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-borpo-noisy-5e-5-005-norm2
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-borpo-noisy-5e-5-005-norm2
|
| 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-borpo-noisy-5e-5-005-norm2", device="cuda")
|
| 24 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 25 |
+
print(output["generated_text"])
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Training procedure
|
| 29 |
+
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/silvias/huggingface/runs/p8p8dr5w)
|
| 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": 50.92685067313058,
|
| 5 |
+
"train_runtime": 8087.8942,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.504,
|
| 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": 50.92685067313058,
|
| 5 |
+
"train_runtime": 8087.8942,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.504,
|
| 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": 677.5082397460938,
|
| 14 |
+
"learning_rate": 7.8125e-06,
|
| 15 |
+
"log_odds_chosen": 4.972014427185059,
|
| 16 |
+
"log_odds_ratio": -9.77397632598877,
|
| 17 |
+
"logits/chosen": 138.50723266601562,
|
| 18 |
+
"logits/rejected": 153.07931518554688,
|
| 19 |
+
"logps/chosen": -20.706424713134766,
|
| 20 |
+
"logps/rejected": -25.678659439086914,
|
| 21 |
+
"loss": 295.6143,
|
| 22 |
+
"nll_loss": 8.335786819458008,
|
| 23 |
+
"rewards/accuracies": 0.53125,
|
| 24 |
+
"rewards/chosen": -1.0353213548660278,
|
| 25 |
+
"rewards/margins": 0.24861177802085876,
|
| 26 |
+
"rewards/rejected": -1.2839329242706299,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 424.99365234375,
|
| 32 |
+
"learning_rate": 1.5625e-05,
|
| 33 |
+
"log_odds_chosen": 2.362799644470215,
|
| 34 |
+
"log_odds_ratio": -7.1902971267700195,
|
| 35 |
+
"logits/chosen": 135.88121032714844,
|
| 36 |
+
"logits/rejected": 160.3680419921875,
|
| 37 |
+
"logps/chosen": -15.390497207641602,
|
| 38 |
+
"logps/rejected": -17.75222396850586,
|
| 39 |
+
"loss": 244.5906,
|
| 40 |
+
"nll_loss": 7.254161834716797,
|
| 41 |
+
"rewards/accuracies": 0.543749988079071,
|
| 42 |
+
"rewards/chosen": -0.7695248126983643,
|
| 43 |
+
"rewards/margins": 0.1180863231420517,
|
| 44 |
+
"rewards/rejected": -0.8876112103462219,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 270.6084899902344,
|
| 50 |
+
"learning_rate": 2.34375e-05,
|
| 51 |
+
"log_odds_chosen": 4.370176792144775,
|
| 52 |
+
"log_odds_ratio": -7.611123085021973,
|
| 53 |
+
"logits/chosen": 126.235595703125,
|
| 54 |
+
"logits/rejected": 153.06771850585938,
|
| 55 |
+
"logps/chosen": -19.671573638916016,
|
| 56 |
+
"logps/rejected": -24.04007339477539,
|
| 57 |
+
"loss": 286.111,
|
| 58 |
+
"nll_loss": 8.265220642089844,
|
| 59 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 60 |
+
"rewards/chosen": -0.9835788011550903,
|
| 61 |
+
"rewards/margins": 0.218424990773201,
|
| 62 |
+
"rewards/rejected": -1.2020035982131958,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 2947.77685546875,
|
| 68 |
+
"learning_rate": 3.125e-05,
|
| 69 |
+
"log_odds_chosen": -0.3353792726993561,
|
| 70 |
+
"log_odds_ratio": -5.916926383972168,
|
| 71 |
+
"logits/chosen": 159.31063842773438,
|
| 72 |
+
"logits/rejected": 163.8131561279297,
|
| 73 |
+
"logps/chosen": -13.794464111328125,
|
| 74 |
+
"logps/rejected": -13.460695266723633,
|
| 75 |
+
"loss": 223.2703,
|
| 76 |
+
"nll_loss": 6.456460475921631,
|
| 77 |
+
"rewards/accuracies": 0.543749988079071,
|
| 78 |
+
"rewards/chosen": -0.6897231936454773,
|
| 79 |
+
"rewards/margins": -0.016688454896211624,
|
| 80 |
+
"rewards/rejected": -0.6730347871780396,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 194.74801635742188,
|
| 86 |
+
"learning_rate": 3.90625e-05,
|
| 87 |
+
"log_odds_chosen": 1.4447908401489258,
|
| 88 |
+
"log_odds_ratio": -1.7767770290374756,
|
| 89 |
+
"logits/chosen": 184.77932739257812,
|
| 90 |
+
"logits/rejected": 192.9888916015625,
|
| 91 |
+
"logps/chosen": -4.297823429107666,
|
| 92 |
+
"logps/rejected": -5.7312822341918945,
|
| 93 |
+
"loss": 82.643,
|
| 94 |
+
"nll_loss": 2.5885708332061768,
|
| 95 |
+
"rewards/accuracies": 0.5625,
|
| 96 |
+
"rewards/chosen": -0.21489116549491882,
|
| 97 |
+
"rewards/margins": 0.0716729462146759,
|
| 98 |
+
"rewards/rejected": -0.2865641117095947,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 331.9227600097656,
|
| 104 |
+
"learning_rate": 4.6875e-05,
|
| 105 |
+
"log_odds_chosen": 0.1392047107219696,
|
| 106 |
+
"log_odds_ratio": -0.8184011578559875,
|
| 107 |
+
"logits/chosen": 195.76513671875,
|
| 108 |
+
"logits/rejected": 218.49557495117188,
|
| 109 |
+
"logps/chosen": -1.6848766803741455,
|
| 110 |
+
"logps/rejected": -1.8006423711776733,
|
| 111 |
+
"loss": 62.6908,
|
| 112 |
+
"nll_loss": 1.9485574960708618,
|
| 113 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 114 |
+
"rewards/chosen": -0.08424384146928787,
|
| 115 |
+
"rewards/margins": 0.005788279231637716,
|
| 116 |
+
"rewards/rejected": -0.09003211557865143,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 115.5135498046875,
|
| 122 |
+
"learning_rate": 4.998613757348784e-05,
|
| 123 |
+
"log_odds_chosen": 0.006599183194339275,
|
| 124 |
+
"log_odds_ratio": -0.8610709309577942,
|
| 125 |
+
"logits/chosen": 240.92092895507812,
|
| 126 |
+
"logits/rejected": 237.75436401367188,
|
| 127 |
+
"logps/chosen": -1.599082112312317,
|
| 128 |
+
"logps/rejected": -1.6070514917373657,
|
| 129 |
+
"loss": 56.7162,
|
| 130 |
+
"nll_loss": 1.7848625183105469,
|
| 131 |
+
"rewards/accuracies": 0.512499988079071,
|
| 132 |
+
"rewards/chosen": -0.0799541026353836,
|
| 133 |
+
"rewards/margins": 0.00039846962317824364,
|
| 134 |
+
"rewards/rejected": -0.08035256713628769,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 102.91368865966797,
|
| 140 |
+
"learning_rate": 4.990147841143462e-05,
|
| 141 |
+
"log_odds_chosen": 0.20803086459636688,
|
| 142 |
+
"log_odds_ratio": -0.7011880874633789,
|
| 143 |
+
"logits/chosen": 223.38040161132812,
|
| 144 |
+
"logits/rejected": 223.494873046875,
|
| 145 |
+
"logps/chosen": -1.3678431510925293,
|
| 146 |
+
"logps/rejected": -1.5363763570785522,
|
| 147 |
+
"loss": 53.8445,
|
| 148 |
+
"nll_loss": 1.6675764322280884,
|
| 149 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 150 |
+
"rewards/chosen": -0.06839215010404587,
|
| 151 |
+
"rewards/margins": 0.008426659740507603,
|
| 152 |
+
"rewards/rejected": -0.07681881636381149,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 72.4631576538086,
|
| 158 |
+
"learning_rate": 4.97401218720448e-05,
|
| 159 |
+
"log_odds_chosen": 0.17005965113639832,
|
| 160 |
+
"log_odds_ratio": -0.780523419380188,
|
| 161 |
+
"logits/chosen": 195.12318420410156,
|
| 162 |
+
"logits/rejected": 192.85745239257812,
|
| 163 |
+
"logps/chosen": -1.381034255027771,
|
| 164 |
+
"logps/rejected": -1.5216766595840454,
|
| 165 |
+
"loss": 51.6844,
|
| 166 |
+
"nll_loss": 1.6356455087661743,
|
| 167 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 168 |
+
"rewards/chosen": -0.06905172020196915,
|
| 169 |
+
"rewards/margins": 0.007032120134681463,
|
| 170 |
+
"rewards/rejected": -0.07608383893966675,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 189.4646759033203,
|
| 176 |
+
"learning_rate": 4.9502564938797946e-05,
|
| 177 |
+
"log_odds_chosen": 0.17814292013645172,
|
| 178 |
+
"log_odds_ratio": -0.7335196733474731,
|
| 179 |
+
"logits/chosen": 187.52923583984375,
|
| 180 |
+
"logits/rejected": 187.25880432128906,
|
| 181 |
+
"logps/chosen": -1.3315727710723877,
|
| 182 |
+
"logps/rejected": -1.4765068292617798,
|
| 183 |
+
"loss": 52.9646,
|
| 184 |
+
"nll_loss": 1.67325758934021,
|
| 185 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 186 |
+
"rewards/chosen": -0.06657864153385162,
|
| 187 |
+
"rewards/margins": 0.0072467029094696045,
|
| 188 |
+
"rewards/rejected": -0.07382534444332123,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 109.14909362792969,
|
| 194 |
+
"learning_rate": 4.918953929490768e-05,
|
| 195 |
+
"log_odds_chosen": 0.11172562837600708,
|
| 196 |
+
"log_odds_ratio": -0.7358081340789795,
|
| 197 |
+
"logits/chosen": 181.12258911132812,
|
| 198 |
+
"logits/rejected": 176.92349243164062,
|
| 199 |
+
"logps/chosen": -1.2745424509048462,
|
| 200 |
+
"logps/rejected": -1.3694889545440674,
|
| 201 |
+
"loss": 48.8487,
|
| 202 |
+
"nll_loss": 1.5679895877838135,
|
| 203 |
+
"rewards/accuracies": 0.543749988079071,
|
| 204 |
+
"rewards/chosen": -0.06372712552547455,
|
| 205 |
+
"rewards/margins": 0.004747326020151377,
|
| 206 |
+
"rewards/rejected": -0.06847444176673889,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 174.18399047851562,
|
| 212 |
+
"learning_rate": 4.88020090697132e-05,
|
| 213 |
+
"log_odds_chosen": 0.21670827269554138,
|
| 214 |
+
"log_odds_ratio": -0.6783889532089233,
|
| 215 |
+
"logits/chosen": 178.7930908203125,
|
| 216 |
+
"logits/rejected": 172.34970092773438,
|
| 217 |
+
"logps/chosen": -1.1696356534957886,
|
| 218 |
+
"logps/rejected": -1.3248437643051147,
|
| 219 |
+
"loss": 47.1801,
|
| 220 |
+
"nll_loss": 1.456142783164978,
|
| 221 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 222 |
+
"rewards/chosen": -0.05848178267478943,
|
| 223 |
+
"rewards/margins": 0.007760416716337204,
|
| 224 |
+
"rewards/rejected": -0.06624220311641693,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 140.9652557373047,
|
| 230 |
+
"learning_rate": 4.834116786912897e-05,
|
| 231 |
+
"log_odds_chosen": 0.18169096112251282,
|
| 232 |
+
"log_odds_ratio": -0.7172664403915405,
|
| 233 |
+
"logits/chosen": 180.29550170898438,
|
| 234 |
+
"logits/rejected": 182.17674255371094,
|
| 235 |
+
"logps/chosen": -1.2322423458099365,
|
| 236 |
+
"logps/rejected": -1.3746764659881592,
|
| 237 |
+
"loss": 46.7575,
|
| 238 |
+
"nll_loss": 1.4799208641052246,
|
| 239 |
+
"rewards/accuracies": 0.53125,
|
| 240 |
+
"rewards/chosen": -0.06161211058497429,
|
| 241 |
+
"rewards/margins": 0.007121713366359472,
|
| 242 |
+
"rewards/rejected": -0.0687338262796402,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 222.3861083984375,
|
| 248 |
+
"learning_rate": 4.7808435099299045e-05,
|
| 249 |
+
"log_odds_chosen": 0.2717851996421814,
|
| 250 |
+
"log_odds_ratio": -0.6738412380218506,
|
| 251 |
+
"logits/chosen": 187.9241943359375,
|
| 252 |
+
"logits/rejected": 175.0797576904297,
|
| 253 |
+
"logps/chosen": -1.1564773321151733,
|
| 254 |
+
"logps/rejected": -1.3804590702056885,
|
| 255 |
+
"loss": 46.2865,
|
| 256 |
+
"nll_loss": 1.4590284824371338,
|
| 257 |
+
"rewards/accuracies": 0.581250011920929,
|
| 258 |
+
"rewards/chosen": -0.05782386660575867,
|
| 259 |
+
"rewards/margins": 0.01119908131659031,
|
| 260 |
+
"rewards/rejected": -0.06902293860912323,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 71.39551544189453,
|
| 266 |
+
"learning_rate": 4.720545159477922e-05,
|
| 267 |
+
"log_odds_chosen": 0.2626160979270935,
|
| 268 |
+
"log_odds_ratio": -0.6702799201011658,
|
| 269 |
+
"logits/chosen": 178.911865234375,
|
| 270 |
+
"logits/rejected": 184.33798217773438,
|
| 271 |
+
"logps/chosen": -1.1193958520889282,
|
| 272 |
+
"logps/rejected": -1.3148605823516846,
|
| 273 |
+
"loss": 45.1045,
|
| 274 |
+
"nll_loss": 1.4315224885940552,
|
| 275 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 276 |
+
"rewards/chosen": -0.05596979334950447,
|
| 277 |
+
"rewards/margins": 0.009773238562047482,
|
| 278 |
+
"rewards/rejected": -0.06574302911758423,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 77.85042572021484,
|
| 284 |
+
"learning_rate": 4.653407456471222e-05,
|
| 285 |
+
"log_odds_chosen": 0.25611892342567444,
|
| 286 |
+
"log_odds_ratio": -0.7023987770080566,
|
| 287 |
+
"logits/chosen": 186.37899780273438,
|
| 288 |
+
"logits/rejected": 181.420654296875,
|
| 289 |
+
"logps/chosen": -1.1519978046417236,
|
| 290 |
+
"logps/rejected": -1.3404672145843506,
|
| 291 |
+
"loss": 44.5742,
|
| 292 |
+
"nll_loss": 1.4464070796966553,
|
| 293 |
+
"rewards/accuracies": 0.59375,
|
| 294 |
+
"rewards/chosen": -0.05759989097714424,
|
| 295 |
+
"rewards/margins": 0.009423469193279743,
|
| 296 |
+
"rewards/rejected": -0.06702335178852081,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 131.22291564941406,
|
| 302 |
+
"learning_rate": 4.579637187256222e-05,
|
| 303 |
+
"log_odds_chosen": 0.21689875423908234,
|
| 304 |
+
"log_odds_ratio": -0.6867625713348389,
|
| 305 |
+
"logits/chosen": 193.07936096191406,
|
| 306 |
+
"logits/rejected": 179.24037170410156,
|
| 307 |
+
"logps/chosen": -1.0784428119659424,
|
| 308 |
+
"logps/rejected": -1.2363873720169067,
|
| 309 |
+
"loss": 44.1766,
|
| 310 |
+
"nll_loss": 1.3691766262054443,
|
| 311 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 312 |
+
"rewards/chosen": -0.053922139108181,
|
| 313 |
+
"rewards/margins": 0.007897228002548218,
|
| 314 |
+
"rewards/rejected": -0.061819374561309814,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 102.12734985351562,
|
| 320 |
+
"learning_rate": 4.499461566702685e-05,
|
| 321 |
+
"log_odds_chosen": 0.28654736280441284,
|
| 322 |
+
"log_odds_ratio": -0.6533454060554504,
|
| 323 |
+
"logits/chosen": 198.9041290283203,
|
| 324 |
+
"logits/rejected": 193.521728515625,
|
| 325 |
+
"logps/chosen": -1.0521138906478882,
|
| 326 |
+
"logps/rejected": -1.2458927631378174,
|
| 327 |
+
"loss": 45.3582,
|
| 328 |
+
"nll_loss": 1.402296781539917,
|
| 329 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 330 |
+
"rewards/chosen": -0.05260569974780083,
|
| 331 |
+
"rewards/margins": 0.00968894176185131,
|
| 332 |
+
"rewards/rejected": -0.06229463964700699,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 66.06409454345703,
|
| 338 |
+
"learning_rate": 4.413127538374411e-05,
|
| 339 |
+
"log_odds_chosen": 0.35550013184547424,
|
| 340 |
+
"log_odds_ratio": -0.6025335192680359,
|
| 341 |
+
"logits/chosen": 186.05125427246094,
|
| 342 |
+
"logits/rejected": 179.599609375,
|
| 343 |
+
"logps/chosen": -0.9773751497268677,
|
| 344 |
+
"logps/rejected": -1.2151715755462646,
|
| 345 |
+
"loss": 43.5128,
|
| 346 |
+
"nll_loss": 1.3024979829788208,
|
| 347 |
+
"rewards/accuracies": 0.668749988079071,
|
| 348 |
+
"rewards/chosen": -0.048868756741285324,
|
| 349 |
+
"rewards/margins": 0.011889824643731117,
|
| 350 |
+
"rewards/rejected": -0.06075857952237129,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 54.939449310302734,
|
| 356 |
+
"learning_rate": 4.320901013934887e-05,
|
| 357 |
+
"log_odds_chosen": 0.1582535207271576,
|
| 358 |
+
"log_odds_ratio": -0.7311251759529114,
|
| 359 |
+
"logits/chosen": 187.7299346923828,
|
| 360 |
+
"logits/rejected": 176.6575927734375,
|
| 361 |
+
"logps/chosen": -1.0948975086212158,
|
| 362 |
+
"logps/rejected": -1.2327477931976318,
|
| 363 |
+
"loss": 43.5056,
|
| 364 |
+
"nll_loss": 1.4032866954803467,
|
| 365 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 366 |
+
"rewards/chosen": -0.05474487692117691,
|
| 367 |
+
"rewards/margins": 0.006892507430166006,
|
| 368 |
+
"rewards/rejected": -0.061637382954359055,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 107.38319396972656,
|
| 374 |
+
"learning_rate": 4.223066054130568e-05,
|
| 375 |
+
"log_odds_chosen": 0.27941060066223145,
|
| 376 |
+
"log_odds_ratio": -0.6358767151832581,
|
| 377 |
+
"logits/chosen": 184.79893493652344,
|
| 378 |
+
"logits/rejected": 176.64239501953125,
|
| 379 |
+
"logps/chosen": -1.011204481124878,
|
| 380 |
+
"logps/rejected": -1.2123634815216064,
|
| 381 |
+
"loss": 42.5508,
|
| 382 |
+
"nll_loss": 1.3267438411712646,
|
| 383 |
+
"rewards/accuracies": 0.625,
|
| 384 |
+
"rewards/chosen": -0.050560224801301956,
|
| 385 |
+
"rewards/margins": 0.010057945735752583,
|
| 386 |
+
"rewards/rejected": -0.06061817333102226,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.037925925925926,
|
| 391 |
+
"grad_norm": 69.81970977783203,
|
| 392 |
+
"learning_rate": 4.1199239938743797e-05,
|
| 393 |
+
"log_odds_chosen": 0.506846010684967,
|
| 394 |
+
"log_odds_ratio": -0.5906091332435608,
|
| 395 |
+
"logits/chosen": 173.7986297607422,
|
| 396 |
+
"logits/rejected": 167.03512573242188,
|
| 397 |
+
"logps/chosen": -0.878280758857727,
|
| 398 |
+
"logps/rejected": -1.224012851715088,
|
| 399 |
+
"loss": 33.7043,
|
| 400 |
+
"nll_loss": 1.1471017599105835,
|
| 401 |
+
"rewards/accuracies": 0.6643356680870056,
|
| 402 |
+
"rewards/chosen": -0.04391403868794441,
|
| 403 |
+
"rewards/margins": 0.01728660613298416,
|
| 404 |
+
"rewards/rejected": -0.061200641095638275,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0853333333333333,
|
| 409 |
+
"grad_norm": 69.83367156982422,
|
| 410 |
+
"learning_rate": 4.0117925141242174e-05,
|
| 411 |
+
"log_odds_chosen": 0.726813018321991,
|
| 412 |
+
"log_odds_ratio": -0.48568326234817505,
|
| 413 |
+
"logits/chosen": 172.2089385986328,
|
| 414 |
+
"logits/rejected": 163.337646484375,
|
| 415 |
+
"logps/chosen": -0.841424286365509,
|
| 416 |
+
"logps/rejected": -1.3027280569076538,
|
| 417 |
+
"loss": 36.2183,
|
| 418 |
+
"nll_loss": 1.106263279914856,
|
| 419 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 420 |
+
"rewards/chosen": -0.04207121208310127,
|
| 421 |
+
"rewards/margins": 0.023065198212862015,
|
| 422 |
+
"rewards/rejected": -0.06513641774654388,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1327407407407408,
|
| 427 |
+
"grad_norm": 99.912841796875,
|
| 428 |
+
"learning_rate": 3.899004663415084e-05,
|
| 429 |
+
"log_odds_chosen": 0.6838423013687134,
|
| 430 |
+
"log_odds_ratio": -0.5168099403381348,
|
| 431 |
+
"logits/chosen": 165.5643310546875,
|
| 432 |
+
"logits/rejected": 164.38926696777344,
|
| 433 |
+
"logps/chosen": -0.8310489654541016,
|
| 434 |
+
"logps/rejected": -1.2339271306991577,
|
| 435 |
+
"loss": 36.5605,
|
| 436 |
+
"nll_loss": 1.1273092031478882,
|
| 437 |
+
"rewards/accuracies": 0.75,
|
| 438 |
+
"rewards/chosen": -0.04155245050787926,
|
| 439 |
+
"rewards/margins": 0.020143909379839897,
|
| 440 |
+
"rewards/rejected": -0.061696358025074005,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1801481481481482,
|
| 445 |
+
"grad_norm": 73.86228942871094,
|
| 446 |
+
"learning_rate": 3.781907832058587e-05,
|
| 447 |
+
"log_odds_chosen": 0.6109746694564819,
|
| 448 |
+
"log_odds_ratio": -0.5379955172538757,
|
| 449 |
+
"logits/chosen": 163.86184692382812,
|
| 450 |
+
"logits/rejected": 161.99966430664062,
|
| 451 |
+
"logps/chosen": -0.8782113790512085,
|
| 452 |
+
"logps/rejected": -1.2740932703018188,
|
| 453 |
+
"loss": 34.8236,
|
| 454 |
+
"nll_loss": 1.1110111474990845,
|
| 455 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 456 |
+
"rewards/chosen": -0.043910570442676544,
|
| 457 |
+
"rewards/margins": 0.019794095307588577,
|
| 458 |
+
"rewards/rejected": -0.06370466202497482,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2275555555555555,
|
| 463 |
+
"grad_norm": 105.75894165039062,
|
| 464 |
+
"learning_rate": 3.660862682169282e-05,
|
| 465 |
+
"log_odds_chosen": 0.5121253728866577,
|
| 466 |
+
"log_odds_ratio": -0.5994678735733032,
|
| 467 |
+
"logits/chosen": 166.21932983398438,
|
| 468 |
+
"logits/rejected": 163.54107666015625,
|
| 469 |
+
"logps/chosen": -0.8976671099662781,
|
| 470 |
+
"logps/rejected": -1.1910202503204346,
|
| 471 |
+
"loss": 34.8684,
|
| 472 |
+
"nll_loss": 1.1224849224090576,
|
| 473 |
+
"rewards/accuracies": 0.706250011920929,
|
| 474 |
+
"rewards/chosen": -0.0448833629488945,
|
| 475 |
+
"rewards/margins": 0.01466765720397234,
|
| 476 |
+
"rewards/rejected": -0.059551019221544266,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.274962962962963,
|
| 481 |
+
"grad_norm": 97.21427154541016,
|
| 482 |
+
"learning_rate": 3.5362420368134356e-05,
|
| 483 |
+
"log_odds_chosen": 0.5886248350143433,
|
| 484 |
+
"log_odds_ratio": -0.5262036323547363,
|
| 485 |
+
"logits/chosen": 165.6793670654297,
|
| 486 |
+
"logits/rejected": 166.072509765625,
|
| 487 |
+
"logps/chosen": -0.7888522148132324,
|
| 488 |
+
"logps/rejected": -1.1330220699310303,
|
| 489 |
+
"loss": 35.0844,
|
| 490 |
+
"nll_loss": 1.073457956314087,
|
| 491 |
+
"rewards/accuracies": 0.731249988079071,
|
| 492 |
+
"rewards/chosen": -0.03944260999560356,
|
| 493 |
+
"rewards/margins": 0.017208490520715714,
|
| 494 |
+
"rewards/rejected": -0.056651096791028976,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3223703703703704,
|
| 499 |
+
"grad_norm": 49.054542541503906,
|
| 500 |
+
"learning_rate": 3.408429731701635e-05,
|
| 501 |
+
"log_odds_chosen": 0.6223124265670776,
|
| 502 |
+
"log_odds_ratio": -0.554074764251709,
|
| 503 |
+
"logits/chosen": 161.89114379882812,
|
| 504 |
+
"logits/rejected": 166.47528076171875,
|
| 505 |
+
"logps/chosen": -0.8866279721260071,
|
| 506 |
+
"logps/rejected": -1.2842514514923096,
|
| 507 |
+
"loss": 36.4437,
|
| 508 |
+
"nll_loss": 1.141160011291504,
|
| 509 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 510 |
+
"rewards/chosen": -0.04433140158653259,
|
| 511 |
+
"rewards/margins": 0.019881170243024826,
|
| 512 |
+
"rewards/rejected": -0.06421257555484772,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.3697777777777778,
|
| 517 |
+
"grad_norm": 41.25300216674805,
|
| 518 |
+
"learning_rate": 3.2778194329621104e-05,
|
| 519 |
+
"log_odds_chosen": 0.6255282163619995,
|
| 520 |
+
"log_odds_ratio": -0.5217418670654297,
|
| 521 |
+
"logits/chosen": 167.75601196289062,
|
| 522 |
+
"logits/rejected": 155.63418579101562,
|
| 523 |
+
"logps/chosen": -0.8616455793380737,
|
| 524 |
+
"logps/rejected": -1.2749989032745361,
|
| 525 |
+
"loss": 35.9371,
|
| 526 |
+
"nll_loss": 1.1240451335906982,
|
| 527 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 528 |
+
"rewards/chosen": -0.043082281947135925,
|
| 529 |
+
"rewards/margins": 0.02066767029464245,
|
| 530 |
+
"rewards/rejected": -0.06374995410442352,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.417185185185185,
|
| 535 |
+
"grad_norm": 43.59724426269531,
|
| 536 |
+
"learning_rate": 3.144813424636031e-05,
|
| 537 |
+
"log_odds_chosen": 0.5113286972045898,
|
| 538 |
+
"log_odds_ratio": -0.5680266618728638,
|
| 539 |
+
"logits/chosen": 165.27484130859375,
|
| 540 |
+
"logits/rejected": 169.46823120117188,
|
| 541 |
+
"logps/chosen": -0.8289966583251953,
|
| 542 |
+
"logps/rejected": -1.1291935443878174,
|
| 543 |
+
"loss": 34.8101,
|
| 544 |
+
"nll_loss": 1.084172010421753,
|
| 545 |
+
"rewards/accuracies": 0.699999988079071,
|
| 546 |
+
"rewards/chosen": -0.041449833661317825,
|
| 547 |
+
"rewards/margins": 0.01500984001904726,
|
| 548 |
+
"rewards/rejected": -0.05645967274904251,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4645925925925927,
|
| 553 |
+
"grad_norm": 42.254207611083984,
|
| 554 |
+
"learning_rate": 3.0098213696293542e-05,
|
| 555 |
+
"log_odds_chosen": 0.5680588483810425,
|
| 556 |
+
"log_odds_ratio": -0.5456587672233582,
|
| 557 |
+
"logits/chosen": 173.2566375732422,
|
| 558 |
+
"logits/rejected": 164.67221069335938,
|
| 559 |
+
"logps/chosen": -0.8088580369949341,
|
| 560 |
+
"logps/rejected": -1.1407465934753418,
|
| 561 |
+
"loss": 35.4576,
|
| 562 |
+
"nll_loss": 1.1147387027740479,
|
| 563 |
+
"rewards/accuracies": 0.699999988079071,
|
| 564 |
+
"rewards/chosen": -0.040442898869514465,
|
| 565 |
+
"rewards/margins": 0.016594436019659042,
|
| 566 |
+
"rewards/rejected": -0.057037342339754105,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.512,
|
| 571 |
+
"grad_norm": 45.57976531982422,
|
| 572 |
+
"learning_rate": 2.8732590479375165e-05,
|
| 573 |
+
"log_odds_chosen": 0.540373682975769,
|
| 574 |
+
"log_odds_ratio": -0.543350338935852,
|
| 575 |
+
"logits/chosen": 172.65396118164062,
|
| 576 |
+
"logits/rejected": 173.95144653320312,
|
| 577 |
+
"logps/chosen": -0.910565972328186,
|
| 578 |
+
"logps/rejected": -1.2549046277999878,
|
| 579 |
+
"loss": 35.9476,
|
| 580 |
+
"nll_loss": 1.1549952030181885,
|
| 581 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 582 |
+
"rewards/chosen": -0.045528292655944824,
|
| 583 |
+
"rewards/margins": 0.017216943204402924,
|
| 584 |
+
"rewards/rejected": -0.06274524331092834,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5594074074074074,
|
| 589 |
+
"grad_norm": 54.967220306396484,
|
| 590 |
+
"learning_rate": 2.7355470760292956e-05,
|
| 591 |
+
"log_odds_chosen": 0.6586408019065857,
|
| 592 |
+
"log_odds_ratio": -0.5135148763656616,
|
| 593 |
+
"logits/chosen": 164.9529266357422,
|
| 594 |
+
"logits/rejected": 176.58729553222656,
|
| 595 |
+
"logps/chosen": -0.8376239538192749,
|
| 596 |
+
"logps/rejected": -1.241012454032898,
|
| 597 |
+
"loss": 36.1099,
|
| 598 |
+
"nll_loss": 1.1062064170837402,
|
| 599 |
+
"rewards/accuracies": 0.71875,
|
| 600 |
+
"rewards/chosen": -0.041881196200847626,
|
| 601 |
+
"rewards/margins": 0.020169423893094063,
|
| 602 |
+
"rewards/rejected": -0.062050629407167435,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6068148148148147,
|
| 607 |
+
"grad_norm": 48.87372970581055,
|
| 608 |
+
"learning_rate": 2.597109611334169e-05,
|
| 609 |
+
"log_odds_chosen": 0.6633528470993042,
|
| 610 |
+
"log_odds_ratio": -0.5211383104324341,
|
| 611 |
+
"logits/chosen": 169.59494018554688,
|
| 612 |
+
"logits/rejected": 169.11012268066406,
|
| 613 |
+
"logps/chosen": -0.8359187245368958,
|
| 614 |
+
"logps/rejected": -1.2383381128311157,
|
| 615 |
+
"loss": 35.4707,
|
| 616 |
+
"nll_loss": 1.0976989269256592,
|
| 617 |
+
"rewards/accuracies": 0.762499988079071,
|
| 618 |
+
"rewards/chosen": -0.041795939207077026,
|
| 619 |
+
"rewards/margins": 0.02012096717953682,
|
| 620 |
+
"rewards/rejected": -0.06191689893603325,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6542222222222223,
|
| 625 |
+
"grad_norm": 38.77116012573242,
|
| 626 |
+
"learning_rate": 2.458373045823404e-05,
|
| 627 |
+
"log_odds_chosen": 0.6313592195510864,
|
| 628 |
+
"log_odds_ratio": -0.519364058971405,
|
| 629 |
+
"logits/chosen": 174.59616088867188,
|
| 630 |
+
"logits/rejected": 163.9171600341797,
|
| 631 |
+
"logps/chosen": -0.826526939868927,
|
| 632 |
+
"logps/rejected": -1.2227591276168823,
|
| 633 |
+
"loss": 35.0289,
|
| 634 |
+
"nll_loss": 1.1160304546356201,
|
| 635 |
+
"rewards/accuracies": 0.762499988079071,
|
| 636 |
+
"rewards/chosen": -0.04132634773850441,
|
| 637 |
+
"rewards/margins": 0.019811619073152542,
|
| 638 |
+
"rewards/rejected": -0.06113796308636665,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7016296296296296,
|
| 643 |
+
"grad_norm": 38.79117965698242,
|
| 644 |
+
"learning_rate": 2.3197646927086697e-05,
|
| 645 |
+
"log_odds_chosen": 0.3873888850212097,
|
| 646 |
+
"log_odds_ratio": -0.6197336912155151,
|
| 647 |
+
"logits/chosen": 164.18601989746094,
|
| 648 |
+
"logits/rejected": 166.98605346679688,
|
| 649 |
+
"logps/chosen": -0.8658819198608398,
|
| 650 |
+
"logps/rejected": -1.0931422710418701,
|
| 651 |
+
"loss": 35.4427,
|
| 652 |
+
"nll_loss": 1.1255178451538086,
|
| 653 |
+
"rewards/accuracies": 0.643750011920929,
|
| 654 |
+
"rewards/chosen": -0.043294090777635574,
|
| 655 |
+
"rewards/margins": 0.011363019235432148,
|
| 656 |
+
"rewards/rejected": -0.054657112807035446,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.749037037037037,
|
| 661 |
+
"grad_norm": 41.68294906616211,
|
| 662 |
+
"learning_rate": 2.1817114703032176e-05,
|
| 663 |
+
"log_odds_chosen": 0.5725528001785278,
|
| 664 |
+
"log_odds_ratio": -0.5303586721420288,
|
| 665 |
+
"logits/chosen": 168.3577880859375,
|
| 666 |
+
"logits/rejected": 167.2732391357422,
|
| 667 |
+
"logps/chosen": -0.8526233434677124,
|
| 668 |
+
"logps/rejected": -1.2170125246047974,
|
| 669 |
+
"loss": 35.5527,
|
| 670 |
+
"nll_loss": 1.1156741380691528,
|
| 671 |
+
"rewards/accuracies": 0.706250011920929,
|
| 672 |
+
"rewards/chosen": -0.04263117164373398,
|
| 673 |
+
"rewards/margins": 0.01821945607662201,
|
| 674 |
+
"rewards/rejected": -0.06085062772035599,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.7964444444444445,
|
| 679 |
+
"grad_norm": 57.764461517333984,
|
| 680 |
+
"learning_rate": 2.0446385870993467e-05,
|
| 681 |
+
"log_odds_chosen": 0.6254726648330688,
|
| 682 |
+
"log_odds_ratio": -0.5300140976905823,
|
| 683 |
+
"logits/chosen": 176.73265075683594,
|
| 684 |
+
"logits/rejected": 166.2118377685547,
|
| 685 |
+
"logps/chosen": -0.8647019267082214,
|
| 686 |
+
"logps/rejected": -1.2612402439117432,
|
| 687 |
+
"loss": 35.7179,
|
| 688 |
+
"nll_loss": 1.1093218326568604,
|
| 689 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 690 |
+
"rewards/chosen": -0.04323510080575943,
|
| 691 |
+
"rewards/margins": 0.019826913252472878,
|
| 692 |
+
"rewards/rejected": -0.06306201219558716,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8438518518518519,
|
| 697 |
+
"grad_norm": 57.696022033691406,
|
| 698 |
+
"learning_rate": 1.9089682321121834e-05,
|
| 699 |
+
"log_odds_chosen": 0.7172800302505493,
|
| 700 |
+
"log_odds_ratio": -0.5064095258712769,
|
| 701 |
+
"logits/chosen": 168.99301147460938,
|
| 702 |
+
"logits/rejected": 158.69894409179688,
|
| 703 |
+
"logps/chosen": -0.8419252634048462,
|
| 704 |
+
"logps/rejected": -1.2953702211380005,
|
| 705 |
+
"loss": 35.0776,
|
| 706 |
+
"nll_loss": 1.1046844720840454,
|
| 707 |
+
"rewards/accuracies": 0.75,
|
| 708 |
+
"rewards/chosen": -0.04209625720977783,
|
| 709 |
+
"rewards/margins": 0.022672254592180252,
|
| 710 |
+
"rewards/rejected": -0.06476850807666779,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8912592592592592,
|
| 715 |
+
"grad_norm": 38.26820373535156,
|
| 716 |
+
"learning_rate": 1.775118274523545e-05,
|
| 717 |
+
"log_odds_chosen": 0.5704585313796997,
|
| 718 |
+
"log_odds_ratio": -0.5714629888534546,
|
| 719 |
+
"logits/chosen": 170.9229736328125,
|
| 720 |
+
"logits/rejected": 165.9208526611328,
|
| 721 |
+
"logps/chosen": -0.8371521830558777,
|
| 722 |
+
"logps/rejected": -1.18575119972229,
|
| 723 |
+
"loss": 35.9511,
|
| 724 |
+
"nll_loss": 1.0667616128921509,
|
| 725 |
+
"rewards/accuracies": 0.6875,
|
| 726 |
+
"rewards/chosen": -0.041857607662677765,
|
| 727 |
+
"rewards/margins": 0.017429957166314125,
|
| 728 |
+
"rewards/rejected": -0.05928756669163704,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9386666666666668,
|
| 733 |
+
"grad_norm": 37.01579666137695,
|
| 734 |
+
"learning_rate": 1.643500976631037e-05,
|
| 735 |
+
"log_odds_chosen": 0.5442375540733337,
|
| 736 |
+
"log_odds_ratio": -0.5546382665634155,
|
| 737 |
+
"logits/chosen": 175.0882110595703,
|
| 738 |
+
"logits/rejected": 163.4840850830078,
|
| 739 |
+
"logps/chosen": -0.8371946215629578,
|
| 740 |
+
"logps/rejected": -1.1808607578277588,
|
| 741 |
+
"loss": 34.5686,
|
| 742 |
+
"nll_loss": 1.0792796611785889,
|
| 743 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 744 |
+
"rewards/chosen": -0.04185972735285759,
|
| 745 |
+
"rewards/margins": 0.01718330569565296,
|
| 746 |
+
"rewards/rejected": -0.0590430311858654,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.986074074074074,
|
| 751 |
+
"grad_norm": 58.07009506225586,
|
| 752 |
+
"learning_rate": 1.514521724066537e-05,
|
| 753 |
+
"log_odds_chosen": 0.5513701438903809,
|
| 754 |
+
"log_odds_ratio": -0.5487216114997864,
|
| 755 |
+
"logits/chosen": 160.052734375,
|
| 756 |
+
"logits/rejected": 165.99490356445312,
|
| 757 |
+
"logps/chosen": -0.804582953453064,
|
| 758 |
+
"logps/rejected": -1.125427484512329,
|
| 759 |
+
"loss": 33.0603,
|
| 760 |
+
"nll_loss": 1.0380481481552124,
|
| 761 |
+
"rewards/accuracies": 0.706250011920929,
|
| 762 |
+
"rewards/chosen": -0.04022914916276932,
|
| 763 |
+
"rewards/margins": 0.01604221947491169,
|
| 764 |
+
"rewards/rejected": -0.056271374225616455,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0284444444444443,
|
| 769 |
+
"grad_norm": 46.680702209472656,
|
| 770 |
+
"learning_rate": 1.3885777771950348e-05,
|
| 771 |
+
"log_odds_chosen": 0.9105411171913147,
|
| 772 |
+
"log_odds_ratio": -0.4484807252883911,
|
| 773 |
+
"logits/chosen": 155.98167419433594,
|
| 774 |
+
"logits/rejected": 156.97645568847656,
|
| 775 |
+
"logps/chosen": -0.7199848890304565,
|
| 776 |
+
"logps/rejected": -1.2084262371063232,
|
| 777 |
+
"loss": 27.155,
|
| 778 |
+
"nll_loss": 0.9470744132995605,
|
| 779 |
+
"rewards/accuracies": 0.8251748085021973,
|
| 780 |
+
"rewards/chosen": -0.03599924594163895,
|
| 781 |
+
"rewards/margins": 0.024422070011496544,
|
| 782 |
+
"rewards/rejected": -0.06042131036520004,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.075851851851852,
|
| 787 |
+
"grad_norm": 47.282562255859375,
|
| 788 |
+
"learning_rate": 1.2660570475395683e-05,
|
| 789 |
+
"log_odds_chosen": 1.1580384969711304,
|
| 790 |
+
"log_odds_ratio": -0.39033469557762146,
|
| 791 |
+
"logits/chosen": 142.10780334472656,
|
| 792 |
+
"logits/rejected": 143.4950408935547,
|
| 793 |
+
"logps/chosen": -0.6312204599380493,
|
| 794 |
+
"logps/rejected": -1.257129192352295,
|
| 795 |
+
"loss": 26.9195,
|
| 796 |
+
"nll_loss": 0.8665180206298828,
|
| 797 |
+
"rewards/accuracies": 0.824999988079071,
|
| 798 |
+
"rewards/chosen": -0.031561024487018585,
|
| 799 |
+
"rewards/margins": 0.03129544109106064,
|
| 800 |
+
"rewards/rejected": -0.06285646557807922,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1232592592592594,
|
| 805 |
+
"grad_norm": 44.752628326416016,
|
| 806 |
+
"learning_rate": 1.1473369030008974e-05,
|
| 807 |
+
"log_odds_chosen": 1.052072286605835,
|
| 808 |
+
"log_odds_ratio": -0.412436306476593,
|
| 809 |
+
"logits/chosen": 148.757568359375,
|
| 810 |
+
"logits/rejected": 149.47320556640625,
|
| 811 |
+
"logps/chosen": -0.6186680793762207,
|
| 812 |
+
"logps/rejected": -1.1820106506347656,
|
| 813 |
+
"loss": 28.038,
|
| 814 |
+
"nll_loss": 0.87525475025177,
|
| 815 |
+
"rewards/accuracies": 0.84375,
|
| 816 |
+
"rewards/chosen": -0.030933406203985214,
|
| 817 |
+
"rewards/margins": 0.0281671229749918,
|
| 818 |
+
"rewards/rejected": -0.05910053104162216,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.1706666666666665,
|
| 823 |
+
"grad_norm": 45.039642333984375,
|
| 824 |
+
"learning_rate": 1.0327830055518842e-05,
|
| 825 |
+
"log_odds_chosen": 1.1950232982635498,
|
| 826 |
+
"log_odds_ratio": -0.3665129542350769,
|
| 827 |
+
"logits/chosen": 148.42233276367188,
|
| 828 |
+
"logits/rejected": 144.0989227294922,
|
| 829 |
+
"logps/chosen": -0.6338512897491455,
|
| 830 |
+
"logps/rejected": -1.2817251682281494,
|
| 831 |
+
"loss": 26.4916,
|
| 832 |
+
"nll_loss": 0.8120902180671692,
|
| 833 |
+
"rewards/accuracies": 0.84375,
|
| 834 |
+
"rewards/chosen": -0.031692564487457275,
|
| 835 |
+
"rewards/margins": 0.03239370137453079,
|
| 836 |
+
"rewards/rejected": -0.06408625841140747,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.218074074074074,
|
| 841 |
+
"grad_norm": 44.143611907958984,
|
| 842 |
+
"learning_rate": 9.227481849865235e-06,
|
| 843 |
+
"log_odds_chosen": 1.0343241691589355,
|
| 844 |
+
"log_odds_ratio": -0.41224583983421326,
|
| 845 |
+
"logits/chosen": 142.4004364013672,
|
| 846 |
+
"logits/rejected": 147.73538208007812,
|
| 847 |
+
"logps/chosen": -0.6527639627456665,
|
| 848 |
+
"logps/rejected": -1.1827428340911865,
|
| 849 |
+
"loss": 27.6183,
|
| 850 |
+
"nll_loss": 0.8605834245681763,
|
| 851 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 852 |
+
"rewards/chosen": -0.032638195902109146,
|
| 853 |
+
"rewards/margins": 0.026498943567276,
|
| 854 |
+
"rewards/rejected": -0.059137143194675446,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2654814814814817,
|
| 859 |
+
"grad_norm": 41.318138122558594,
|
| 860 |
+
"learning_rate": 8.175713521924978e-06,
|
| 861 |
+
"log_odds_chosen": 1.1422860622406006,
|
| 862 |
+
"log_odds_ratio": -0.3735355734825134,
|
| 863 |
+
"logits/chosen": 145.45065307617188,
|
| 864 |
+
"logits/rejected": 151.5826416015625,
|
| 865 |
+
"logps/chosen": -0.6259297132492065,
|
| 866 |
+
"logps/rejected": -1.2332427501678467,
|
| 867 |
+
"loss": 27.1915,
|
| 868 |
+
"nll_loss": 0.8626583218574524,
|
| 869 |
+
"rewards/accuracies": 0.862500011920929,
|
| 870 |
+
"rewards/chosen": -0.03129648417234421,
|
| 871 |
+
"rewards/margins": 0.030365657061338425,
|
| 872 |
+
"rewards/rejected": -0.06166214868426323,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.3128888888888888,
|
| 877 |
+
"grad_norm": 44.362037658691406,
|
| 878 |
+
"learning_rate": 7.1757645529443665e-06,
|
| 879 |
+
"log_odds_chosen": 1.273262858390808,
|
| 880 |
+
"log_odds_ratio": -0.35038676857948303,
|
| 881 |
+
"logits/chosen": 142.28067016601562,
|
| 882 |
+
"logits/rejected": 141.07127380371094,
|
| 883 |
+
"logps/chosen": -0.6050504446029663,
|
| 884 |
+
"logps/rejected": -1.2771079540252686,
|
| 885 |
+
"loss": 26.2994,
|
| 886 |
+
"nll_loss": 0.849342942237854,
|
| 887 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 888 |
+
"rewards/chosen": -0.030252521857619286,
|
| 889 |
+
"rewards/margins": 0.03360287845134735,
|
| 890 |
+
"rewards/rejected": -0.06385539472103119,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3602962962962963,
|
| 895 |
+
"grad_norm": 37.62168884277344,
|
| 896 |
+
"learning_rate": 6.230714818829733e-06,
|
| 897 |
+
"log_odds_chosen": 1.244945764541626,
|
| 898 |
+
"log_odds_ratio": -0.36415261030197144,
|
| 899 |
+
"logits/chosen": 138.27842712402344,
|
| 900 |
+
"logits/rejected": 143.33724975585938,
|
| 901 |
+
"logps/chosen": -0.6403986811637878,
|
| 902 |
+
"logps/rejected": -1.302079200744629,
|
| 903 |
+
"loss": 26.4474,
|
| 904 |
+
"nll_loss": 0.8333737254142761,
|
| 905 |
+
"rewards/accuracies": 0.831250011920929,
|
| 906 |
+
"rewards/chosen": -0.03201993554830551,
|
| 907 |
+
"rewards/margins": 0.03308403491973877,
|
| 908 |
+
"rewards/rejected": -0.06510396301746368,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.407703703703704,
|
| 913 |
+
"grad_norm": 40.77067947387695,
|
| 914 |
+
"learning_rate": 5.343475104027743e-06,
|
| 915 |
+
"log_odds_chosen": 1.3023749589920044,
|
| 916 |
+
"log_odds_ratio": -0.3504185378551483,
|
| 917 |
+
"logits/chosen": 136.38796997070312,
|
| 918 |
+
"logits/rejected": 143.35763549804688,
|
| 919 |
+
"logps/chosen": -0.5728102326393127,
|
| 920 |
+
"logps/rejected": -1.279425859451294,
|
| 921 |
+
"loss": 26.3726,
|
| 922 |
+
"nll_loss": 0.8203259706497192,
|
| 923 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 924 |
+
"rewards/chosen": -0.028640514239668846,
|
| 925 |
+
"rewards/margins": 0.03533077985048294,
|
| 926 |
+
"rewards/rejected": -0.06397129595279694,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.455111111111111,
|
| 931 |
+
"grad_norm": 45.096004486083984,
|
| 932 |
+
"learning_rate": 4.516778136213037e-06,
|
| 933 |
+
"log_odds_chosen": 1.3704315423965454,
|
| 934 |
+
"log_odds_ratio": -0.3305371403694153,
|
| 935 |
+
"logits/chosen": 137.95346069335938,
|
| 936 |
+
"logits/rejected": 139.55905151367188,
|
| 937 |
+
"logps/chosen": -0.591722309589386,
|
| 938 |
+
"logps/rejected": -1.3149653673171997,
|
| 939 |
+
"loss": 26.1383,
|
| 940 |
+
"nll_loss": 0.8089566230773926,
|
| 941 |
+
"rewards/accuracies": 0.856249988079071,
|
| 942 |
+
"rewards/chosen": -0.02958611212670803,
|
| 943 |
+
"rewards/margins": 0.036162156611680984,
|
| 944 |
+
"rewards/rejected": -0.06574826687574387,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5025185185185186,
|
| 949 |
+
"grad_norm": 47.061092376708984,
|
| 950 |
+
"learning_rate": 3.7531701693965554e-06,
|
| 951 |
+
"log_odds_chosen": 1.1507422924041748,
|
| 952 |
+
"log_odds_ratio": -0.408168226480484,
|
| 953 |
+
"logits/chosen": 135.37014770507812,
|
| 954 |
+
"logits/rejected": 141.3232421875,
|
| 955 |
+
"logps/chosen": -0.6197280883789062,
|
| 956 |
+
"logps/rejected": -1.2372686862945557,
|
| 957 |
+
"loss": 25.815,
|
| 958 |
+
"nll_loss": 0.8066200017929077,
|
| 959 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 960 |
+
"rewards/chosen": -0.030986404046416283,
|
| 961 |
+
"rewards/margins": 0.03087702952325344,
|
| 962 |
+
"rewards/rejected": -0.061863433569669724,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.549925925925926,
|
| 967 |
+
"grad_norm": 44.70289611816406,
|
| 968 |
+
"learning_rate": 3.055003141378948e-06,
|
| 969 |
+
"log_odds_chosen": 1.2243850231170654,
|
| 970 |
+
"log_odds_ratio": -0.36160460114479065,
|
| 971 |
+
"logits/chosen": 140.19070434570312,
|
| 972 |
+
"logits/rejected": 143.959716796875,
|
| 973 |
+
"logps/chosen": -0.632489025592804,
|
| 974 |
+
"logps/rejected": -1.2881219387054443,
|
| 975 |
+
"loss": 25.857,
|
| 976 |
+
"nll_loss": 0.8426779508590698,
|
| 977 |
+
"rewards/accuracies": 0.856249988079071,
|
| 978 |
+
"rewards/chosen": -0.031624455004930496,
|
| 979 |
+
"rewards/margins": 0.03278164193034172,
|
| 980 |
+
"rewards/rejected": -0.06440608948469162,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.5973333333333333,
|
| 985 |
+
"grad_norm": 40.76127243041992,
|
| 986 |
+
"learning_rate": 2.424427429704365e-06,
|
| 987 |
+
"log_odds_chosen": 1.3216700553894043,
|
| 988 |
+
"log_odds_ratio": -0.3369961977005005,
|
| 989 |
+
"logits/chosen": 138.0738067626953,
|
| 990 |
+
"logits/rejected": 139.0469970703125,
|
| 991 |
+
"logps/chosen": -0.6244256496429443,
|
| 992 |
+
"logps/rejected": -1.3335908651351929,
|
| 993 |
+
"loss": 27.0255,
|
| 994 |
+
"nll_loss": 0.8462891578674316,
|
| 995 |
+
"rewards/accuracies": 0.893750011920929,
|
| 996 |
+
"rewards/chosen": -0.031221281737089157,
|
| 997 |
+
"rewards/margins": 0.035458266735076904,
|
| 998 |
+
"rewards/rejected": -0.06667955219745636,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.644740740740741,
|
| 1003 |
+
"grad_norm": 42.352970123291016,
|
| 1004 |
+
"learning_rate": 1.8633852284264508e-06,
|
| 1005 |
+
"log_odds_chosen": 1.190826654434204,
|
| 1006 |
+
"log_odds_ratio": -0.3770466446876526,
|
| 1007 |
+
"logits/chosen": 140.8355255126953,
|
| 1008 |
+
"logits/rejected": 142.74783325195312,
|
| 1009 |
+
"logps/chosen": -0.6085912585258484,
|
| 1010 |
+
"logps/rejected": -1.2392776012420654,
|
| 1011 |
+
"loss": 25.9705,
|
| 1012 |
+
"nll_loss": 0.7998742461204529,
|
| 1013 |
+
"rewards/accuracies": 0.893750011920929,
|
| 1014 |
+
"rewards/chosen": -0.03042956255376339,
|
| 1015 |
+
"rewards/margins": 0.031534306704998016,
|
| 1016 |
+
"rewards/rejected": -0.061963874846696854,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.6921481481481484,
|
| 1021 |
+
"grad_norm": 40.149749755859375,
|
| 1022 |
+
"learning_rate": 1.3736045660864034e-06,
|
| 1023 |
+
"log_odds_chosen": 1.1410518884658813,
|
| 1024 |
+
"log_odds_ratio": -0.3837069869041443,
|
| 1025 |
+
"logits/chosen": 147.14105224609375,
|
| 1026 |
+
"logits/rejected": 141.97027587890625,
|
| 1027 |
+
"logps/chosen": -0.6317769289016724,
|
| 1028 |
+
"logps/rejected": -1.2591302394866943,
|
| 1029 |
+
"loss": 26.657,
|
| 1030 |
+
"nll_loss": 0.8469123840332031,
|
| 1031 |
+
"rewards/accuracies": 0.84375,
|
| 1032 |
+
"rewards/chosen": -0.0315888449549675,
|
| 1033 |
+
"rewards/margins": 0.03136766701936722,
|
| 1034 |
+
"rewards/rejected": -0.06295650452375412,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.7395555555555555,
|
| 1039 |
+
"grad_norm": 60.75810623168945,
|
| 1040 |
+
"learning_rate": 9.565939833279192e-07,
|
| 1041 |
+
"log_odds_chosen": 1.31714928150177,
|
| 1042 |
+
"log_odds_ratio": -0.3334599435329437,
|
| 1043 |
+
"logits/chosen": 140.9696807861328,
|
| 1044 |
+
"logits/rejected": 147.40383911132812,
|
| 1045 |
+
"logps/chosen": -0.5831328630447388,
|
| 1046 |
+
"logps/rejected": -1.263795018196106,
|
| 1047 |
+
"loss": 27.04,
|
| 1048 |
+
"nll_loss": 0.8344683647155762,
|
| 1049 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1050 |
+
"rewards/chosen": -0.029156645759940147,
|
| 1051 |
+
"rewards/margins": 0.03403311222791672,
|
| 1052 |
+
"rewards/rejected": -0.06318975985050201,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.786962962962963,
|
| 1057 |
+
"grad_norm": 43.854427337646484,
|
| 1058 |
+
"learning_rate": 6.136378865420872e-07,
|
| 1059 |
+
"log_odds_chosen": 1.1479493379592896,
|
| 1060 |
+
"log_odds_ratio": -0.4157779812812805,
|
| 1061 |
+
"logits/chosen": 139.2587890625,
|
| 1062 |
+
"logits/rejected": 147.88906860351562,
|
| 1063 |
+
"logps/chosen": -0.6736854314804077,
|
| 1064 |
+
"logps/rejected": -1.3039884567260742,
|
| 1065 |
+
"loss": 26.5112,
|
| 1066 |
+
"nll_loss": 0.8539054989814758,
|
| 1067 |
+
"rewards/accuracies": 0.84375,
|
| 1068 |
+
"rewards/chosen": -0.033684272319078445,
|
| 1069 |
+
"rewards/margins": 0.03151514753699303,
|
| 1070 |
+
"rewards/rejected": -0.06519941985607147,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.83437037037037,
|
| 1075 |
+
"grad_norm": 52.22172164916992,
|
| 1076 |
+
"learning_rate": 3.45792591853214e-07,
|
| 1077 |
+
"log_odds_chosen": 1.2029494047164917,
|
| 1078 |
+
"log_odds_ratio": -0.38027894496917725,
|
| 1079 |
+
"logits/chosen": 151.04905700683594,
|
| 1080 |
+
"logits/rejected": 142.887939453125,
|
| 1081 |
+
"logps/chosen": -0.6414423584938049,
|
| 1082 |
+
"logps/rejected": -1.3054555654525757,
|
| 1083 |
+
"loss": 26.8389,
|
| 1084 |
+
"nll_loss": 0.8671246767044067,
|
| 1085 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1086 |
+
"rewards/chosen": -0.032072119414806366,
|
| 1087 |
+
"rewards/margins": 0.033200670033693314,
|
| 1088 |
+
"rewards/rejected": -0.06527277827262878,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.8817777777777778,
|
| 1093 |
+
"grad_norm": 47.876564025878906,
|
| 1094 |
+
"learning_rate": 1.538830716302092e-07,
|
| 1095 |
+
"log_odds_chosen": 1.184543490409851,
|
| 1096 |
+
"log_odds_ratio": -0.3588692843914032,
|
| 1097 |
+
"logits/chosen": 142.39617919921875,
|
| 1098 |
+
"logits/rejected": 142.94223022460938,
|
| 1099 |
+
"logps/chosen": -0.6531465649604797,
|
| 1100 |
+
"logps/rejected": -1.2773756980895996,
|
| 1101 |
+
"loss": 26.2345,
|
| 1102 |
+
"nll_loss": 0.8135266304016113,
|
| 1103 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1104 |
+
"rewards/chosen": -0.03265732526779175,
|
| 1105 |
+
"rewards/margins": 0.031211454421281815,
|
| 1106 |
+
"rewards/rejected": -0.06386878341436386,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9291851851851853,
|
| 1111 |
+
"grad_norm": 41.12287902832031,
|
| 1112 |
+
"learning_rate": 3.8500413544415025e-08,
|
| 1113 |
+
"log_odds_chosen": 1.222886323928833,
|
| 1114 |
+
"log_odds_ratio": -0.37252649664878845,
|
| 1115 |
+
"logits/chosen": 142.91061401367188,
|
| 1116 |
+
"logits/rejected": 140.1123046875,
|
| 1117 |
+
"logps/chosen": -0.613469123840332,
|
| 1118 |
+
"logps/rejected": -1.2194663286209106,
|
| 1119 |
+
"loss": 26.3091,
|
| 1120 |
+
"nll_loss": 0.8243390917778015,
|
| 1121 |
+
"rewards/accuracies": 0.84375,
|
| 1122 |
+
"rewards/chosen": -0.03067345917224884,
|
| 1123 |
+
"rewards/margins": 0.030299866572022438,
|
| 1124 |
+
"rewards/rejected": -0.06097331643104553,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.9765925925925925,
|
| 1129 |
+
"grad_norm": 49.64106369018555,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 1.233667016029358,
|
| 1132 |
+
"log_odds_ratio": -0.3490303158760071,
|
| 1133 |
+
"logits/chosen": 141.09365844726562,
|
| 1134 |
+
"logits/rejected": 138.0718231201172,
|
| 1135 |
+
"logps/chosen": -0.6048657894134521,
|
| 1136 |
+
"logps/rejected": -1.2514102458953857,
|
| 1137 |
+
"loss": 25.6403,
|
| 1138 |
+
"nll_loss": 0.8243409395217896,
|
| 1139 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 1140 |
+
"rewards/chosen": -0.030243292450904846,
|
| 1141 |
+
"rewards/margins": 0.03232722356915474,
|
| 1142 |
+
"rewards/rejected": -0.06257051229476929,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.9765925925925925,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 50.92685067313058,
|
| 1150 |
+
"train_runtime": 8087.8942,
|
| 1151 |
+
"train_samples_per_second": 2.504,
|
| 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 |
+
}
|