Model save
Browse files- README.md +67 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1210 -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-shuffled-5e-5-005-v141
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-orpo-shuffled-5e-5-005-v141
|
| 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-shuffled-5e-5-005-v141", 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/ovpis4k9)
|
| 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: 0.15.2
|
| 38 |
+
- Transformers: 4.49.0
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.2.0
|
| 41 |
+
- Tokenizers: 0.21.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite ORPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{hong2024orpo,
|
| 49 |
+
title = {{ORPO: Monolithic Preference Optimization without Reference Model}},
|
| 50 |
+
author = {Jiwoo Hong and Noah Lee and James Thorne},
|
| 51 |
+
year = 2024,
|
| 52 |
+
eprint = {arXiv:2403.07691}
|
| 53 |
+
}
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
Cite TRL as:
|
| 57 |
+
|
| 58 |
+
```bibtex
|
| 59 |
+
@misc{vonwerra2022trl,
|
| 60 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 61 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 62 |
+
year = 2020,
|
| 63 |
+
journal = {GitHub repository},
|
| 64 |
+
publisher = {GitHub},
|
| 65 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 66 |
+
}
|
| 67 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.9595687331536387,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 25.71915608903636,
|
| 5 |
+
"train_runtime": 4429.3261,
|
| 6 |
+
"train_samples": 5936,
|
| 7 |
+
"train_samples_per_second": 5.361,
|
| 8 |
+
"train_steps_per_second": 0.083
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.9595687331536387,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 25.71915608903636,
|
| 5 |
+
"train_runtime": 4429.3261,
|
| 6 |
+
"train_samples": 5936,
|
| 7 |
+
"train_samples_per_second": 5.361,
|
| 8 |
+
"train_steps_per_second": 0.083
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1210 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.9595687331536387,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 368,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05390835579514825,
|
| 13 |
+
"grad_norm": 684.0,
|
| 14 |
+
"learning_rate": 6.7567567567567575e-06,
|
| 15 |
+
"log_odds_chosen": 5.920718193054199,
|
| 16 |
+
"log_odds_ratio": -7.724557399749756,
|
| 17 |
+
"logps/chosen": -24.315807342529297,
|
| 18 |
+
"logps/rejected": -30.238388061523438,
|
| 19 |
+
"loss": 193.2993,
|
| 20 |
+
"nll_loss": 11.694974899291992,
|
| 21 |
+
"rewards/accuracies": 0.484375,
|
| 22 |
+
"rewards/chosen": -1.2157905101776123,
|
| 23 |
+
"rewards/margins": 0.2961289882659912,
|
| 24 |
+
"rewards/rejected": -1.511919379234314,
|
| 25 |
+
"step": 5
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.1078167115902965,
|
| 29 |
+
"grad_norm": 218.0,
|
| 30 |
+
"learning_rate": 1.3513513513513515e-05,
|
| 31 |
+
"log_odds_chosen": 4.100552082061768,
|
| 32 |
+
"log_odds_ratio": -5.447903633117676,
|
| 33 |
+
"logps/chosen": -20.245830535888672,
|
| 34 |
+
"logps/rejected": -24.348098754882812,
|
| 35 |
+
"loss": 161.9053,
|
| 36 |
+
"nll_loss": 9.846681594848633,
|
| 37 |
+
"rewards/accuracies": 0.4781250059604645,
|
| 38 |
+
"rewards/chosen": -1.012291669845581,
|
| 39 |
+
"rewards/margins": 0.2051134556531906,
|
| 40 |
+
"rewards/rejected": -1.2174049615859985,
|
| 41 |
+
"step": 10
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 0.16172506738544473,
|
| 45 |
+
"grad_norm": 205.0,
|
| 46 |
+
"learning_rate": 2.0270270270270273e-05,
|
| 47 |
+
"log_odds_chosen": 3.8738200664520264,
|
| 48 |
+
"log_odds_ratio": -6.855284690856934,
|
| 49 |
+
"logps/chosen": -21.959657669067383,
|
| 50 |
+
"logps/rejected": -25.834619522094727,
|
| 51 |
+
"loss": 166.4944,
|
| 52 |
+
"nll_loss": 10.06313419342041,
|
| 53 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 54 |
+
"rewards/chosen": -1.0979830026626587,
|
| 55 |
+
"rewards/margins": 0.19374798238277435,
|
| 56 |
+
"rewards/rejected": -1.2917309999465942,
|
| 57 |
+
"step": 15
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.215633423180593,
|
| 61 |
+
"grad_norm": 648.0,
|
| 62 |
+
"learning_rate": 2.702702702702703e-05,
|
| 63 |
+
"log_odds_chosen": 0.7052766680717468,
|
| 64 |
+
"log_odds_ratio": -7.368393898010254,
|
| 65 |
+
"logps/chosen": -20.677827835083008,
|
| 66 |
+
"logps/rejected": -21.383352279663086,
|
| 67 |
+
"loss": 141.3241,
|
| 68 |
+
"nll_loss": 8.464337348937988,
|
| 69 |
+
"rewards/accuracies": 0.518750011920929,
|
| 70 |
+
"rewards/chosen": -1.0338914394378662,
|
| 71 |
+
"rewards/margins": 0.03527634218335152,
|
| 72 |
+
"rewards/rejected": -1.0691677331924438,
|
| 73 |
+
"step": 20
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.2695417789757412,
|
| 77 |
+
"grad_norm": 182.0,
|
| 78 |
+
"learning_rate": 3.3783783783783784e-05,
|
| 79 |
+
"log_odds_chosen": -0.0883919969201088,
|
| 80 |
+
"log_odds_ratio": -2.0752124786376953,
|
| 81 |
+
"logps/chosen": -6.3797478675842285,
|
| 82 |
+
"logps/rejected": -6.272956371307373,
|
| 83 |
+
"loss": 62.7556,
|
| 84 |
+
"nll_loss": 3.8184638023376465,
|
| 85 |
+
"rewards/accuracies": 0.559374988079071,
|
| 86 |
+
"rewards/chosen": -0.3189873993396759,
|
| 87 |
+
"rewards/margins": -0.0053395600989460945,
|
| 88 |
+
"rewards/rejected": -0.3136478364467621,
|
| 89 |
+
"step": 25
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.32345013477088946,
|
| 93 |
+
"grad_norm": 83.0,
|
| 94 |
+
"learning_rate": 4.0540540540540545e-05,
|
| 95 |
+
"log_odds_chosen": 0.24213147163391113,
|
| 96 |
+
"log_odds_ratio": -0.7704355716705322,
|
| 97 |
+
"logps/chosen": -1.9613044261932373,
|
| 98 |
+
"logps/rejected": -2.175011157989502,
|
| 99 |
+
"loss": 36.6421,
|
| 100 |
+
"nll_loss": 2.2516086101531982,
|
| 101 |
+
"rewards/accuracies": 0.565625011920929,
|
| 102 |
+
"rewards/chosen": -0.09806521981954575,
|
| 103 |
+
"rewards/margins": 0.010685333982110023,
|
| 104 |
+
"rewards/rejected": -0.10875056684017181,
|
| 105 |
+
"step": 30
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.37735849056603776,
|
| 109 |
+
"grad_norm": 78.0,
|
| 110 |
+
"learning_rate": 4.72972972972973e-05,
|
| 111 |
+
"log_odds_chosen": 0.27958303689956665,
|
| 112 |
+
"log_odds_ratio": -0.7305089831352234,
|
| 113 |
+
"logps/chosen": -1.6892141103744507,
|
| 114 |
+
"logps/rejected": -1.930132269859314,
|
| 115 |
+
"loss": 30.9266,
|
| 116 |
+
"nll_loss": 1.8963871002197266,
|
| 117 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 118 |
+
"rewards/chosen": -0.08446070551872253,
|
| 119 |
+
"rewards/margins": 0.01204589568078518,
|
| 120 |
+
"rewards/rejected": -0.09650660306215286,
|
| 121 |
+
"step": 35
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.431266846361186,
|
| 125 |
+
"grad_norm": 62.5,
|
| 126 |
+
"learning_rate": 4.998986632578596e-05,
|
| 127 |
+
"log_odds_chosen": 0.2721244692802429,
|
| 128 |
+
"log_odds_ratio": -0.7381750345230103,
|
| 129 |
+
"logps/chosen": -1.7667934894561768,
|
| 130 |
+
"logps/rejected": -1.9824241399765015,
|
| 131 |
+
"loss": 31.6068,
|
| 132 |
+
"nll_loss": 1.9385162591934204,
|
| 133 |
+
"rewards/accuracies": 0.606249988079071,
|
| 134 |
+
"rewards/chosen": -0.0883396789431572,
|
| 135 |
+
"rewards/margins": 0.010781526565551758,
|
| 136 |
+
"rewards/rejected": -0.09912119805812836,
|
| 137 |
+
"step": 40
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.48517520215633425,
|
| 141 |
+
"grad_norm": 46.75,
|
| 142 |
+
"learning_rate": 4.992796806510936e-05,
|
| 143 |
+
"log_odds_chosen": 0.1983262300491333,
|
| 144 |
+
"log_odds_ratio": -0.7065704464912415,
|
| 145 |
+
"logps/chosen": -1.4196051359176636,
|
| 146 |
+
"logps/rejected": -1.579620122909546,
|
| 147 |
+
"loss": 28.8934,
|
| 148 |
+
"nll_loss": 1.7705100774765015,
|
| 149 |
+
"rewards/accuracies": 0.59375,
|
| 150 |
+
"rewards/chosen": -0.0709802582859993,
|
| 151 |
+
"rewards/margins": 0.008000737987458706,
|
| 152 |
+
"rewards/rejected": -0.07898099720478058,
|
| 153 |
+
"step": 45
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.5390835579514824,
|
| 157 |
+
"grad_norm": 33.0,
|
| 158 |
+
"learning_rate": 4.980994056736854e-05,
|
| 159 |
+
"log_odds_chosen": 0.1537928730249405,
|
| 160 |
+
"log_odds_ratio": -0.7110344171524048,
|
| 161 |
+
"logps/chosen": -1.3953869342803955,
|
| 162 |
+
"logps/rejected": -1.5007522106170654,
|
| 163 |
+
"loss": 27.7613,
|
| 164 |
+
"nll_loss": 1.699528694152832,
|
| 165 |
+
"rewards/accuracies": 0.612500011920929,
|
| 166 |
+
"rewards/chosen": -0.06976933777332306,
|
| 167 |
+
"rewards/margins": 0.005268273409456015,
|
| 168 |
+
"rewards/rejected": -0.07503761351108551,
|
| 169 |
+
"step": 50
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 0.5929919137466307,
|
| 173 |
+
"grad_norm": 56.0,
|
| 174 |
+
"learning_rate": 4.9636049590020475e-05,
|
| 175 |
+
"log_odds_chosen": 0.2435292750597,
|
| 176 |
+
"log_odds_ratio": -0.6781546473503113,
|
| 177 |
+
"logps/chosen": -1.3146178722381592,
|
| 178 |
+
"logps/rejected": -1.4984357357025146,
|
| 179 |
+
"loss": 26.7408,
|
| 180 |
+
"nll_loss": 1.6373910903930664,
|
| 181 |
+
"rewards/accuracies": 0.6031249761581421,
|
| 182 |
+
"rewards/chosen": -0.06573089212179184,
|
| 183 |
+
"rewards/margins": 0.009190889075398445,
|
| 184 |
+
"rewards/rejected": -0.07492177933454514,
|
| 185 |
+
"step": 55
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.6469002695417789,
|
| 189 |
+
"grad_norm": 65.0,
|
| 190 |
+
"learning_rate": 4.940668667592439e-05,
|
| 191 |
+
"log_odds_chosen": 0.13836432993412018,
|
| 192 |
+
"log_odds_ratio": -0.7060701847076416,
|
| 193 |
+
"logps/chosen": -1.2619632482528687,
|
| 194 |
+
"logps/rejected": -1.3755035400390625,
|
| 195 |
+
"loss": 25.7309,
|
| 196 |
+
"nll_loss": 1.5728803873062134,
|
| 197 |
+
"rewards/accuracies": 0.590624988079071,
|
| 198 |
+
"rewards/chosen": -0.06309816986322403,
|
| 199 |
+
"rewards/margins": 0.005677012261003256,
|
| 200 |
+
"rewards/rejected": -0.06877517700195312,
|
| 201 |
+
"step": 60
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.7008086253369272,
|
| 205 |
+
"grad_norm": 204.0,
|
| 206 |
+
"learning_rate": 4.912236827172148e-05,
|
| 207 |
+
"log_odds_chosen": 0.13251568377017975,
|
| 208 |
+
"log_odds_ratio": -0.7122939229011536,
|
| 209 |
+
"logps/chosen": -1.2452380657196045,
|
| 210 |
+
"logps/rejected": -1.3495495319366455,
|
| 211 |
+
"loss": 25.7811,
|
| 212 |
+
"nll_loss": 1.5757039785385132,
|
| 213 |
+
"rewards/accuracies": 0.5218750238418579,
|
| 214 |
+
"rewards/chosen": -0.062261901795864105,
|
| 215 |
+
"rewards/margins": 0.005215569399297237,
|
| 216 |
+
"rewards/rejected": -0.06747747957706451,
|
| 217 |
+
"step": 65
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 0.7547169811320755,
|
| 221 |
+
"grad_norm": 95.0,
|
| 222 |
+
"learning_rate": 4.878373456497416e-05,
|
| 223 |
+
"log_odds_chosen": 0.16559086740016937,
|
| 224 |
+
"log_odds_ratio": -0.6909226179122925,
|
| 225 |
+
"logps/chosen": -1.2156684398651123,
|
| 226 |
+
"logps/rejected": -1.3327122926712036,
|
| 227 |
+
"loss": 25.469,
|
| 228 |
+
"nll_loss": 1.5572690963745117,
|
| 229 |
+
"rewards/accuracies": 0.546875,
|
| 230 |
+
"rewards/chosen": -0.06078341603279114,
|
| 231 |
+
"rewards/margins": 0.005852191708981991,
|
| 232 |
+
"rewards/rejected": -0.0666356161236763,
|
| 233 |
+
"step": 70
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.8086253369272237,
|
| 237 |
+
"grad_norm": 130.0,
|
| 238 |
+
"learning_rate": 4.839154804268313e-05,
|
| 239 |
+
"log_odds_chosen": 0.24878835678100586,
|
| 240 |
+
"log_odds_ratio": -0.6528117656707764,
|
| 241 |
+
"logps/chosen": -1.171622633934021,
|
| 242 |
+
"logps/rejected": -1.3498855829238892,
|
| 243 |
+
"loss": 25.0146,
|
| 244 |
+
"nll_loss": 1.530771255493164,
|
| 245 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 246 |
+
"rewards/chosen": -0.05858113244175911,
|
| 247 |
+
"rewards/margins": 0.00891314446926117,
|
| 248 |
+
"rewards/rejected": -0.06749428063631058,
|
| 249 |
+
"step": 75
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.862533692722372,
|
| 253 |
+
"grad_norm": 119.5,
|
| 254 |
+
"learning_rate": 4.7946691774428144e-05,
|
| 255 |
+
"log_odds_chosen": 0.2302316427230835,
|
| 256 |
+
"log_odds_ratio": -0.6437969207763672,
|
| 257 |
+
"logps/chosen": -1.1122291088104248,
|
| 258 |
+
"logps/rejected": -1.2697585821151733,
|
| 259 |
+
"loss": 24.4222,
|
| 260 |
+
"nll_loss": 1.4941973686218262,
|
| 261 |
+
"rewards/accuracies": 0.6156250238418579,
|
| 262 |
+
"rewards/chosen": -0.05561145395040512,
|
| 263 |
+
"rewards/margins": 0.007876471616327763,
|
| 264 |
+
"rewards/rejected": -0.0634879320859909,
|
| 265 |
+
"step": 80
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.9164420485175202,
|
| 269 |
+
"grad_norm": 48.0,
|
| 270 |
+
"learning_rate": 4.745016742399804e-05,
|
| 271 |
+
"log_odds_chosen": 0.18154019117355347,
|
| 272 |
+
"log_odds_ratio": -0.6740174889564514,
|
| 273 |
+
"logps/chosen": -1.0852258205413818,
|
| 274 |
+
"logps/rejected": -1.2075302600860596,
|
| 275 |
+
"loss": 23.767,
|
| 276 |
+
"nll_loss": 1.451735496520996,
|
| 277 |
+
"rewards/accuracies": 0.578125,
|
| 278 |
+
"rewards/chosen": -0.05426129698753357,
|
| 279 |
+
"rewards/margins": 0.006115221884101629,
|
| 280 |
+
"rewards/rejected": -0.06037651747465134,
|
| 281 |
+
"step": 85
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.9703504043126685,
|
| 285 |
+
"grad_norm": 121.0,
|
| 286 |
+
"learning_rate": 4.690309299398736e-05,
|
| 287 |
+
"log_odds_chosen": 0.1481303721666336,
|
| 288 |
+
"log_odds_ratio": -0.6858943700790405,
|
| 289 |
+
"logps/chosen": -1.0602507591247559,
|
| 290 |
+
"logps/rejected": -1.1643625497817993,
|
| 291 |
+
"loss": 23.2092,
|
| 292 |
+
"nll_loss": 1.416282296180725,
|
| 293 |
+
"rewards/accuracies": 0.565625011920929,
|
| 294 |
+
"rewards/chosen": -0.05301254242658615,
|
| 295 |
+
"rewards/margins": 0.005205585155636072,
|
| 296 |
+
"rewards/rejected": -0.05821812152862549,
|
| 297 |
+
"step": 90
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 1.0215633423180592,
|
| 301 |
+
"grad_norm": 65.0,
|
| 302 |
+
"learning_rate": 4.630670030843785e-05,
|
| 303 |
+
"log_odds_chosen": 0.2527507245540619,
|
| 304 |
+
"log_odds_ratio": -0.6484514474868774,
|
| 305 |
+
"logps/chosen": -0.9999241232872009,
|
| 306 |
+
"logps/rejected": -1.1713911294937134,
|
| 307 |
+
"loss": 21.3461,
|
| 308 |
+
"nll_loss": 1.3719265460968018,
|
| 309 |
+
"rewards/accuracies": 0.625,
|
| 310 |
+
"rewards/chosen": -0.04999620094895363,
|
| 311 |
+
"rewards/margins": 0.00857335701584816,
|
| 312 |
+
"rewards/rejected": -0.05856956169009209,
|
| 313 |
+
"step": 95
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 1.0754716981132075,
|
| 317 |
+
"grad_norm": 39.0,
|
| 318 |
+
"learning_rate": 4.566233223919298e-05,
|
| 319 |
+
"log_odds_chosen": 0.5452004671096802,
|
| 320 |
+
"log_odds_ratio": -0.554630696773529,
|
| 321 |
+
"logps/chosen": -0.8976788520812988,
|
| 322 |
+
"logps/rejected": -1.2380362749099731,
|
| 323 |
+
"loss": 20.9491,
|
| 324 |
+
"nll_loss": 1.2815890312194824,
|
| 325 |
+
"rewards/accuracies": 0.690625011920929,
|
| 326 |
+
"rewards/chosen": -0.04488394409418106,
|
| 327 |
+
"rewards/margins": 0.017017865553498268,
|
| 328 |
+
"rewards/rejected": -0.06190181523561478,
|
| 329 |
+
"step": 100
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.1293800539083558,
|
| 333 |
+
"grad_norm": 36.25,
|
| 334 |
+
"learning_rate": 4.4971439682211125e-05,
|
| 335 |
+
"log_odds_chosen": 0.5778145790100098,
|
| 336 |
+
"log_odds_ratio": -0.5227857828140259,
|
| 337 |
+
"logps/chosen": -0.8327204585075378,
|
| 338 |
+
"logps/rejected": -1.1775634288787842,
|
| 339 |
+
"loss": 19.2746,
|
| 340 |
+
"nll_loss": 1.1785202026367188,
|
| 341 |
+
"rewards/accuracies": 0.7281249761581421,
|
| 342 |
+
"rewards/chosen": -0.04163602367043495,
|
| 343 |
+
"rewards/margins": 0.017242146655917168,
|
| 344 |
+
"rewards/rejected": -0.05887817218899727,
|
| 345 |
+
"step": 105
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.1832884097035041,
|
| 349 |
+
"grad_norm": 51.0,
|
| 350 |
+
"learning_rate": 4.4235578290645194e-05,
|
| 351 |
+
"log_odds_chosen": 0.564378023147583,
|
| 352 |
+
"log_odds_ratio": -0.5335876941680908,
|
| 353 |
+
"logps/chosen": -0.8683027029037476,
|
| 354 |
+
"logps/rejected": -1.225411057472229,
|
| 355 |
+
"loss": 20.1128,
|
| 356 |
+
"nll_loss": 1.2303677797317505,
|
| 357 |
+
"rewards/accuracies": 0.75,
|
| 358 |
+
"rewards/chosen": -0.0434151366353035,
|
| 359 |
+
"rewards/margins": 0.017855416983366013,
|
| 360 |
+
"rewards/rejected": -0.06127055361866951,
|
| 361 |
+
"step": 110
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 1.2371967654986522,
|
| 365 |
+
"grad_norm": 34.5,
|
| 366 |
+
"learning_rate": 4.3456404972045216e-05,
|
| 367 |
+
"log_odds_chosen": 0.6416139602661133,
|
| 368 |
+
"log_odds_ratio": -0.5017353296279907,
|
| 369 |
+
"logps/chosen": -0.8975638151168823,
|
| 370 |
+
"logps/rejected": -1.311712384223938,
|
| 371 |
+
"loss": 19.926,
|
| 372 |
+
"nll_loss": 1.220287799835205,
|
| 373 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 374 |
+
"rewards/chosen": -0.044878192245960236,
|
| 375 |
+
"rewards/margins": 0.02070743218064308,
|
| 376 |
+
"rewards/rejected": -0.06558562815189362,
|
| 377 |
+
"step": 115
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.2911051212938005,
|
| 381 |
+
"grad_norm": 29.375,
|
| 382 |
+
"learning_rate": 4.263567415757058e-05,
|
| 383 |
+
"log_odds_chosen": 0.5536849498748779,
|
| 384 |
+
"log_odds_ratio": -0.5178900957107544,
|
| 385 |
+
"logps/chosen": -0.8433765172958374,
|
| 386 |
+
"logps/rejected": -1.1795700788497925,
|
| 387 |
+
"loss": 18.7183,
|
| 388 |
+
"nll_loss": 1.1439992189407349,
|
| 389 |
+
"rewards/accuracies": 0.7718750238418579,
|
| 390 |
+
"rewards/chosen": -0.04216882586479187,
|
| 391 |
+
"rewards/margins": 0.016809683293104172,
|
| 392 |
+
"rewards/rejected": -0.05897850915789604,
|
| 393 |
+
"step": 120
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 1.3450134770889488,
|
| 397 |
+
"grad_norm": 40.25,
|
| 398 |
+
"learning_rate": 4.177523385161264e-05,
|
| 399 |
+
"log_odds_chosen": 0.5022929906845093,
|
| 400 |
+
"log_odds_ratio": -0.5432526469230652,
|
| 401 |
+
"logps/chosen": -0.830506443977356,
|
| 402 |
+
"logps/rejected": -1.1401817798614502,
|
| 403 |
+
"loss": 19.2524,
|
| 404 |
+
"nll_loss": 1.176114559173584,
|
| 405 |
+
"rewards/accuracies": 0.762499988079071,
|
| 406 |
+
"rewards/chosen": -0.041525326669216156,
|
| 407 |
+
"rewards/margins": 0.015483764931559563,
|
| 408 |
+
"rewards/rejected": -0.05700908973813057,
|
| 409 |
+
"step": 125
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.398921832884097,
|
| 413 |
+
"grad_norm": 45.5,
|
| 414 |
+
"learning_rate": 4.087702147072241e-05,
|
| 415 |
+
"log_odds_chosen": 0.7220330238342285,
|
| 416 |
+
"log_odds_ratio": -0.4684603810310364,
|
| 417 |
+
"logps/chosen": -0.8251385688781738,
|
| 418 |
+
"logps/rejected": -1.28031325340271,
|
| 419 |
+
"loss": 18.9627,
|
| 420 |
+
"nll_loss": 1.1617438793182373,
|
| 421 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 422 |
+
"rewards/chosen": -0.04125693440437317,
|
| 423 |
+
"rewards/margins": 0.022758739069104195,
|
| 424 |
+
"rewards/rejected": -0.06401567161083221,
|
| 425 |
+
"step": 130
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.4528301886792452,
|
| 429 |
+
"grad_norm": 42.75,
|
| 430 |
+
"learning_rate": 3.9943059481212795e-05,
|
| 431 |
+
"log_odds_chosen": 0.6586893796920776,
|
| 432 |
+
"log_odds_ratio": -0.4884856641292572,
|
| 433 |
+
"logps/chosen": -0.8284580111503601,
|
| 434 |
+
"logps/rejected": -1.2332229614257812,
|
| 435 |
+
"loss": 18.9402,
|
| 436 |
+
"nll_loss": 1.1593413352966309,
|
| 437 |
+
"rewards/accuracies": 0.800000011920929,
|
| 438 |
+
"rewards/chosen": -0.041422899812459946,
|
| 439 |
+
"rewards/margins": 0.02023824118077755,
|
| 440 |
+
"rewards/rejected": -0.061661142855882645,
|
| 441 |
+
"step": 135
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.5067385444743935,
|
| 445 |
+
"grad_norm": 40.0,
|
| 446 |
+
"learning_rate": 3.89754508452579e-05,
|
| 447 |
+
"log_odds_chosen": 0.5820509195327759,
|
| 448 |
+
"log_odds_ratio": -0.507893443107605,
|
| 449 |
+
"logps/chosen": -0.8388631939888,
|
| 450 |
+
"logps/rejected": -1.1933777332305908,
|
| 451 |
+
"loss": 18.6589,
|
| 452 |
+
"nll_loss": 1.1407865285873413,
|
| 453 |
+
"rewards/accuracies": 0.793749988079071,
|
| 454 |
+
"rewards/chosen": -0.04194316640496254,
|
| 455 |
+
"rewards/margins": 0.01772572472691536,
|
| 456 |
+
"rewards/rejected": -0.0596688874065876,
|
| 457 |
+
"step": 140
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.5606469002695418,
|
| 461 |
+
"grad_norm": 29.625,
|
| 462 |
+
"learning_rate": 3.797637428574326e-05,
|
| 463 |
+
"log_odds_chosen": 0.5326761603355408,
|
| 464 |
+
"log_odds_ratio": -0.5247709155082703,
|
| 465 |
+
"logps/chosen": -0.8665750622749329,
|
| 466 |
+
"logps/rejected": -1.1989167928695679,
|
| 467 |
+
"loss": 20.4033,
|
| 468 |
+
"nll_loss": 1.2489675283432007,
|
| 469 |
+
"rewards/accuracies": 0.765625,
|
| 470 |
+
"rewards/chosen": -0.043328750878572464,
|
| 471 |
+
"rewards/margins": 0.01661708578467369,
|
| 472 |
+
"rewards/rejected": -0.05994584038853645,
|
| 473 |
+
"step": 145
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.61455525606469,
|
| 477 |
+
"grad_norm": 46.0,
|
| 478 |
+
"learning_rate": 3.694807938052887e-05,
|
| 479 |
+
"log_odds_chosen": 0.5733472108840942,
|
| 480 |
+
"log_odds_ratio": -0.5206255912780762,
|
| 481 |
+
"logps/chosen": -0.8152993321418762,
|
| 482 |
+
"logps/rejected": -1.168148159980774,
|
| 483 |
+
"loss": 18.796,
|
| 484 |
+
"nll_loss": 1.1487157344818115,
|
| 485 |
+
"rewards/accuracies": 0.75,
|
| 486 |
+
"rewards/chosen": -0.04076496511697769,
|
| 487 |
+
"rewards/margins": 0.017642442137002945,
|
| 488 |
+
"rewards/rejected": -0.05840740725398064,
|
| 489 |
+
"step": 150
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.6684636118598384,
|
| 493 |
+
"grad_norm": 47.5,
|
| 494 |
+
"learning_rate": 3.589288149717119e-05,
|
| 495 |
+
"log_odds_chosen": 0.5549741387367249,
|
| 496 |
+
"log_odds_ratio": -0.5415369868278503,
|
| 497 |
+
"logps/chosen": -0.8091446161270142,
|
| 498 |
+
"logps/rejected": -1.1404829025268555,
|
| 499 |
+
"loss": 18.6901,
|
| 500 |
+
"nll_loss": 1.1410565376281738,
|
| 501 |
+
"rewards/accuracies": 0.746874988079071,
|
| 502 |
+
"rewards/chosen": -0.04045723378658295,
|
| 503 |
+
"rewards/margins": 0.016566917300224304,
|
| 504 |
+
"rewards/rejected": -0.057024143636226654,
|
| 505 |
+
"step": 155
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 1.7223719676549867,
|
| 509 |
+
"grad_norm": 41.25,
|
| 510 |
+
"learning_rate": 3.481315657950931e-05,
|
| 511 |
+
"log_odds_chosen": 0.6500066518783569,
|
| 512 |
+
"log_odds_ratio": -0.49195536971092224,
|
| 513 |
+
"logps/chosen": -0.8281671404838562,
|
| 514 |
+
"logps/rejected": -1.2156283855438232,
|
| 515 |
+
"loss": 18.8162,
|
| 516 |
+
"nll_loss": 1.1514135599136353,
|
| 517 |
+
"rewards/accuracies": 0.768750011920929,
|
| 518 |
+
"rewards/chosen": -0.04140836000442505,
|
| 519 |
+
"rewards/margins": 0.01937306486070156,
|
| 520 |
+
"rewards/rejected": -0.06078142672777176,
|
| 521 |
+
"step": 160
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.7762803234501348,
|
| 525 |
+
"grad_norm": 22.5,
|
| 526 |
+
"learning_rate": 3.3711335797853977e-05,
|
| 527 |
+
"log_odds_chosen": 0.7445698976516724,
|
| 528 |
+
"log_odds_ratio": -0.4790865480899811,
|
| 529 |
+
"logps/chosen": -0.8083413243293762,
|
| 530 |
+
"logps/rejected": -1.2825214862823486,
|
| 531 |
+
"loss": 18.5217,
|
| 532 |
+
"nll_loss": 1.1336489915847778,
|
| 533 |
+
"rewards/accuracies": 0.809374988079071,
|
| 534 |
+
"rewards/chosen": -0.04041706770658493,
|
| 535 |
+
"rewards/margins": 0.0237090103328228,
|
| 536 |
+
"rewards/rejected": -0.06412607431411743,
|
| 537 |
+
"step": 165
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 1.830188679245283,
|
| 541 |
+
"grad_norm": 39.0,
|
| 542 |
+
"learning_rate": 3.2589900074825696e-05,
|
| 543 |
+
"log_odds_chosen": 0.6931416988372803,
|
| 544 |
+
"log_odds_ratio": -0.4947904944419861,
|
| 545 |
+
"logps/chosen": -0.7949181795120239,
|
| 546 |
+
"logps/rejected": -1.2068727016448975,
|
| 547 |
+
"loss": 18.9459,
|
| 548 |
+
"nll_loss": 1.1593761444091797,
|
| 549 |
+
"rewards/accuracies": 0.778124988079071,
|
| 550 |
+
"rewards/chosen": -0.03974590823054314,
|
| 551 |
+
"rewards/margins": 0.02059771679341793,
|
| 552 |
+
"rewards/rejected": -0.060343630611896515,
|
| 553 |
+
"step": 170
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 1.8840970350404311,
|
| 557 |
+
"grad_norm": 47.25,
|
| 558 |
+
"learning_rate": 3.1451374499167586e-05,
|
| 559 |
+
"log_odds_chosen": 0.5809050798416138,
|
| 560 |
+
"log_odds_ratio": -0.5110514163970947,
|
| 561 |
+
"logps/chosen": -0.8269885182380676,
|
| 562 |
+
"logps/rejected": -1.1618260145187378,
|
| 563 |
+
"loss": 18.2167,
|
| 564 |
+
"nll_loss": 1.112992286682129,
|
| 565 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 566 |
+
"rewards/chosen": -0.04134942218661308,
|
| 567 |
+
"rewards/margins": 0.016741881147027016,
|
| 568 |
+
"rewards/rejected": -0.05809130519628525,
|
| 569 |
+
"step": 175
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 1.9380053908355794,
|
| 573 |
+
"grad_norm": 33.75,
|
| 574 |
+
"learning_rate": 3.029832264011133e-05,
|
| 575 |
+
"log_odds_chosen": 0.5635145902633667,
|
| 576 |
+
"log_odds_ratio": -0.5329084396362305,
|
| 577 |
+
"logps/chosen": -0.8633018732070923,
|
| 578 |
+
"logps/rejected": -1.2152156829833984,
|
| 579 |
+
"loss": 19.8266,
|
| 580 |
+
"nll_loss": 1.2125170230865479,
|
| 581 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 582 |
+
"rewards/chosen": -0.043165091425180435,
|
| 583 |
+
"rewards/margins": 0.017595689743757248,
|
| 584 |
+
"rewards/rejected": -0.06076078489422798,
|
| 585 |
+
"step": 180
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.9919137466307277,
|
| 589 |
+
"grad_norm": 26.375,
|
| 590 |
+
"learning_rate": 2.91333407750982e-05,
|
| 591 |
+
"log_odds_chosen": 0.6361402869224548,
|
| 592 |
+
"log_odds_ratio": -0.5029143691062927,
|
| 593 |
+
"logps/chosen": -0.8135613203048706,
|
| 594 |
+
"logps/rejected": -1.1941924095153809,
|
| 595 |
+
"loss": 18.7586,
|
| 596 |
+
"nll_loss": 1.1472654342651367,
|
| 597 |
+
"rewards/accuracies": 0.784375011920929,
|
| 598 |
+
"rewards/chosen": -0.04067806527018547,
|
| 599 |
+
"rewards/margins": 0.019031552597880363,
|
| 600 |
+
"rewards/rejected": -0.05970961973071098,
|
| 601 |
+
"step": 185
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 2.0431266846361185,
|
| 605 |
+
"grad_norm": 28.625,
|
| 606 |
+
"learning_rate": 2.7959052043852464e-05,
|
| 607 |
+
"log_odds_chosen": 1.1101665496826172,
|
| 608 |
+
"log_odds_ratio": -0.3634699881076813,
|
| 609 |
+
"logps/chosen": -0.6150734424591064,
|
| 610 |
+
"logps/rejected": -1.237518548965454,
|
| 611 |
+
"loss": 13.763,
|
| 612 |
+
"nll_loss": 0.8872840404510498,
|
| 613 |
+
"rewards/accuracies": 0.8980262875556946,
|
| 614 |
+
"rewards/chosen": -0.030753672122955322,
|
| 615 |
+
"rewards/margins": 0.031122254207730293,
|
| 616 |
+
"rewards/rejected": -0.06187593191862106,
|
| 617 |
+
"step": 190
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 2.0970350404312668,
|
| 621 |
+
"grad_norm": 26.625,
|
| 622 |
+
"learning_rate": 2.677810054197009e-05,
|
| 623 |
+
"log_odds_chosen": 1.4561563730239868,
|
| 624 |
+
"log_odds_ratio": -0.28991881012916565,
|
| 625 |
+
"logps/chosen": -0.5946121215820312,
|
| 626 |
+
"logps/rejected": -1.417449712753296,
|
| 627 |
+
"loss": 14.7392,
|
| 628 |
+
"nll_loss": 0.906706690788269,
|
| 629 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 630 |
+
"rewards/chosen": -0.029730603098869324,
|
| 631 |
+
"rewards/margins": 0.04114187881350517,
|
| 632 |
+
"rewards/rejected": -0.0708724781870842,
|
| 633 |
+
"step": 195
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.150943396226415,
|
| 637 |
+
"grad_norm": 36.5,
|
| 638 |
+
"learning_rate": 2.559314536732212e-05,
|
| 639 |
+
"log_odds_chosen": 1.3712774515151978,
|
| 640 |
+
"log_odds_ratio": -0.2994323670864105,
|
| 641 |
+
"logps/chosen": -0.5892333984375,
|
| 642 |
+
"logps/rejected": -1.355689287185669,
|
| 643 |
+
"loss": 13.7443,
|
| 644 |
+
"nll_loss": 0.8440448045730591,
|
| 645 |
+
"rewards/accuracies": 0.934374988079071,
|
| 646 |
+
"rewards/chosen": -0.02946167066693306,
|
| 647 |
+
"rewards/margins": 0.03832279518246651,
|
| 648 |
+
"rewards/rejected": -0.06778446584939957,
|
| 649 |
+
"step": 200
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 2.2048517520215634,
|
| 653 |
+
"grad_norm": 25.75,
|
| 654 |
+
"learning_rate": 2.4406854632677883e-05,
|
| 655 |
+
"log_odds_chosen": 1.447076439857483,
|
| 656 |
+
"log_odds_ratio": -0.29277992248535156,
|
| 657 |
+
"logps/chosen": -0.5634891390800476,
|
| 658 |
+
"logps/rejected": -1.386910080909729,
|
| 659 |
+
"loss": 13.4321,
|
| 660 |
+
"nll_loss": 0.8248661756515503,
|
| 661 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 662 |
+
"rewards/chosen": -0.02817445993423462,
|
| 663 |
+
"rewards/margins": 0.04117104411125183,
|
| 664 |
+
"rewards/rejected": -0.06934549659490585,
|
| 665 |
+
"step": 205
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 2.2587601078167117,
|
| 669 |
+
"grad_norm": 27.75,
|
| 670 |
+
"learning_rate": 2.3221899458029918e-05,
|
| 671 |
+
"log_odds_chosen": 1.4350051879882812,
|
| 672 |
+
"log_odds_ratio": -0.2998976409435272,
|
| 673 |
+
"logps/chosen": -0.5992333292961121,
|
| 674 |
+
"logps/rejected": -1.4122200012207031,
|
| 675 |
+
"loss": 14.0534,
|
| 676 |
+
"nll_loss": 0.8633407354354858,
|
| 677 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 678 |
+
"rewards/chosen": -0.02996166981756687,
|
| 679 |
+
"rewards/margins": 0.04064933583140373,
|
| 680 |
+
"rewards/rejected": -0.07061101496219635,
|
| 681 |
+
"step": 210
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 2.31266846361186,
|
| 685 |
+
"grad_norm": 28.25,
|
| 686 |
+
"learning_rate": 2.204094795614755e-05,
|
| 687 |
+
"log_odds_chosen": 1.4530997276306152,
|
| 688 |
+
"log_odds_ratio": -0.2895173728466034,
|
| 689 |
+
"logps/chosen": -0.5493010878562927,
|
| 690 |
+
"logps/rejected": -1.3495293855667114,
|
| 691 |
+
"loss": 13.5655,
|
| 692 |
+
"nll_loss": 0.8333690762519836,
|
| 693 |
+
"rewards/accuracies": 0.925000011920929,
|
| 694 |
+
"rewards/chosen": -0.027465051040053368,
|
| 695 |
+
"rewards/margins": 0.04001142084598541,
|
| 696 |
+
"rewards/rejected": -0.06747647374868393,
|
| 697 |
+
"step": 215
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 2.3665768194070083,
|
| 701 |
+
"grad_norm": 26.125,
|
| 702 |
+
"learning_rate": 2.0866659224901812e-05,
|
| 703 |
+
"log_odds_chosen": 1.4535211324691772,
|
| 704 |
+
"log_odds_ratio": -0.2938300371170044,
|
| 705 |
+
"logps/chosen": -0.5619036555290222,
|
| 706 |
+
"logps/rejected": -1.3884742259979248,
|
| 707 |
+
"loss": 13.1624,
|
| 708 |
+
"nll_loss": 0.8079586029052734,
|
| 709 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 710 |
+
"rewards/chosen": -0.02809518575668335,
|
| 711 |
+
"rewards/margins": 0.04132852703332901,
|
| 712 |
+
"rewards/rejected": -0.06942371279001236,
|
| 713 |
+
"step": 220
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 2.420485175202156,
|
| 717 |
+
"grad_norm": 44.75,
|
| 718 |
+
"learning_rate": 1.970167735988867e-05,
|
| 719 |
+
"log_odds_chosen": 1.4392869472503662,
|
| 720 |
+
"log_odds_ratio": -0.291705459356308,
|
| 721 |
+
"logps/chosen": -0.5864852666854858,
|
| 722 |
+
"logps/rejected": -1.387615442276001,
|
| 723 |
+
"loss": 13.6079,
|
| 724 |
+
"nll_loss": 0.8359111547470093,
|
| 725 |
+
"rewards/accuracies": 0.940625011920929,
|
| 726 |
+
"rewards/chosen": -0.029324263334274292,
|
| 727 |
+
"rewards/margins": 0.0400565080344677,
|
| 728 |
+
"rewards/rejected": -0.06938077509403229,
|
| 729 |
+
"step": 225
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 2.4743935309973044,
|
| 733 |
+
"grad_norm": 25.75,
|
| 734 |
+
"learning_rate": 1.854862550083241e-05,
|
| 735 |
+
"log_odds_chosen": 1.425066590309143,
|
| 736 |
+
"log_odds_ratio": -0.29249343276023865,
|
| 737 |
+
"logps/chosen": -0.558022141456604,
|
| 738 |
+
"logps/rejected": -1.3558303117752075,
|
| 739 |
+
"loss": 13.6401,
|
| 740 |
+
"nll_loss": 0.83788001537323,
|
| 741 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 742 |
+
"rewards/chosen": -0.02790110372006893,
|
| 743 |
+
"rewards/margins": 0.03989041596651077,
|
| 744 |
+
"rewards/rejected": -0.06779151409864426,
|
| 745 |
+
"step": 230
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 2.5283018867924527,
|
| 749 |
+
"grad_norm": 31.125,
|
| 750 |
+
"learning_rate": 1.7410099925174307e-05,
|
| 751 |
+
"log_odds_chosen": 1.3924018144607544,
|
| 752 |
+
"log_odds_ratio": -0.3068452477455139,
|
| 753 |
+
"logps/chosen": -0.5728594064712524,
|
| 754 |
+
"logps/rejected": -1.3289473056793213,
|
| 755 |
+
"loss": 13.4421,
|
| 756 |
+
"nll_loss": 0.8247901797294617,
|
| 757 |
+
"rewards/accuracies": 0.921875,
|
| 758 |
+
"rewards/chosen": -0.028642971068620682,
|
| 759 |
+
"rewards/margins": 0.03780439496040344,
|
| 760 |
+
"rewards/rejected": -0.06644736230373383,
|
| 761 |
+
"step": 235
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 2.582210242587601,
|
| 765 |
+
"grad_norm": 25.375,
|
| 766 |
+
"learning_rate": 1.628866420214603e-05,
|
| 767 |
+
"log_odds_chosen": 1.424349308013916,
|
| 768 |
+
"log_odds_ratio": -0.3004930913448334,
|
| 769 |
+
"logps/chosen": -0.5925456285476685,
|
| 770 |
+
"logps/rejected": -1.407995581626892,
|
| 771 |
+
"loss": 14.0136,
|
| 772 |
+
"nll_loss": 0.8608237504959106,
|
| 773 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 774 |
+
"rewards/chosen": -0.029627284035086632,
|
| 775 |
+
"rewards/margins": 0.04077249392867088,
|
| 776 |
+
"rewards/rejected": -0.07039977610111237,
|
| 777 |
+
"step": 240
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 2.6361185983827493,
|
| 781 |
+
"grad_norm": 28.375,
|
| 782 |
+
"learning_rate": 1.5186843420490698e-05,
|
| 783 |
+
"log_odds_chosen": 1.4117343425750732,
|
| 784 |
+
"log_odds_ratio": -0.29593607783317566,
|
| 785 |
+
"logps/chosen": -0.5560811758041382,
|
| 786 |
+
"logps/rejected": -1.315725564956665,
|
| 787 |
+
"loss": 14.3972,
|
| 788 |
+
"nll_loss": 0.8850256204605103,
|
| 789 |
+
"rewards/accuracies": 0.921875,
|
| 790 |
+
"rewards/chosen": -0.02780405804514885,
|
| 791 |
+
"rewards/margins": 0.03798222541809082,
|
| 792 |
+
"rewards/rejected": -0.06578627973794937,
|
| 793 |
+
"step": 245
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 2.6900269541778976,
|
| 797 |
+
"grad_norm": 25.0,
|
| 798 |
+
"learning_rate": 1.4107118502828814e-05,
|
| 799 |
+
"log_odds_chosen": 1.5662610530853271,
|
| 800 |
+
"log_odds_ratio": -0.27495214343070984,
|
| 801 |
+
"logps/chosen": -0.5344637036323547,
|
| 802 |
+
"logps/rejected": -1.4078654050827026,
|
| 803 |
+
"loss": 13.1208,
|
| 804 |
+
"nll_loss": 0.8063005208969116,
|
| 805 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 806 |
+
"rewards/chosen": -0.026723185554146767,
|
| 807 |
+
"rewards/margins": 0.043670084327459335,
|
| 808 |
+
"rewards/rejected": -0.07039327919483185,
|
| 809 |
+
"step": 250
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 2.743935309973046,
|
| 813 |
+
"grad_norm": 25.625,
|
| 814 |
+
"learning_rate": 1.3051920619471145e-05,
|
| 815 |
+
"log_odds_chosen": 1.4818260669708252,
|
| 816 |
+
"log_odds_ratio": -0.2981964945793152,
|
| 817 |
+
"logps/chosen": -0.5319411754608154,
|
| 818 |
+
"logps/rejected": -1.3410605192184448,
|
| 819 |
+
"loss": 13.8473,
|
| 820 |
+
"nll_loss": 0.8505468368530273,
|
| 821 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 822 |
+
"rewards/chosen": -0.02659705840051174,
|
| 823 |
+
"rewards/margins": 0.04045596718788147,
|
| 824 |
+
"rewards/rejected": -0.06705302000045776,
|
| 825 |
+
"step": 255
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 2.797843665768194,
|
| 829 |
+
"grad_norm": 20.875,
|
| 830 |
+
"learning_rate": 1.2023625714256743e-05,
|
| 831 |
+
"log_odds_chosen": 1.3853600025177002,
|
| 832 |
+
"log_odds_ratio": -0.3066624402999878,
|
| 833 |
+
"logps/chosen": -0.5818651914596558,
|
| 834 |
+
"logps/rejected": -1.3556396961212158,
|
| 835 |
+
"loss": 13.8096,
|
| 836 |
+
"nll_loss": 0.8477646112442017,
|
| 837 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 838 |
+
"rewards/chosen": -0.02909325622022152,
|
| 839 |
+
"rewards/margins": 0.03868872672319412,
|
| 840 |
+
"rewards/rejected": -0.06778198480606079,
|
| 841 |
+
"step": 260
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 2.8517520215633425,
|
| 845 |
+
"grad_norm": 25.875,
|
| 846 |
+
"learning_rate": 1.1024549154742098e-05,
|
| 847 |
+
"log_odds_chosen": 1.3135344982147217,
|
| 848 |
+
"log_odds_ratio": -0.3150535225868225,
|
| 849 |
+
"logps/chosen": -0.5827392339706421,
|
| 850 |
+
"logps/rejected": -1.295180320739746,
|
| 851 |
+
"loss": 13.5359,
|
| 852 |
+
"nll_loss": 0.8302389979362488,
|
| 853 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 854 |
+
"rewards/chosen": -0.029136961326003075,
|
| 855 |
+
"rewards/margins": 0.03562205657362938,
|
| 856 |
+
"rewards/rejected": -0.0647590309381485,
|
| 857 |
+
"step": 265
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 2.9056603773584904,
|
| 861 |
+
"grad_norm": 26.75,
|
| 862 |
+
"learning_rate": 1.005694051878721e-05,
|
| 863 |
+
"log_odds_chosen": 1.3994379043579102,
|
| 864 |
+
"log_odds_ratio": -0.3157843351364136,
|
| 865 |
+
"logps/chosen": -0.5678838491439819,
|
| 866 |
+
"logps/rejected": -1.3345766067504883,
|
| 867 |
+
"loss": 13.6712,
|
| 868 |
+
"nll_loss": 0.8386627435684204,
|
| 869 |
+
"rewards/accuracies": 0.90625,
|
| 870 |
+
"rewards/chosen": -0.028394188731908798,
|
| 871 |
+
"rewards/margins": 0.03833463788032532,
|
| 872 |
+
"rewards/rejected": -0.06672882288694382,
|
| 873 |
+
"step": 270
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.9595687331536387,
|
| 877 |
+
"grad_norm": 25.5,
|
| 878 |
+
"learning_rate": 9.1229785292776e-06,
|
| 879 |
+
"log_odds_chosen": 1.4966779947280884,
|
| 880 |
+
"log_odds_ratio": -0.28297117352485657,
|
| 881 |
+
"logps/chosen": -0.5280012488365173,
|
| 882 |
+
"logps/rejected": -1.359018325805664,
|
| 883 |
+
"loss": 13.0722,
|
| 884 |
+
"nll_loss": 0.8028658628463745,
|
| 885 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 886 |
+
"rewards/chosen": -0.026400065049529076,
|
| 887 |
+
"rewards/margins": 0.04155084863305092,
|
| 888 |
+
"rewards/rejected": -0.06795090436935425,
|
| 889 |
+
"step": 275
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 3.01078167115903,
|
| 893 |
+
"grad_norm": 24.5,
|
| 894 |
+
"learning_rate": 8.22476614838737e-06,
|
| 895 |
+
"log_odds_chosen": 1.5248451232910156,
|
| 896 |
+
"log_odds_ratio": -0.2791544198989868,
|
| 897 |
+
"logps/chosen": -0.540464460849762,
|
| 898 |
+
"logps/rejected": -1.3801342248916626,
|
| 899 |
+
"loss": 12.3085,
|
| 900 |
+
"nll_loss": 0.7958097457885742,
|
| 901 |
+
"rewards/accuracies": 0.9407894611358643,
|
| 902 |
+
"rewards/chosen": -0.027023224160075188,
|
| 903 |
+
"rewards/margins": 0.04198349267244339,
|
| 904 |
+
"rewards/rejected": -0.06900670379400253,
|
| 905 |
+
"step": 280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 3.0646900269541777,
|
| 909 |
+
"grad_norm": 20.75,
|
| 910 |
+
"learning_rate": 7.364325842429417e-06,
|
| 911 |
+
"log_odds_chosen": 2.0129177570343018,
|
| 912 |
+
"log_odds_ratio": -0.1976321041584015,
|
| 913 |
+
"logps/chosen": -0.43267732858657837,
|
| 914 |
+
"logps/rejected": -1.4722133874893188,
|
| 915 |
+
"loss": 10.7779,
|
| 916 |
+
"nll_loss": 0.6637340784072876,
|
| 917 |
+
"rewards/accuracies": 0.981249988079071,
|
| 918 |
+
"rewards/chosen": -0.02163386344909668,
|
| 919 |
+
"rewards/margins": 0.05197680741548538,
|
| 920 |
+
"rewards/rejected": -0.07361067086458206,
|
| 921 |
+
"step": 285
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 3.118598382749326,
|
| 925 |
+
"grad_norm": 33.5,
|
| 926 |
+
"learning_rate": 6.543595027954785e-06,
|
| 927 |
+
"log_odds_chosen": 2.077929735183716,
|
| 928 |
+
"log_odds_ratio": -0.2039026916027069,
|
| 929 |
+
"logps/chosen": -0.4455755650997162,
|
| 930 |
+
"logps/rejected": -1.5429691076278687,
|
| 931 |
+
"loss": 10.7447,
|
| 932 |
+
"nll_loss": 0.6613457798957825,
|
| 933 |
+
"rewards/accuracies": 0.96875,
|
| 934 |
+
"rewards/chosen": -0.02227878011763096,
|
| 935 |
+
"rewards/margins": 0.05486968904733658,
|
| 936 |
+
"rewards/rejected": -0.07714846730232239,
|
| 937 |
+
"step": 290
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 3.1725067385444743,
|
| 941 |
+
"grad_norm": 26.375,
|
| 942 |
+
"learning_rate": 5.764421709354811e-06,
|
| 943 |
+
"log_odds_chosen": 2.2563636302948,
|
| 944 |
+
"log_odds_ratio": -0.17267544567584991,
|
| 945 |
+
"logps/chosen": -0.4047982692718506,
|
| 946 |
+
"logps/rejected": -1.6122785806655884,
|
| 947 |
+
"loss": 10.4634,
|
| 948 |
+
"nll_loss": 0.6453262567520142,
|
| 949 |
+
"rewards/accuracies": 0.96875,
|
| 950 |
+
"rewards/chosen": -0.02023991569876671,
|
| 951 |
+
"rewards/margins": 0.06037401407957077,
|
| 952 |
+
"rewards/rejected": -0.08061392605304718,
|
| 953 |
+
"step": 295
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 3.2264150943396226,
|
| 957 |
+
"grad_norm": 22.25,
|
| 958 |
+
"learning_rate": 5.02856031778888e-06,
|
| 959 |
+
"log_odds_chosen": 2.167142152786255,
|
| 960 |
+
"log_odds_ratio": -0.19436317682266235,
|
| 961 |
+
"logps/chosen": -0.43167462944984436,
|
| 962 |
+
"logps/rejected": -1.5412400960922241,
|
| 963 |
+
"loss": 10.7259,
|
| 964 |
+
"nll_loss": 0.6606518030166626,
|
| 965 |
+
"rewards/accuracies": 0.949999988079071,
|
| 966 |
+
"rewards/chosen": -0.021583732217550278,
|
| 967 |
+
"rewards/margins": 0.055478282272815704,
|
| 968 |
+
"rewards/rejected": -0.07706201076507568,
|
| 969 |
+
"step": 300
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 3.280323450134771,
|
| 973 |
+
"grad_norm": 22.75,
|
| 974 |
+
"learning_rate": 4.337667760807018e-06,
|
| 975 |
+
"log_odds_chosen": 2.055614471435547,
|
| 976 |
+
"log_odds_ratio": -0.20666727423667908,
|
| 977 |
+
"logps/chosen": -0.4421593248844147,
|
| 978 |
+
"logps/rejected": -1.5431147813796997,
|
| 979 |
+
"loss": 10.7846,
|
| 980 |
+
"nll_loss": 0.6637014746665955,
|
| 981 |
+
"rewards/accuracies": 0.971875011920929,
|
| 982 |
+
"rewards/chosen": -0.022107966244220734,
|
| 983 |
+
"rewards/margins": 0.05504777282476425,
|
| 984 |
+
"rewards/rejected": -0.07715573906898499,
|
| 985 |
+
"step": 305
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 3.334231805929919,
|
| 989 |
+
"grad_norm": 27.875,
|
| 990 |
+
"learning_rate": 3.69329969156216e-06,
|
| 991 |
+
"log_odds_chosen": 2.183558940887451,
|
| 992 |
+
"log_odds_ratio": -0.17826569080352783,
|
| 993 |
+
"logps/chosen": -0.43023914098739624,
|
| 994 |
+
"logps/rejected": -1.6062309741973877,
|
| 995 |
+
"loss": 10.7638,
|
| 996 |
+
"nll_loss": 0.663822591304779,
|
| 997 |
+
"rewards/accuracies": 0.971875011920929,
|
| 998 |
+
"rewards/chosen": -0.021511957049369812,
|
| 999 |
+
"rewards/margins": 0.058799587190151215,
|
| 1000 |
+
"rewards/rejected": -0.08031155169010162,
|
| 1001 |
+
"step": 310
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 3.3881401617250675,
|
| 1005 |
+
"grad_norm": 27.875,
|
| 1006 |
+
"learning_rate": 3.096907006012639e-06,
|
| 1007 |
+
"log_odds_chosen": 2.1112418174743652,
|
| 1008 |
+
"log_odds_ratio": -0.19758489727973938,
|
| 1009 |
+
"logps/chosen": -0.4420015215873718,
|
| 1010 |
+
"logps/rejected": -1.5825992822647095,
|
| 1011 |
+
"loss": 10.8219,
|
| 1012 |
+
"nll_loss": 0.6664899587631226,
|
| 1013 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1014 |
+
"rewards/chosen": -0.02210007607936859,
|
| 1015 |
+
"rewards/margins": 0.05702989175915718,
|
| 1016 |
+
"rewards/rejected": -0.07912997156381607,
|
| 1017 |
+
"step": 315
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 3.442048517520216,
|
| 1021 |
+
"grad_norm": 23.875,
|
| 1022 |
+
"learning_rate": 2.549832576001965e-06,
|
| 1023 |
+
"log_odds_chosen": 2.2864553928375244,
|
| 1024 |
+
"log_odds_ratio": -0.1727539300918579,
|
| 1025 |
+
"logps/chosen": -0.42124342918395996,
|
| 1026 |
+
"logps/rejected": -1.643252968788147,
|
| 1027 |
+
"loss": 10.4624,
|
| 1028 |
+
"nll_loss": 0.6452642679214478,
|
| 1029 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 1030 |
+
"rewards/chosen": -0.021062172949314117,
|
| 1031 |
+
"rewards/margins": 0.061100464314222336,
|
| 1032 |
+
"rewards/rejected": -0.08216263353824615,
|
| 1033 |
+
"step": 320
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 3.4959568733153636,
|
| 1037 |
+
"grad_norm": 24.75,
|
| 1038 |
+
"learning_rate": 2.0533082255718594e-06,
|
| 1039 |
+
"log_odds_chosen": 2.240377187728882,
|
| 1040 |
+
"log_odds_ratio": -0.18875715136528015,
|
| 1041 |
+
"logps/chosen": -0.39375439286231995,
|
| 1042 |
+
"logps/rejected": -1.5621250867843628,
|
| 1043 |
+
"loss": 9.993,
|
| 1044 |
+
"nll_loss": 0.6151254177093506,
|
| 1045 |
+
"rewards/accuracies": 0.96875,
|
| 1046 |
+
"rewards/chosen": -0.019687717780470848,
|
| 1047 |
+
"rewards/margins": 0.058418530970811844,
|
| 1048 |
+
"rewards/rejected": -0.07810626178979874,
|
| 1049 |
+
"step": 325
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 3.5498652291105124,
|
| 1053 |
+
"grad_norm": 28.5,
|
| 1054 |
+
"learning_rate": 1.6084519573168727e-06,
|
| 1055 |
+
"log_odds_chosen": 2.1224582195281982,
|
| 1056 |
+
"log_odds_ratio": -0.2051163613796234,
|
| 1057 |
+
"logps/chosen": -0.4335021376609802,
|
| 1058 |
+
"logps/rejected": -1.5549887418746948,
|
| 1059 |
+
"loss": 10.6481,
|
| 1060 |
+
"nll_loss": 0.6552503705024719,
|
| 1061 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1062 |
+
"rewards/chosen": -0.0216751080006361,
|
| 1063 |
+
"rewards/margins": 0.05607432872056961,
|
| 1064 |
+
"rewards/rejected": -0.07774944603443146,
|
| 1065 |
+
"step": 330
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 3.6037735849056602,
|
| 1069 |
+
"grad_norm": 21.625,
|
| 1070 |
+
"learning_rate": 1.2162654350258402e-06,
|
| 1071 |
+
"log_odds_chosen": 2.2741246223449707,
|
| 1072 |
+
"log_odds_ratio": -0.18007881939411163,
|
| 1073 |
+
"logps/chosen": -0.438800573348999,
|
| 1074 |
+
"logps/rejected": -1.675838828086853,
|
| 1075 |
+
"loss": 10.4112,
|
| 1076 |
+
"nll_loss": 0.6416950821876526,
|
| 1077 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1078 |
+
"rewards/chosen": -0.02194002829492092,
|
| 1079 |
+
"rewards/margins": 0.06185191869735718,
|
| 1080 |
+
"rewards/rejected": -0.08379194885492325,
|
| 1081 |
+
"step": 335
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 3.6576819407008085,
|
| 1085 |
+
"grad_norm": 27.5,
|
| 1086 |
+
"learning_rate": 8.776317282785163e-07,
|
| 1087 |
+
"log_odds_chosen": 2.210236072540283,
|
| 1088 |
+
"log_odds_ratio": -0.1864241361618042,
|
| 1089 |
+
"logps/chosen": -0.4287256598472595,
|
| 1090 |
+
"logps/rejected": -1.5987118482589722,
|
| 1091 |
+
"loss": 10.3796,
|
| 1092 |
+
"nll_loss": 0.6394068598747253,
|
| 1093 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1094 |
+
"rewards/chosen": -0.021436283364892006,
|
| 1095 |
+
"rewards/margins": 0.05849931389093399,
|
| 1096 |
+
"rewards/rejected": -0.07993558794260025,
|
| 1097 |
+
"step": 340
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 3.711590296495957,
|
| 1101 |
+
"grad_norm": 25.875,
|
| 1102 |
+
"learning_rate": 5.933133240756117e-07,
|
| 1103 |
+
"log_odds_chosen": 2.208949327468872,
|
| 1104 |
+
"log_odds_ratio": -0.18415012955665588,
|
| 1105 |
+
"logps/chosen": -0.44286638498306274,
|
| 1106 |
+
"logps/rejected": -1.6374422311782837,
|
| 1107 |
+
"loss": 10.6099,
|
| 1108 |
+
"nll_loss": 0.6539115905761719,
|
| 1109 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1110 |
+
"rewards/chosen": -0.022143317386507988,
|
| 1111 |
+
"rewards/margins": 0.05972879007458687,
|
| 1112 |
+
"rewards/rejected": -0.0818721130490303,
|
| 1113 |
+
"step": 345
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 3.765498652291105,
|
| 1117 |
+
"grad_norm": 21.375,
|
| 1118 |
+
"learning_rate": 3.6395040997952714e-07,
|
| 1119 |
+
"log_odds_chosen": 2.108987331390381,
|
| 1120 |
+
"log_odds_ratio": -0.19528530538082123,
|
| 1121 |
+
"logps/chosen": -0.42990946769714355,
|
| 1122 |
+
"logps/rejected": -1.5616236925125122,
|
| 1123 |
+
"loss": 10.8233,
|
| 1124 |
+
"nll_loss": 0.6666921377182007,
|
| 1125 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1126 |
+
"rewards/chosen": -0.021495472639799118,
|
| 1127 |
+
"rewards/margins": 0.05658571049571037,
|
| 1128 |
+
"rewards/rejected": -0.0780811756849289,
|
| 1129 |
+
"step": 350
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 3.8194070080862534,
|
| 1133 |
+
"grad_norm": 24.25,
|
| 1134 |
+
"learning_rate": 1.9005943263146198e-07,
|
| 1135 |
+
"log_odds_chosen": 2.360919237136841,
|
| 1136 |
+
"log_odds_ratio": -0.1664571613073349,
|
| 1137 |
+
"logps/chosen": -0.3908630311489105,
|
| 1138 |
+
"logps/rejected": -1.650352120399475,
|
| 1139 |
+
"loss": 10.2428,
|
| 1140 |
+
"nll_loss": 0.6318550109863281,
|
| 1141 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1142 |
+
"rewards/chosen": -0.019543152302503586,
|
| 1143 |
+
"rewards/margins": 0.0629744604229927,
|
| 1144 |
+
"rewards/rejected": -0.082517609000206,
|
| 1145 |
+
"step": 355
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 3.8733153638814017,
|
| 1149 |
+
"grad_norm": 29.25,
|
| 1150 |
+
"learning_rate": 7.203193489064009e-08,
|
| 1151 |
+
"log_odds_chosen": 2.1713168621063232,
|
| 1152 |
+
"log_odds_ratio": -0.1880660057067871,
|
| 1153 |
+
"logps/chosen": -0.41322654485702515,
|
| 1154 |
+
"logps/rejected": -1.57550048828125,
|
| 1155 |
+
"loss": 10.3482,
|
| 1156 |
+
"nll_loss": 0.6373571157455444,
|
| 1157 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1158 |
+
"rewards/chosen": -0.020661327987909317,
|
| 1159 |
+
"rewards/margins": 0.058113694190979004,
|
| 1160 |
+
"rewards/rejected": -0.07877503335475922,
|
| 1161 |
+
"step": 360
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 3.92722371967655,
|
| 1165 |
+
"grad_norm": 24.75,
|
| 1166 |
+
"learning_rate": 1.0133674214043987e-08,
|
| 1167 |
+
"log_odds_chosen": 2.3266632556915283,
|
| 1168 |
+
"log_odds_ratio": -0.16368618607521057,
|
| 1169 |
+
"logps/chosen": -0.3924325108528137,
|
| 1170 |
+
"logps/rejected": -1.6150646209716797,
|
| 1171 |
+
"loss": 9.8053,
|
| 1172 |
+
"nll_loss": 0.6046454310417175,
|
| 1173 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1174 |
+
"rewards/chosen": -0.019621625542640686,
|
| 1175 |
+
"rewards/margins": 0.06113160774111748,
|
| 1176 |
+
"rewards/rejected": -0.08075322955846786,
|
| 1177 |
+
"step": 365
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 3.9595687331536387,
|
| 1181 |
+
"step": 368,
|
| 1182 |
+
"total_flos": 0.0,
|
| 1183 |
+
"train_loss": 25.71915608903636,
|
| 1184 |
+
"train_runtime": 4429.3261,
|
| 1185 |
+
"train_samples_per_second": 5.361,
|
| 1186 |
+
"train_steps_per_second": 0.083
|
| 1187 |
+
}
|
| 1188 |
+
],
|
| 1189 |
+
"logging_steps": 5,
|
| 1190 |
+
"max_steps": 368,
|
| 1191 |
+
"num_input_tokens_seen": 0,
|
| 1192 |
+
"num_train_epochs": 4,
|
| 1193 |
+
"save_steps": 100,
|
| 1194 |
+
"stateful_callbacks": {
|
| 1195 |
+
"TrainerControl": {
|
| 1196 |
+
"args": {
|
| 1197 |
+
"should_epoch_stop": false,
|
| 1198 |
+
"should_evaluate": false,
|
| 1199 |
+
"should_log": false,
|
| 1200 |
+
"should_save": true,
|
| 1201 |
+
"should_training_stop": true
|
| 1202 |
+
},
|
| 1203 |
+
"attributes": {}
|
| 1204 |
+
}
|
| 1205 |
+
},
|
| 1206 |
+
"total_flos": 0.0,
|
| 1207 |
+
"train_batch_size": 1,
|
| 1208 |
+
"trial_name": null,
|
| 1209 |
+
"trial_params": null
|
| 1210 |
+
}
|