6715133190155029,
|
| 287 |
+
"logits/chosen": 204.8794403076172,
|
| 288 |
+
"logits/rejected": 193.88922119140625,
|
| 289 |
+
"logps/chosen": -1.177809238433838,
|
| 290 |
+
"logps/rejected": -1.3896067142486572,
|
| 291 |
+
"loss": 56.0751,
|
| 292 |
+
"nll_loss": 1.4720895290374756,
|
| 293 |
+
"rewards/accuracies": 0.612500011920929,
|
| 294 |
+
"rewards/chosen": -0.588904619216919,
|
| 295 |
+
"rewards/margins": 0.10589867830276489,
|
| 296 |
+
"rewards/rejected": -0.6948033571243286,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 69.81222534179688,
|
| 302 |
+
"learning_rate": 6.869455780884331e-05,
|
| 303 |
+
"log_odds_chosen": 0.23772254586219788,
|
| 304 |
+
"log_odds_ratio": -0.6677379608154297,
|
| 305 |
+
"logits/chosen": 197.03091430664062,
|
| 306 |
+
"logits/rejected": 172.4070587158203,
|
| 307 |
+
"logps/chosen": -1.120238184928894,
|
| 308 |
+
"logps/rejected": -1.296899437904358,
|
| 309 |
+
"loss": 55.6046,
|
| 310 |
+
"nll_loss": 1.4089494943618774,
|
| 311 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 312 |
+
"rewards/chosen": -0.560119092464447,
|
| 313 |
+
"rewards/margins": 0.08833061158657074,
|
| 314 |
+
"rewards/rejected": -0.648449718952179,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 95.9283676147461,
|
| 320 |
+
"learning_rate": 6.749192350054026e-05,
|
| 321 |
+
"log_odds_chosen": 0.31241756677627563,
|
| 322 |
+
"log_odds_ratio": -0.6360076665878296,
|
| 323 |
+
"logits/chosen": 196.90533447265625,
|
| 324 |
+
"logits/rejected": 185.89779663085938,
|
| 325 |
+
"logps/chosen": -1.0959351062774658,
|
| 326 |
+
"logps/rejected": -1.3153069019317627,
|
| 327 |
+
"loss": 57.8707,
|
| 328 |
+
"nll_loss": 1.4560327529907227,
|
| 329 |
+
"rewards/accuracies": 0.637499988079071,
|
| 330 |
+
"rewards/chosen": -0.5479675531387329,
|
| 331 |
+
"rewards/margins": 0.10968589782714844,
|
| 332 |
+
"rewards/rejected": -0.6576534509658813,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 71.1992416381836,
|
| 338 |
+
"learning_rate": 6.619691307561615e-05,
|
| 339 |
+
"log_odds_chosen": 0.406308114528656,
|
| 340 |
+
"log_odds_ratio": -0.5838354825973511,
|
| 341 |
+
"logits/chosen": 188.7007293701172,
|
| 342 |
+
"logits/rejected": 175.3910369873047,
|
| 343 |
+
"logps/chosen": -1.0477182865142822,
|
| 344 |
+
"logps/rejected": -1.3336520195007324,
|
| 345 |
+
"loss": 55.6848,
|
| 346 |
+
"nll_loss": 1.3589107990264893,
|
| 347 |
+
"rewards/accuracies": 0.6875,
|
| 348 |
+
"rewards/chosen": -0.5238591432571411,
|
| 349 |
+
"rewards/margins": 0.1429668366909027,
|
| 350 |
+
"rewards/rejected": -0.6668260097503662,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 104.38742065429688,
|
| 356 |
+
"learning_rate": 6.48135152090233e-05,
|
| 357 |
+
"log_odds_chosen": 0.1892620474100113,
|
| 358 |
+
"log_odds_ratio": -0.7005770206451416,
|
| 359 |
+
"logits/chosen": 185.79942321777344,
|
| 360 |
+
"logits/rejected": 171.77197265625,
|
| 361 |
+
"logps/chosen": -1.1602965593338013,
|
| 362 |
+
"logps/rejected": -1.3126600980758667,
|
| 363 |
+
"loss": 56.5279,
|
| 364 |
+
"nll_loss": 1.4600017070770264,
|
| 365 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 366 |
+
"rewards/chosen": -0.5801482796669006,
|
| 367 |
+
"rewards/margins": 0.07618186622858047,
|
| 368 |
+
"rewards/rejected": -0.6563300490379333,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 231.78350830078125,
|
| 374 |
+
"learning_rate": 6.33459908119585e-05,
|
| 375 |
+
"log_odds_chosen": 0.29791754484176636,
|
| 376 |
+
"log_odds_ratio": -0.6211925745010376,
|
| 377 |
+
"logits/chosen": 185.2686004638672,
|
| 378 |
+
"logits/rejected": 171.43771362304688,
|
| 379 |
+
"logps/chosen": -1.0696375370025635,
|
| 380 |
+
"logps/rejected": -1.2886693477630615,
|
| 381 |
+
"loss": 54.4392,
|
| 382 |
+
"nll_loss": 1.3739655017852783,
|
| 383 |
+
"rewards/accuracies": 0.668749988079071,
|
| 384 |
+
"rewards/chosen": -0.5348187685012817,
|
| 385 |
+
"rewards/margins": 0.10951590538024902,
|
| 386 |
+
"rewards/rejected": -0.6443346738815308,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.037925925925926,
|
| 391 |
+
"grad_norm": 86.91617584228516,
|
| 392 |
+
"learning_rate": 6.179885990811568e-05,
|
| 393 |
+
"log_odds_chosen": 0.8971813917160034,
|
| 394 |
+
"log_odds_ratio": -0.496082067489624,
|
| 395 |
+
"logits/chosen": 178.14752197265625,
|
| 396 |
+
"logits/rejected": 169.9856414794922,
|
| 397 |
+
"logps/chosen": -0.9173161387443542,
|
| 398 |
+
"logps/rejected": -1.5359400510787964,
|
| 399 |
+
"loss": 41.2989,
|
| 400 |
+
"nll_loss": 1.1888840198516846,
|
| 401 |
+
"rewards/accuracies": 0.811188817024231,
|
| 402 |
+
"rewards/chosen": -0.4586580693721771,
|
| 403 |
+
"rewards/margins": 0.30931198596954346,
|
| 404 |
+
"rewards/rejected": -0.7679700255393982,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0853333333333333,
|
| 409 |
+
"grad_norm": 85.66360473632812,
|
| 410 |
+
"learning_rate": 6.0176887711863254e-05,
|
| 411 |
+
"log_odds_chosen": 1.2501261234283447,
|
| 412 |
+
"log_odds_ratio": -0.3368678689002991,
|
| 413 |
+
"logits/chosen": 172.72744750976562,
|
| 414 |
+
"logits/rejected": 157.48776245117188,
|
| 415 |
+
"logps/chosen": -0.82439124584198,
|
| 416 |
+
"logps/rejected": -1.6705620288848877,
|
| 417 |
+
"loss": 42.4544,
|
| 418 |
+
"nll_loss": 1.1247361898422241,
|
| 419 |
+
"rewards/accuracies": 0.90625,
|
| 420 |
+
"rewards/chosen": -0.41219562292099,
|
| 421 |
+
"rewards/margins": 0.42308536171913147,
|
| 422 |
+
"rewards/rejected": -0.8352810144424438,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1327407407407408,
|
| 427 |
+
"grad_norm": 58.510986328125,
|
| 428 |
+
"learning_rate": 5.848506995122624e-05,
|
| 429 |
+
"log_odds_chosen": 1.2685009241104126,
|
| 430 |
+
"log_odds_ratio": -0.3527768850326538,
|
| 431 |
+
"logits/chosen": 159.47116088867188,
|
| 432 |
+
"logits/rejected": 157.1888885498047,
|
| 433 |
+
"logps/chosen": -0.8306115865707397,
|
| 434 |
+
"logps/rejected": -1.68930983543396,
|
| 435 |
+
"loss": 42.8811,
|
| 436 |
+
"nll_loss": 1.1527321338653564,
|
| 437 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 438 |
+
"rewards/chosen": -0.4153057932853699,
|
| 439 |
+
"rewards/margins": 0.4293490946292877,
|
| 440 |
+
"rewards/rejected": -0.84465491771698,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1801481481481482,
|
| 445 |
+
"grad_norm": 80.0993881225586,
|
| 446 |
+
"learning_rate": 5.672861748087879e-05,
|
| 447 |
+
"log_odds_chosen": 1.1959865093231201,
|
| 448 |
+
"log_odds_ratio": -0.37415602803230286,
|
| 449 |
+
"logits/chosen": 163.66038513183594,
|
| 450 |
+
"logits/rejected": 159.62203979492188,
|
| 451 |
+
"logps/chosen": -0.8715580701828003,
|
| 452 |
+
"logps/rejected": -1.7014503479003906,
|
| 453 |
+
"loss": 41.053,
|
| 454 |
+
"nll_loss": 1.1238315105438232,
|
| 455 |
+
"rewards/accuracies": 0.831250011920929,
|
| 456 |
+
"rewards/chosen": -0.43577903509140015,
|
| 457 |
+
"rewards/margins": 0.41494613885879517,
|
| 458 |
+
"rewards/rejected": -0.8507251739501953,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2275555555555555,
|
| 463 |
+
"grad_norm": 82.62030792236328,
|
| 464 |
+
"learning_rate": 5.491294023253923e-05,
|
| 465 |
+
"log_odds_chosen": 1.1291576623916626,
|
| 466 |
+
"log_odds_ratio": -0.3804819583892822,
|
| 467 |
+
"logits/chosen": 178.26722717285156,
|
| 468 |
+
"logits/rejected": 170.98712158203125,
|
| 469 |
+
"logps/chosen": -0.8340187072753906,
|
| 470 |
+
"logps/rejected": -1.5631098747253418,
|
| 471 |
+
"loss": 41.0951,
|
| 472 |
+
"nll_loss": 1.1230322122573853,
|
| 473 |
+
"rewards/accuracies": 0.862500011920929,
|
| 474 |
+
"rewards/chosen": -0.4170093536376953,
|
| 475 |
+
"rewards/margins": 0.3645455539226532,
|
| 476 |
+
"rewards/rejected": -0.7815549373626709,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.274962962962963,
|
| 481 |
+
"grad_norm": 64.6174087524414,
|
| 482 |
+
"learning_rate": 5.304363055220153e-05,
|
| 483 |
+
"log_odds_chosen": 1.1577521562576294,
|
| 484 |
+
"log_odds_ratio": -0.3698500394821167,
|
| 485 |
+
"logits/chosen": 176.21475219726562,
|
| 486 |
+
"logits/rejected": 174.70472717285156,
|
| 487 |
+
"logps/chosen": -0.7851073741912842,
|
| 488 |
+
"logps/rejected": -1.5308529138565063,
|
| 489 |
+
"loss": 41.1429,
|
| 490 |
+
"nll_loss": 1.0836269855499268,
|
| 491 |
+
"rewards/accuracies": 0.875,
|
| 492 |
+
"rewards/chosen": -0.3925536870956421,
|
| 493 |
+
"rewards/margins": 0.37287265062332153,
|
| 494 |
+
"rewards/rejected": -0.7654264569282532,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3223703703703704,
|
| 499 |
+
"grad_norm": 55.24971389770508,
|
| 500 |
+
"learning_rate": 5.112644597552452e-05,
|
| 501 |
+
"log_odds_chosen": 1.2705775499343872,
|
| 502 |
+
"log_odds_ratio": -0.3516805171966553,
|
| 503 |
+
"logits/chosen": 158.27615356445312,
|
| 504 |
+
"logits/rejected": 164.78353881835938,
|
| 505 |
+
"logps/chosen": -0.8647182583808899,
|
| 506 |
+
"logps/rejected": -1.7130002975463867,
|
| 507 |
+
"loss": 42.35,
|
| 508 |
+
"nll_loss": 1.1466747522354126,
|
| 509 |
+
"rewards/accuracies": 0.856249988079071,
|
| 510 |
+
"rewards/chosen": -0.43235912919044495,
|
| 511 |
+
"rewards/margins": 0.4241410195827484,
|
| 512 |
+
"rewards/rejected": -0.8565001487731934,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.3697777777777778,
|
| 517 |
+
"grad_norm": 62.0118408203125,
|
| 518 |
+
"learning_rate": 4.9167291494431645e-05,
|
| 519 |
+
"log_odds_chosen": 1.2424061298370361,
|
| 520 |
+
"log_odds_ratio": -0.3542417883872986,
|
| 521 |
+
"logits/chosen": 168.1389617919922,
|
| 522 |
+
"logits/rejected": 153.880859375,
|
| 523 |
+
"logps/chosen": -0.8555068969726562,
|
| 524 |
+
"logps/rejected": -1.7085926532745361,
|
| 525 |
+
"loss": 41.8085,
|
| 526 |
+
"nll_loss": 1.1373878717422485,
|
| 527 |
+
"rewards/accuracies": 0.875,
|
| 528 |
+
"rewards/chosen": -0.4277534484863281,
|
| 529 |
+
"rewards/margins": 0.4265429973602295,
|
| 530 |
+
"rewards/rejected": -0.8542963266372681,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.417185185185185,
|
| 535 |
+
"grad_norm": 61.60399627685547,
|
| 536 |
+
"learning_rate": 4.717220136954046e-05,
|
| 537 |
+
"log_odds_chosen": 1.2572475671768188,
|
| 538 |
+
"log_odds_ratio": -0.341666042804718,
|
| 539 |
+
"logits/chosen": 162.69091796875,
|
| 540 |
+
"logits/rejected": 168.27194213867188,
|
| 541 |
+
"logps/chosen": -0.7956852316856384,
|
| 542 |
+
"logps/rejected": -1.6127551794052124,
|
| 543 |
+
"loss": 40.5585,
|
| 544 |
+
"nll_loss": 1.0824170112609863,
|
| 545 |
+
"rewards/accuracies": 0.893750011920929,
|
| 546 |
+
"rewards/chosen": -0.3978426158428192,
|
| 547 |
+
"rewards/margins": 0.4085349142551422,
|
| 548 |
+
"rewards/rejected": -0.8063775897026062,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4645925925925927,
|
| 553 |
+
"grad_norm": 57.484771728515625,
|
| 554 |
+
"learning_rate": 4.514732054444031e-05,
|
| 555 |
+
"log_odds_chosen": 1.2566419839859009,
|
| 556 |
+
"log_odds_ratio": -0.37278324365615845,
|
| 557 |
+
"logits/chosen": 171.4698944091797,
|
| 558 |
+
"logits/rejected": 158.44186401367188,
|
| 559 |
+
"logps/chosen": -0.8029295802116394,
|
| 560 |
+
"logps/rejected": -1.641107201576233,
|
| 561 |
+
"loss": 41.6586,
|
| 562 |
+
"nll_loss": 1.1319265365600586,
|
| 563 |
+
"rewards/accuracies": 0.84375,
|
| 564 |
+
"rewards/chosen": -0.4014647901058197,
|
| 565 |
+
"rewards/margins": 0.41908884048461914,
|
| 566 |
+
"rewards/rejected": -0.8205536007881165,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.512,
|
| 571 |
+
"grad_norm": 55.12516784667969,
|
| 572 |
+
"learning_rate": 4.309888571906274e-05,
|
| 573 |
+
"log_odds_chosen": 1.149753451347351,
|
| 574 |
+
"log_odds_ratio": -0.3672695755958557,
|
| 575 |
+
"logits/chosen": 164.41958618164062,
|
| 576 |
+
"logits/rejected": 164.40335083007812,
|
| 577 |
+
"logps/chosen": -0.9021244049072266,
|
| 578 |
+
"logps/rejected": -1.6924254894256592,
|
| 579 |
+
"loss": 42.0332,
|
| 580 |
+
"nll_loss": 1.1726073026657104,
|
| 581 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 582 |
+
"rewards/chosen": -0.4510622024536133,
|
| 583 |
+
"rewards/margins": 0.3951506018638611,
|
| 584 |
+
"rewards/rejected": -0.8462127447128296,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5594074074074074,
|
| 589 |
+
"grad_norm": 46.9222297668457,
|
| 590 |
+
"learning_rate": 4.1033206140439427e-05,
|
| 591 |
+
"log_odds_chosen": 1.319567084312439,
|
| 592 |
+
"log_odds_ratio": -0.32131245732307434,
|
| 593 |
+
"logits/chosen": 151.97567749023438,
|
| 594 |
+
"logits/rejected": 166.11029052734375,
|
| 595 |
+
"logps/chosen": -0.8229783773422241,
|
| 596 |
+
"logps/rejected": -1.708121657371521,
|
| 597 |
+
"loss": 41.3777,
|
| 598 |
+
"nll_loss": 1.1105647087097168,
|
| 599 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 600 |
+
"rewards/chosen": -0.41148918867111206,
|
| 601 |
+
"rewards/margins": 0.4425717294216156,
|
| 602 |
+
"rewards/rejected": -0.8540608286857605,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6068148148148147,
|
| 607 |
+
"grad_norm": 75.31800842285156,
|
| 608 |
+
"learning_rate": 3.8956644170012536e-05,
|
| 609 |
+
"log_odds_chosen": 1.4432159662246704,
|
| 610 |
+
"log_odds_ratio": -0.33049893379211426,
|
| 611 |
+
"logits/chosen": 156.82803344726562,
|
| 612 |
+
"logits/rejected": 155.05007934570312,
|
| 613 |
+
"logps/chosen": -0.8102866411209106,
|
| 614 |
+
"logps/rejected": -1.783372163772583,
|
| 615 |
+
"loss": 40.0255,
|
| 616 |
+
"nll_loss": 1.0867273807525635,
|
| 617 |
+
"rewards/accuracies": 0.893750011920929,
|
| 618 |
+
"rewards/chosen": -0.4051433205604553,
|
| 619 |
+
"rewards/margins": 0.48654279112815857,
|
| 620 |
+
"rewards/rejected": -0.8916860818862915,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6542222222222223,
|
| 625 |
+
"grad_norm": 68.79954528808594,
|
| 626 |
+
"learning_rate": 3.687559568735105e-05,
|
| 627 |
+
"log_odds_chosen": 1.266623854637146,
|
| 628 |
+
"log_odds_ratio": -0.34371012449264526,
|
| 629 |
+
"logits/chosen": 159.88473510742188,
|
| 630 |
+
"logits/rejected": 147.89791870117188,
|
| 631 |
+
"logps/chosen": -0.8142267465591431,
|
| 632 |
+
"logps/rejected": -1.6857149600982666,
|
| 633 |
+
"loss": 40.7363,
|
| 634 |
+
"nll_loss": 1.139469027519226,
|
| 635 |
+
"rewards/accuracies": 0.875,
|
| 636 |
+
"rewards/chosen": -0.40711337327957153,
|
| 637 |
+
"rewards/margins": 0.4357439875602722,
|
| 638 |
+
"rewards/rejected": -0.8428574800491333,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7016296296296296,
|
| 643 |
+
"grad_norm": 72.72908020019531,
|
| 644 |
+
"learning_rate": 3.479647039063004e-05,
|
| 645 |
+
"log_odds_chosen": 1.2162840366363525,
|
| 646 |
+
"log_odds_ratio": -0.3732939064502716,
|
| 647 |
+
"logits/chosen": 149.99951171875,
|
| 648 |
+
"logits/rejected": 149.6708221435547,
|
| 649 |
+
"logps/chosen": -0.8462786674499512,
|
| 650 |
+
"logps/rejected": -1.6657482385635376,
|
| 651 |
+
"loss": 41.3565,
|
| 652 |
+
"nll_loss": 1.1392968893051147,
|
| 653 |
+
"rewards/accuracies": 0.862500011920929,
|
| 654 |
+
"rewards/chosen": -0.4231393337249756,
|
| 655 |
+
"rewards/margins": 0.4097347855567932,
|
| 656 |
+
"rewards/rejected": -0.8328741192817688,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.749037037037037,
|
| 661 |
+
"grad_norm": 54.563720703125,
|
| 662 |
+
"learning_rate": 3.272567205454826e-05,
|
| 663 |
+
"log_odds_chosen": 1.255048155784607,
|
| 664 |
+
"log_odds_ratio": -0.35441821813583374,
|
| 665 |
+
"logits/chosen": 153.781982421875,
|
| 666 |
+
"logits/rejected": 151.72683715820312,
|
| 667 |
+
"logps/chosen": -0.8422285318374634,
|
| 668 |
+
"logps/rejected": -1.7023115158081055,
|
| 669 |
+
"loss": 41.0615,
|
| 670 |
+
"nll_loss": 1.1301571130752563,
|
| 671 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 672 |
+
"rewards/chosen": -0.4211142659187317,
|
| 673 |
+
"rewards/margins": 0.4300415515899658,
|
| 674 |
+
"rewards/rejected": -0.8511557579040527,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.7964444444444445,
|
| 679 |
+
"grad_norm": 68.65567016601562,
|
| 680 |
+
"learning_rate": 3.06695788064902e-05,
|
| 681 |
+
"log_odds_chosen": 1.3918511867523193,
|
| 682 |
+
"log_odds_ratio": -0.3203474283218384,
|
| 683 |
+
"logits/chosen": 160.94696044921875,
|
| 684 |
+
"logits/rejected": 150.07833862304688,
|
| 685 |
+
"logps/chosen": -0.8545455932617188,
|
| 686 |
+
"logps/rejected": -1.8117406368255615,
|
| 687 |
+
"loss": 42.5014,
|
| 688 |
+
"nll_loss": 1.1190176010131836,
|
| 689 |
+
"rewards/accuracies": 0.918749988079071,
|
| 690 |
+
"rewards/chosen": -0.4272727966308594,
|
| 691 |
+
"rewards/margins": 0.4785975515842438,
|
| 692 |
+
"rewards/rejected": -0.9058703184127808,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8438518518518519,
|
| 697 |
+
"grad_norm": 64.03311920166016,
|
| 698 |
+
"learning_rate": 2.863452348168275e-05,
|
| 699 |
+
"log_odds_chosen": 1.4326978921890259,
|
| 700 |
+
"log_odds_ratio": -0.32069623470306396,
|
| 701 |
+
"logits/chosen": 155.94273376464844,
|
| 702 |
+
"logits/rejected": 141.63304138183594,
|
| 703 |
+
"logps/chosen": -0.8207441568374634,
|
| 704 |
+
"logps/rejected": -1.8126327991485596,
|
| 705 |
+
"loss": 40.4731,
|
| 706 |
+
"nll_loss": 1.1084260940551758,
|
| 707 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 708 |
+
"rewards/chosen": -0.4103720784187317,
|
| 709 |
+
"rewards/margins": 0.4959442615509033,
|
| 710 |
+
"rewards/rejected": -0.9063163995742798,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8912592592592592,
|
| 715 |
+
"grad_norm": 73.11625671386719,
|
| 716 |
+
"learning_rate": 2.6626774117853172e-05,
|
| 717 |
+
"log_odds_chosen": 1.2478724718093872,
|
| 718 |
+
"log_odds_ratio": -0.3590041697025299,
|
| 719 |
+
"logits/chosen": 155.2909698486328,
|
| 720 |
+
"logits/rejected": 149.68272399902344,
|
| 721 |
+
"logps/chosen": -0.8270870447158813,
|
| 722 |
+
"logps/rejected": -1.6576263904571533,
|
| 723 |
+
"loss": 42.3125,
|
| 724 |
+
"nll_loss": 1.0772912502288818,
|
| 725 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 726 |
+
"rewards/chosen": -0.4135435223579407,
|
| 727 |
+
"rewards/margins": 0.415269672870636,
|
| 728 |
+
"rewards/rejected": -0.8288131952285767,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9386666666666668,
|
| 733 |
+
"grad_norm": 51.53399658203125,
|
| 734 |
+
"learning_rate": 2.4652514649465555e-05,
|
| 735 |
+
"log_odds_chosen": 1.105621576309204,
|
| 736 |
+
"log_odds_ratio": -0.3948635756969452,
|
| 737 |
+
"logits/chosen": 163.22994995117188,
|
| 738 |
+
"logits/rejected": 149.87380981445312,
|
| 739 |
+
"logps/chosen": -0.8414310216903687,
|
| 740 |
+
"logps/rejected": -1.586313009262085,
|
| 741 |
+
"loss": 41.2346,
|
| 742 |
+
"nll_loss": 1.1008248329162598,
|
| 743 |
+
"rewards/accuracies": 0.862500011920929,
|
| 744 |
+
"rewards/chosen": -0.4207155108451843,
|
| 745 |
+
"rewards/margins": 0.37244105339050293,
|
| 746 |
+
"rewards/rejected": -0.7931565046310425,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.986074074074074,
|
| 751 |
+
"grad_norm": 592.9937133789062,
|
| 752 |
+
"learning_rate": 2.2717825860998054e-05,
|
| 753 |
+
"log_odds_chosen": 1.1890079975128174,
|
| 754 |
+
"log_odds_ratio": -0.3658009171485901,
|
| 755 |
+
"logits/chosen": 143.8880157470703,
|
| 756 |
+
"logits/rejected": 149.6746063232422,
|
| 757 |
+
"logps/chosen": -0.8134666681289673,
|
| 758 |
+
"logps/rejected": -1.5765702724456787,
|
| 759 |
+
"loss": 39.231,
|
| 760 |
+
"nll_loss": 1.0431172847747803,
|
| 761 |
+
"rewards/accuracies": 0.862500011920929,
|
| 762 |
+
"rewards/chosen": -0.40673333406448364,
|
| 763 |
+
"rewards/margins": 0.38155174255371094,
|
| 764 |
+
"rewards/rejected": -0.7882851362228394,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0284444444444443,
|
| 769 |
+
"grad_norm": 48.64838409423828,
|
| 770 |
+
"learning_rate": 2.082866665792552e-05,
|
| 771 |
+
"log_odds_chosen": 2.0941009521484375,
|
| 772 |
+
"log_odds_ratio": -0.22275511920452118,
|
| 773 |
+
"logits/chosen": 143.320068359375,
|
| 774 |
+
"logits/rejected": 138.30125427246094,
|
| 775 |
+
"logps/chosen": -0.6920493245124817,
|
| 776 |
+
"logps/rejected": -2.0684661865234375,
|
| 777 |
+
"loss": 30.0118,
|
| 778 |
+
"nll_loss": 0.9490633606910706,
|
| 779 |
+
"rewards/accuracies": 0.9370629191398621,
|
| 780 |
+
"rewards/chosen": -0.34602466225624084,
|
| 781 |
+
"rewards/margins": 0.6882083415985107,
|
| 782 |
+
"rewards/rejected": -1.0342330932617188,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.075851851851852,
|
| 787 |
+
"grad_norm": 77.73394012451172,
|
| 788 |
+
"learning_rate": 1.8990855713093523e-05,
|
| 789 |
+
"log_odds_chosen": 3.1007094383239746,
|
| 790 |
+
"log_odds_ratio": -0.10504193603992462,
|
| 791 |
+
"logits/chosen": 118.2657241821289,
|
| 792 |
+
"logits/rejected": 116.21284484863281,
|
| 793 |
+
"logps/chosen": -0.5571001172065735,
|
| 794 |
+
"logps/rejected": -2.6969428062438965,
|
| 795 |
+
"loss": 28.4152,
|
| 796 |
+
"nll_loss": 0.8508051633834839,
|
| 797 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 798 |
+
"rewards/chosen": -0.27855005860328674,
|
| 799 |
+
"rewards/margins": 1.0699213743209839,
|
| 800 |
+
"rewards/rejected": -1.3484714031219482,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1232592592592594,
|
| 805 |
+
"grad_norm": 50.05561447143555,
|
| 806 |
+
"learning_rate": 1.7210053545013458e-05,
|
| 807 |
+
"log_odds_chosen": 2.909291982650757,
|
| 808 |
+
"log_odds_ratio": -0.11771439015865326,
|
| 809 |
+
"logits/chosen": 133.0370330810547,
|
| 810 |
+
"logits/rejected": 126.50312805175781,
|
| 811 |
+
"logps/chosen": -0.565724790096283,
|
| 812 |
+
"logps/rejected": -2.530423641204834,
|
| 813 |
+
"loss": 29.3042,
|
| 814 |
+
"nll_loss": 0.8670442700386047,
|
| 815 |
+
"rewards/accuracies": 0.96875,
|
| 816 |
+
"rewards/chosen": -0.2828623950481415,
|
| 817 |
+
"rewards/margins": 0.9823493957519531,
|
| 818 |
+
"rewards/rejected": -1.265211820602417,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.1706666666666665,
|
| 823 |
+
"grad_norm": 54.2590446472168,
|
| 824 |
+
"learning_rate": 1.549174508327826e-05,
|
| 825 |
+
"log_odds_chosen": 3.056389808654785,
|
| 826 |
+
"log_odds_ratio": -0.09899865090847015,
|
| 827 |
+
"logits/chosen": 129.60385131835938,
|
| 828 |
+
"logits/rejected": 120.37672424316406,
|
| 829 |
+
"logps/chosen": -0.5788933038711548,
|
| 830 |
+
"logps/rejected": -2.6959786415100098,
|
| 831 |
+
"loss": 27.7201,
|
| 832 |
+
"nll_loss": 0.8015221357345581,
|
| 833 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 834 |
+
"rewards/chosen": -0.2894466519355774,
|
| 835 |
+
"rewards/margins": 1.0585426092147827,
|
| 836 |
+
"rewards/rejected": -1.3479893207550049,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.218074074074074,
|
| 841 |
+
"grad_norm": 53.754486083984375,
|
| 842 |
+
"learning_rate": 1.3841222774797851e-05,
|
| 843 |
+
"log_odds_chosen": 2.7191035747528076,
|
| 844 |
+
"log_odds_ratio": -0.11840663105249405,
|
| 845 |
+
"logits/chosen": 128.04299926757812,
|
| 846 |
+
"logits/rejected": 123.1611557006836,
|
| 847 |
+
"logps/chosen": -0.5877354741096497,
|
| 848 |
+
"logps/rejected": -2.4053232669830322,
|
| 849 |
+
"loss": 28.9243,
|
| 850 |
+
"nll_loss": 0.8488882184028625,
|
| 851 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 852 |
+
"rewards/chosen": -0.29386773705482483,
|
| 853 |
+
"rewards/margins": 0.9087940454483032,
|
| 854 |
+
"rewards/rejected": -1.2026616334915161,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2654814814814817,
|
| 859 |
+
"grad_norm": 48.55769729614258,
|
| 860 |
+
"learning_rate": 1.2263570282887465e-05,
|
| 861 |
+
"log_odds_chosen": 2.9994494915008545,
|
| 862 |
+
"log_odds_ratio": -0.1045825332403183,
|
| 863 |
+
"logits/chosen": 124.81593322753906,
|
| 864 |
+
"logits/rejected": 124.41886901855469,
|
| 865 |
+
"logps/chosen": -0.5564694404602051,
|
| 866 |
+
"logps/rejected": -2.5893123149871826,
|
| 867 |
+
"loss": 28.7714,
|
| 868 |
+
"nll_loss": 0.8480992317199707,
|
| 869 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 870 |
+
"rewards/chosen": -0.27823472023010254,
|
| 871 |
+
"rewards/margins": 1.0164214372634888,
|
| 872 |
+
"rewards/rejected": -1.2946561574935913,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.3128888888888888,
|
| 877 |
+
"grad_norm": 54.71540069580078,
|
| 878 |
+
"learning_rate": 1.0763646829416548e-05,
|
| 879 |
+
"log_odds_chosen": 3.0860180854797363,
|
| 880 |
+
"log_odds_ratio": -0.09014725685119629,
|
| 881 |
+
"logits/chosen": 127.33006286621094,
|
| 882 |
+
"logits/rejected": 117.83717346191406,
|
| 883 |
+
"logps/chosen": -0.5473681688308716,
|
| 884 |
+
"logps/rejected": -2.653287649154663,
|
| 885 |
+
"loss": 27.0092,
|
| 886 |
+
"nll_loss": 0.82794588804245,
|
| 887 |
+
"rewards/accuracies": 1.0,
|
| 888 |
+
"rewards/chosen": -0.2736840844154358,
|
| 889 |
+
"rewards/margins": 1.052959680557251,
|
| 890 |
+
"rewards/rejected": -1.3266438245773315,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3602962962962963,
|
| 895 |
+
"grad_norm": 144.38958740234375,
|
| 896 |
+
"learning_rate": 9.346072228244598e-06,
|
| 897 |
+
"log_odds_chosen": 3.072089672088623,
|
| 898 |
+
"log_odds_ratio": -0.10084352642297745,
|
| 899 |
+
"logits/chosen": 128.05050659179688,
|
| 900 |
+
"logits/rejected": 127.39276123046875,
|
| 901 |
+
"logps/chosen": -0.5662072896957397,
|
| 902 |
+
"logps/rejected": -2.6561198234558105,
|
| 903 |
+
"loss": 27.741,
|
| 904 |
+
"nll_loss": 0.8294836282730103,
|
| 905 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 906 |
+
"rewards/chosen": -0.2831036448478699,
|
| 907 |
+
"rewards/margins": 1.044956088066101,
|
| 908 |
+
"rewards/rejected": -1.3280599117279053,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.407703703703704,
|
| 913 |
+
"grad_norm": 69.65536499023438,
|
| 914 |
+
"learning_rate": 8.015212656041612e-06,
|
| 915 |
+
"log_odds_chosen": 3.221827268600464,
|
| 916 |
+
"log_odds_ratio": -0.090447798371315,
|
| 917 |
+
"logits/chosen": 123.1548843383789,
|
| 918 |
+
"logits/rejected": 122.70021057128906,
|
| 919 |
+
"logps/chosen": -0.49670666456222534,
|
| 920 |
+
"logps/rejected": -2.6964945793151855,
|
| 921 |
+
"loss": 27.5962,
|
| 922 |
+
"nll_loss": 0.7993031740188599,
|
| 923 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 924 |
+
"rewards/chosen": -0.24835333228111267,
|
| 925 |
+
"rewards/margins": 1.0998938083648682,
|
| 926 |
+
"rewards/rejected": -1.3482472896575928,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.455111111111111,
|
| 931 |
+
"grad_norm": 52.56804275512695,
|
| 932 |
+
"learning_rate": 6.775167204319555e-06,
|
| 933 |
+
"log_odds_chosen": 3.3616344928741455,
|
| 934 |
+
"log_odds_ratio": -0.08646370470523834,
|
| 935 |
+
"logits/chosen": 127.23355865478516,
|
| 936 |
+
"logits/rejected": 119.65962219238281,
|
| 937 |
+
"logps/chosen": -0.5302814841270447,
|
| 938 |
+
"logps/rejected": -2.8641209602355957,
|
| 939 |
+
"loss": 27.3852,
|
| 940 |
+
"nll_loss": 0.8003301620483398,
|
| 941 |
+
"rewards/accuracies": 1.0,
|
| 942 |
+
"rewards/chosen": -0.26514074206352234,
|
| 943 |
+
"rewards/margins": 1.1669199466705322,
|
| 944 |
+
"rewards/rejected": -1.4320604801177979,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5025185185185186,
|
| 949 |
+
"grad_norm": 52.00509262084961,
|
| 950 |
+
"learning_rate": 5.6297552540948325e-06,
|
| 951 |
+
"log_odds_chosen": 2.9919915199279785,
|
| 952 |
+
"log_odds_ratio": -0.13428373634815216,
|
| 953 |
+
"logits/chosen": 129.68038940429688,
|
| 954 |
+
"logits/rejected": 126.67582702636719,
|
| 955 |
+
"logps/chosen": -0.5471857190132141,
|
| 956 |
+
"logps/rejected": -2.557234525680542,
|
| 957 |
+
"loss": 27.4382,
|
| 958 |
+
"nll_loss": 0.7889086008071899,
|
| 959 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 960 |
+
"rewards/chosen": -0.27359285950660706,
|
| 961 |
+
"rewards/margins": 1.0050245523452759,
|
| 962 |
+
"rewards/rejected": -1.278617262840271,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.549925925925926,
|
| 967 |
+
"grad_norm": 48.82490539550781,
|
| 968 |
+
"learning_rate": 4.5825047120684215e-06,
|
| 969 |
+
"log_odds_chosen": 3.1277198791503906,
|
| 970 |
+
"log_odds_ratio": -0.11080478131771088,
|
| 971 |
+
"logits/chosen": 129.29345703125,
|
| 972 |
+
"logits/rejected": 128.2639617919922,
|
| 973 |
+
"logps/chosen": -0.5835883021354675,
|
| 974 |
+
"logps/rejected": -2.7072691917419434,
|
| 975 |
+
"loss": 26.8792,
|
| 976 |
+
"nll_loss": 0.8251482844352722,
|
| 977 |
+
"rewards/accuracies": 0.987500011920929,
|
| 978 |
+
"rewards/chosen": -0.29179415106773376,
|
| 979 |
+
"rewards/margins": 1.0618404150009155,
|
| 980 |
+
"rewards/rejected": -1.3536345958709717,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.5973333333333333,
|
| 985 |
+
"grad_norm": 44.77346420288086,
|
| 986 |
+
"learning_rate": 3.636641144556547e-06,
|
| 987 |
+
"log_odds_chosen": 3.256783962249756,
|
| 988 |
+
"log_odds_ratio": -0.10093816369771957,
|
| 989 |
+
"logits/chosen": 130.92579650878906,
|
| 990 |
+
"logits/rejected": 121.23966217041016,
|
| 991 |
+
"logps/chosen": -0.5694173574447632,
|
| 992 |
+
"logps/rejected": -2.807206630706787,
|
| 993 |
+
"loss": 28.3661,
|
| 994 |
+
"nll_loss": 0.8377152681350708,
|
| 995 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 996 |
+
"rewards/chosen": -0.2847086787223816,
|
| 997 |
+
"rewards/margins": 1.1188945770263672,
|
| 998 |
+
"rewards/rejected": -1.4036033153533936,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.644740740740741,
|
| 1003 |
+
"grad_norm": 43.979530334472656,
|
| 1004 |
+
"learning_rate": 2.795077842639676e-06,
|
| 1005 |
+
"log_odds_chosen": 3.014188766479492,
|
| 1006 |
+
"log_odds_ratio": -0.10471322387456894,
|
| 1007 |
+
"logits/chosen": 125.88334655761719,
|
| 1008 |
+
"logits/rejected": 120.0384292602539,
|
| 1009 |
+
"logps/chosen": -0.5433251857757568,
|
| 1010 |
+
"logps/rejected": -2.5689566135406494,
|
| 1011 |
+
"loss": 27.0828,
|
| 1012 |
+
"nll_loss": 0.7862499952316284,
|
| 1013 |
+
"rewards/accuracies": 1.0,
|
| 1014 |
+
"rewards/chosen": -0.2716625928878784,
|
| 1015 |
+
"rewards/margins": 1.0128155946731567,
|
| 1016 |
+
"rewards/rejected": -1.2844783067703247,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.6921481481481484,
|
| 1021 |
+
"grad_norm": 50.671897888183594,
|
| 1022 |
+
"learning_rate": 2.060406849129605e-06,
|
| 1023 |
+
"log_odds_chosen": 3.0842363834381104,
|
| 1024 |
+
"log_odds_ratio": -0.10635282844305038,
|
| 1025 |
+
"logits/chosen": 134.56607055664062,
|
| 1026 |
+
"logits/rejected": 120.92826843261719,
|
| 1027 |
+
"logps/chosen": -0.5578547716140747,
|
| 1028 |
+
"logps/rejected": -2.670154333114624,
|
| 1029 |
+
"loss": 27.4779,
|
| 1030 |
+
"nll_loss": 0.8171490430831909,
|
| 1031 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1032 |
+
"rewards/chosen": -0.27892738580703735,
|
| 1033 |
+
"rewards/margins": 1.056149959564209,
|
| 1034 |
+
"rewards/rejected": -1.335077166557312,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.7395555555555555,
|
| 1039 |
+
"grad_norm": 46.54133224487305,
|
| 1040 |
+
"learning_rate": 1.4348909749918786e-06,
|
| 1041 |
+
"log_odds_chosen": 3.2423605918884277,
|
| 1042 |
+
"log_odds_ratio": -0.08490051329135895,
|
| 1043 |
+
"logits/chosen": 128.3448028564453,
|
| 1044 |
+
"logits/rejected": 129.498779296875,
|
| 1045 |
+
"logps/chosen": -0.5114618539810181,
|
| 1046 |
+
"logps/rejected": -2.7147812843322754,
|
| 1047 |
+
"loss": 28.442,
|
| 1048 |
+
"nll_loss": 0.8227435350418091,
|
| 1049 |
+
"rewards/accuracies": 1.0,
|
| 1050 |
+
"rewards/chosen": -0.25573092699050903,
|
| 1051 |
+
"rewards/margins": 1.101659893989563,
|
| 1052 |
+
"rewards/rejected": -1.3573906421661377,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.786962962962963,
|
| 1057 |
+
"grad_norm": 45.60363006591797,
|
| 1058 |
+
"learning_rate": 9.204568298131307e-07,
|
| 1059 |
+
"log_odds_chosen": 3.0159189701080322,
|
| 1060 |
+
"log_odds_ratio": -0.14175482094287872,
|
| 1061 |
+
"logits/chosen": 130.07974243164062,
|
| 1062 |
+
"logits/rejected": 128.87826538085938,
|
| 1063 |
+
"logps/chosen": -0.6083237528800964,
|
| 1064 |
+
"logps/rejected": -2.6694674491882324,
|
| 1065 |
+
"loss": 28.036,
|
| 1066 |
+
"nll_loss": 0.8367668390274048,
|
| 1067 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1068 |
+
"rewards/chosen": -0.3041618764400482,
|
| 1069 |
+
"rewards/margins": 1.0305720567703247,
|
| 1070 |
+
"rewards/rejected": -1.3347337245941162,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.83437037037037,
|
| 1075 |
+
"grad_norm": 91.35395050048828,
|
| 1076 |
+
"learning_rate": 5.186888877798209e-07,
|
| 1077 |
+
"log_odds_chosen": 3.2012171745300293,
|
| 1078 |
+
"log_odds_ratio": -0.09840454906225204,
|
| 1079 |
+
"logits/chosen": 144.08177185058594,
|
| 1080 |
+
"logits/rejected": 126.72806549072266,
|
| 1081 |
+
"logps/chosen": -0.5671483278274536,
|
| 1082 |
+
"logps/rejected": -2.7830772399902344,
|
| 1083 |
+
"loss": 28.1208,
|
| 1084 |
+
"nll_loss": 0.8516788482666016,
|
| 1085 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1086 |
+
"rewards/chosen": -0.2835741639137268,
|
| 1087 |
+
"rewards/margins": 1.1079645156860352,
|
| 1088 |
+
"rewards/rejected": -1.3915386199951172,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.8817777777777778,
|
| 1093 |
+
"grad_norm": 73.92967224121094,
|
| 1094 |
+
"learning_rate": 2.308246074453138e-07,
|
| 1095 |
+
"log_odds_chosen": 3.117356061935425,
|
| 1096 |
+
"log_odds_ratio": -0.09485842287540436,
|
| 1097 |
+
"logits/chosen": 132.63426208496094,
|
| 1098 |
+
"logits/rejected": 125.5457534790039,
|
| 1099 |
+
"logps/chosen": -0.5718491077423096,
|
| 1100 |
+
"logps/rejected": -2.6983981132507324,
|
| 1101 |
+
"loss": 27.0973,
|
| 1102 |
+
"nll_loss": 0.7848010659217834,
|
| 1103 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1104 |
+
"rewards/chosen": -0.2859245538711548,
|
| 1105 |
+
"rewards/margins": 1.063274621963501,
|
| 1106 |
+
"rewards/rejected": -1.3491990566253662,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9291851851851853,
|
| 1111 |
+
"grad_norm": 50.79293441772461,
|
| 1112 |
+
"learning_rate": 5.7750620316622534e-08,
|
| 1113 |
+
"log_odds_chosen": 3.230167865753174,
|
| 1114 |
+
"log_odds_ratio": -0.08979301154613495,
|
| 1115 |
+
"logits/chosen": 133.17994689941406,
|
| 1116 |
+
"logits/rejected": 121.85380554199219,
|
| 1117 |
+
"logps/chosen": -0.5334647297859192,
|
| 1118 |
+
"logps/rejected": -2.7135567665100098,
|
| 1119 |
+
"loss": 27.1197,
|
| 1120 |
+
"nll_loss": 0.8048761487007141,
|
| 1121 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1122 |
+
"rewards/chosen": -0.2667323648929596,
|
| 1123 |
+
"rewards/margins": 1.0900460481643677,
|
| 1124 |
+
"rewards/rejected": -1.3567783832550049,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.9765925925925925,
|
| 1129 |
+
"grad_norm": 59.680809020996094,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 3.1277458667755127,
|
| 1132 |
+
"log_odds_ratio": -0.09231346845626831,
|
| 1133 |
+
"logits/chosen": 130.7123260498047,
|
| 1134 |
+
"logits/rejected": 117.61688232421875,
|
| 1135 |
+
"logps/chosen": -0.545311450958252,
|
| 1136 |
+
"logps/rejected": -2.663321018218994,
|
| 1137 |
+
"loss": 26.588,
|
| 1138 |
+
"nll_loss": 0.802075207233429,
|
| 1139 |
+
"rewards/accuracies": 1.0,
|
| 1140 |
+
"rewards/chosen": -0.272655725479126,
|
| 1141 |
+
"rewards/margins": 1.059004783630371,
|
| 1142 |
+
"rewards/rejected": -1.331660509109497,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.9765925925925925,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 60.23838462224082,
|
| 1150 |
+
"train_runtime": 8115.1309,
|
| 1151 |
+
"train_samples_per_second": 2.495,
|
| 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 |
+
}
|