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-6e-5
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-borpo-noisy-6e-5
|
| 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-6e-5", 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/cm8xb4wa)
|
| 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: 4.46.1
|
| 39 |
+
- Pytorch: 2.4.0
|
| 40 |
+
- Datasets: 3.1.0
|
| 41 |
+
- Tokenizers: 0.20.1
|
| 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": 2.985781990521327,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 127.4701649257115,
|
| 5 |
+
"train_runtime": 3752.5983,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 5.396,
|
| 8 |
+
"train_steps_per_second": 0.084
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.985781990521327,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 127.4701649257115,
|
| 5 |
+
"train_runtime": 3752.5983,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 5.396,
|
| 8 |
+
"train_steps_per_second": 0.084
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.985781990521327,
|
| 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": 0.04739336492890995,
|
| 13 |
+
"grad_norm": 168960.0,
|
| 14 |
+
"learning_rate": 9.375000000000001e-06,
|
| 15 |
+
"log_odds_chosen": 10.756143569946289,
|
| 16 |
+
"log_odds_ratio": -8.201075553894043,
|
| 17 |
+
"logits/chosen": 137.92239379882812,
|
| 18 |
+
"logits/rejected": 154.74513244628906,
|
| 19 |
+
"logps/chosen": -16.90255355834961,
|
| 20 |
+
"logps/rejected": -27.65829849243164,
|
| 21 |
+
"loss": 1407.4271,
|
| 22 |
+
"nll_loss": 8.030233383178711,
|
| 23 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 24 |
+
"rewards/chosen": -8.451276779174805,
|
| 25 |
+
"rewards/margins": 5.377873420715332,
|
| 26 |
+
"rewards/rejected": -13.82914924621582,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.0947867298578199,
|
| 31 |
+
"grad_norm": 8320.0,
|
| 32 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 33 |
+
"log_odds_chosen": -2.0146212577819824,
|
| 34 |
+
"log_odds_ratio": -9.117185592651367,
|
| 35 |
+
"logits/chosen": 139.0816192626953,
|
| 36 |
+
"logits/rejected": 152.4964141845703,
|
| 37 |
+
"logps/chosen": -16.836305618286133,
|
| 38 |
+
"logps/rejected": -14.821057319641113,
|
| 39 |
+
"loss": 114.8602,
|
| 40 |
+
"nll_loss": 6.842395782470703,
|
| 41 |
+
"rewards/accuracies": 0.512499988079071,
|
| 42 |
+
"rewards/chosen": -8.418152809143066,
|
| 43 |
+
"rewards/margins": -1.0076239109039307,
|
| 44 |
+
"rewards/rejected": -7.410528659820557,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14218009478672985,
|
| 49 |
+
"grad_norm": 25344.0,
|
| 50 |
+
"learning_rate": 2.8125e-05,
|
| 51 |
+
"log_odds_chosen": 6.341404914855957,
|
| 52 |
+
"log_odds_ratio": -7.492286682128906,
|
| 53 |
+
"logits/chosen": 115.94664001464844,
|
| 54 |
+
"logits/rejected": 139.38864135742188,
|
| 55 |
+
"logps/chosen": -19.506237030029297,
|
| 56 |
+
"logps/rejected": -25.845422744750977,
|
| 57 |
+
"loss": 2035.5172,
|
| 58 |
+
"nll_loss": 8.318296432495117,
|
| 59 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 60 |
+
"rewards/chosen": -9.753118515014648,
|
| 61 |
+
"rewards/margins": 3.1695926189422607,
|
| 62 |
+
"rewards/rejected": -12.922711372375488,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.1895734597156398,
|
| 67 |
+
"grad_norm": 3424.0,
|
| 68 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 69 |
+
"log_odds_chosen": -0.7289161682128906,
|
| 70 |
+
"log_odds_ratio": -7.691019535064697,
|
| 71 |
+
"logits/chosen": 105.5518569946289,
|
| 72 |
+
"logits/rejected": 116.6485595703125,
|
| 73 |
+
"logps/chosen": -18.734756469726562,
|
| 74 |
+
"logps/rejected": -18.00742530822754,
|
| 75 |
+
"loss": 1202.2451,
|
| 76 |
+
"nll_loss": 8.924882888793945,
|
| 77 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 78 |
+
"rewards/chosen": -9.367378234863281,
|
| 79 |
+
"rewards/margins": -0.3636666238307953,
|
| 80 |
+
"rewards/rejected": -9.00371265411377,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23696682464454977,
|
| 85 |
+
"grad_norm": 8640.0,
|
| 86 |
+
"learning_rate": 4.6875e-05,
|
| 87 |
+
"log_odds_chosen": -1.369574785232544,
|
| 88 |
+
"log_odds_ratio": -8.745997428894043,
|
| 89 |
+
"logits/chosen": 102.84317779541016,
|
| 90 |
+
"logits/rejected": 96.14292907714844,
|
| 91 |
+
"logps/chosen": -18.200498580932617,
|
| 92 |
+
"logps/rejected": -16.830005645751953,
|
| 93 |
+
"loss": 1457.908,
|
| 94 |
+
"nll_loss": 7.214259147644043,
|
| 95 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 96 |
+
"rewards/chosen": -9.100249290466309,
|
| 97 |
+
"rewards/margins": -0.685246467590332,
|
| 98 |
+
"rewards/rejected": -8.415002822875977,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.2843601895734597,
|
| 103 |
+
"grad_norm": 20480.0,
|
| 104 |
+
"learning_rate": 5.625e-05,
|
| 105 |
+
"log_odds_chosen": -2.44787859916687,
|
| 106 |
+
"log_odds_ratio": -9.492902755737305,
|
| 107 |
+
"logits/chosen": 82.8386459350586,
|
| 108 |
+
"logits/rejected": 124.99415588378906,
|
| 109 |
+
"logps/chosen": -19.397884368896484,
|
| 110 |
+
"logps/rejected": -16.947490692138672,
|
| 111 |
+
"loss": 416.4873,
|
| 112 |
+
"nll_loss": 7.92099666595459,
|
| 113 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 114 |
+
"rewards/chosen": -9.698942184448242,
|
| 115 |
+
"rewards/margins": -1.2251958847045898,
|
| 116 |
+
"rewards/rejected": -8.473745346069336,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33175355450236965,
|
| 121 |
+
"grad_norm": 872.0,
|
| 122 |
+
"learning_rate": 5.998336508818541e-05,
|
| 123 |
+
"log_odds_chosen": -3.3048667907714844,
|
| 124 |
+
"log_odds_ratio": -4.905556678771973,
|
| 125 |
+
"logits/chosen": 179.0537567138672,
|
| 126 |
+
"logits/rejected": 157.49603271484375,
|
| 127 |
+
"logps/chosen": -10.089513778686523,
|
| 128 |
+
"logps/rejected": -6.791520595550537,
|
| 129 |
+
"loss": 183.6225,
|
| 130 |
+
"nll_loss": 5.170867443084717,
|
| 131 |
+
"rewards/accuracies": 0.4375,
|
| 132 |
+
"rewards/chosen": -5.044756889343262,
|
| 133 |
+
"rewards/margins": -1.6489967107772827,
|
| 134 |
+
"rewards/rejected": -3.3957602977752686,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3791469194312796,
|
| 139 |
+
"grad_norm": 388.0,
|
| 140 |
+
"learning_rate": 5.988177409372154e-05,
|
| 141 |
+
"log_odds_chosen": 0.18621893227100372,
|
| 142 |
+
"log_odds_ratio": -0.744963526725769,
|
| 143 |
+
"logits/chosen": 250.6142120361328,
|
| 144 |
+
"logits/rejected": 266.593994140625,
|
| 145 |
+
"logps/chosen": -1.9268105030059814,
|
| 146 |
+
"logps/rejected": -2.0770750045776367,
|
| 147 |
+
"loss": 45.8455,
|
| 148 |
+
"nll_loss": 2.393054962158203,
|
| 149 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 150 |
+
"rewards/chosen": -0.9634052515029907,
|
| 151 |
+
"rewards/margins": 0.07513223588466644,
|
| 152 |
+
"rewards/rejected": -1.0385375022888184,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4265402843601896,
|
| 157 |
+
"grad_norm": 296.0,
|
| 158 |
+
"learning_rate": 5.968814624645376e-05,
|
| 159 |
+
"log_odds_chosen": 0.19682058691978455,
|
| 160 |
+
"log_odds_ratio": -0.9012538194656372,
|
| 161 |
+
"logits/chosen": 257.0416259765625,
|
| 162 |
+
"logits/rejected": 231.23562622070312,
|
| 163 |
+
"logps/chosen": -1.8941532373428345,
|
| 164 |
+
"logps/rejected": -2.064608097076416,
|
| 165 |
+
"loss": 36.973,
|
| 166 |
+
"nll_loss": 2.027169704437256,
|
| 167 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 168 |
+
"rewards/chosen": -0.9470766186714172,
|
| 169 |
+
"rewards/margins": 0.08522741496562958,
|
| 170 |
+
"rewards/rejected": -1.032304048538208,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.47393364928909953,
|
| 175 |
+
"grad_norm": 65.5,
|
| 176 |
+
"learning_rate": 5.9403077926557534e-05,
|
| 177 |
+
"log_odds_chosen": 0.11038754880428314,
|
| 178 |
+
"log_odds_ratio": -0.8040679693222046,
|
| 179 |
+
"logits/chosen": 263.97332763671875,
|
| 180 |
+
"logits/rejected": 269.75946044921875,
|
| 181 |
+
"logps/chosen": -1.6346588134765625,
|
| 182 |
+
"logps/rejected": -1.7171961069107056,
|
| 183 |
+
"loss": 35.7226,
|
| 184 |
+
"nll_loss": 1.9796053171157837,
|
| 185 |
+
"rewards/accuracies": 0.5,
|
| 186 |
+
"rewards/chosen": -0.8173294067382812,
|
| 187 |
+
"rewards/margins": 0.04126860201358795,
|
| 188 |
+
"rewards/rejected": -0.8585980534553528,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5213270142180095,
|
| 193 |
+
"grad_norm": 50.75,
|
| 194 |
+
"learning_rate": 5.9027447153889215e-05,
|
| 195 |
+
"log_odds_chosen": 0.09287216514348984,
|
| 196 |
+
"log_odds_ratio": -0.7353734374046326,
|
| 197 |
+
"logits/chosen": 243.19577026367188,
|
| 198 |
+
"logits/rejected": 245.62234497070312,
|
| 199 |
+
"logps/chosen": -1.287972092628479,
|
| 200 |
+
"logps/rejected": -1.3607467412948608,
|
| 201 |
+
"loss": 30.7576,
|
| 202 |
+
"nll_loss": 1.6278873682022095,
|
| 203 |
+
"rewards/accuracies": 0.550000011920929,
|
| 204 |
+
"rewards/chosen": -0.6439860463142395,
|
| 205 |
+
"rewards/margins": 0.03638739511370659,
|
| 206 |
+
"rewards/rejected": -0.6803733706474304,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5687203791469194,
|
| 211 |
+
"grad_norm": 37.25,
|
| 212 |
+
"learning_rate": 5.856241088365584e-05,
|
| 213 |
+
"log_odds_chosen": 0.1018252968788147,
|
| 214 |
+
"log_odds_ratio": -0.7410688400268555,
|
| 215 |
+
"logits/chosen": 220.8896026611328,
|
| 216 |
+
"logits/rejected": 226.8162841796875,
|
| 217 |
+
"logps/chosen": -1.215947151184082,
|
| 218 |
+
"logps/rejected": -1.2763280868530273,
|
| 219 |
+
"loss": 28.7756,
|
| 220 |
+
"nll_loss": 1.4690172672271729,
|
| 221 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 222 |
+
"rewards/chosen": -0.607973575592041,
|
| 223 |
+
"rewards/margins": 0.030190488323569298,
|
| 224 |
+
"rewards/rejected": -0.6381640434265137,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6161137440758294,
|
| 229 |
+
"grad_norm": 94.0,
|
| 230 |
+
"learning_rate": 5.800940144295476e-05,
|
| 231 |
+
"log_odds_chosen": 0.2972797155380249,
|
| 232 |
+
"log_odds_ratio": -0.6765426993370056,
|
| 233 |
+
"logits/chosen": 227.9703369140625,
|
| 234 |
+
"logits/rejected": 230.8743438720703,
|
| 235 |
+
"logps/chosen": -1.2523950338363647,
|
| 236 |
+
"logps/rejected": -1.4669511318206787,
|
| 237 |
+
"loss": 28.0376,
|
| 238 |
+
"nll_loss": 1.5335967540740967,
|
| 239 |
+
"rewards/accuracies": 0.612500011920929,
|
| 240 |
+
"rewards/chosen": -0.6261975169181824,
|
| 241 |
+
"rewards/margins": 0.10727809369564056,
|
| 242 |
+
"rewards/rejected": -0.7334755659103394,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6635071090047393,
|
| 247 |
+
"grad_norm": 80.5,
|
| 248 |
+
"learning_rate": 5.7370122119158855e-05,
|
| 249 |
+
"log_odds_chosen": 0.3697313070297241,
|
| 250 |
+
"log_odds_ratio": -0.656975507736206,
|
| 251 |
+
"logits/chosen": 229.30117797851562,
|
| 252 |
+
"logits/rejected": 225.6231689453125,
|
| 253 |
+
"logps/chosen": -1.0724413394927979,
|
| 254 |
+
"logps/rejected": -1.3780503273010254,
|
| 255 |
+
"loss": 27.2101,
|
| 256 |
+
"nll_loss": 1.4345372915267944,
|
| 257 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 258 |
+
"rewards/chosen": -0.5362206697463989,
|
| 259 |
+
"rewards/margins": 0.15280446410179138,
|
| 260 |
+
"rewards/rejected": -0.6890251636505127,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7109004739336493,
|
| 265 |
+
"grad_norm": 66.0,
|
| 266 |
+
"learning_rate": 5.6646541913735056e-05,
|
| 267 |
+
"log_odds_chosen": 0.23539912700653076,
|
| 268 |
+
"log_odds_ratio": -0.6753562092781067,
|
| 269 |
+
"logits/chosen": 222.90090942382812,
|
| 270 |
+
"logits/rejected": 224.7488555908203,
|
| 271 |
+
"logps/chosen": -1.040056586265564,
|
| 272 |
+
"logps/rejected": -1.2055822610855103,
|
| 273 |
+
"loss": 26.432,
|
| 274 |
+
"nll_loss": 1.4434144496917725,
|
| 275 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 276 |
+
"rewards/chosen": -0.520028293132782,
|
| 277 |
+
"rewards/margins": 0.08276289701461792,
|
| 278 |
+
"rewards/rejected": -0.6027911305427551,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7582938388625592,
|
| 283 |
+
"grad_norm": 57.0,
|
| 284 |
+
"learning_rate": 5.5840889477654665e-05,
|
| 285 |
+
"log_odds_chosen": 0.21213491261005402,
|
| 286 |
+
"log_odds_ratio": -0.7092779874801636,
|
| 287 |
+
"logits/chosen": 224.779052734375,
|
| 288 |
+
"logits/rejected": 223.54104614257812,
|
| 289 |
+
"logps/chosen": -1.135852575302124,
|
| 290 |
+
"logps/rejected": -1.2810156345367432,
|
| 291 |
+
"loss": 25.729,
|
| 292 |
+
"nll_loss": 1.3958572149276733,
|
| 293 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 294 |
+
"rewards/chosen": -0.567926287651062,
|
| 295 |
+
"rewards/margins": 0.07258154451847076,
|
| 296 |
+
"rewards/rejected": -0.6405078172683716,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8056872037914692,
|
| 301 |
+
"grad_norm": 75.5,
|
| 302 |
+
"learning_rate": 5.495564624707466e-05,
|
| 303 |
+
"log_odds_chosen": 0.22340472042560577,
|
| 304 |
+
"log_odds_ratio": -0.6968339085578918,
|
| 305 |
+
"logits/chosen": 214.7851104736328,
|
| 306 |
+
"logits/rejected": 210.5839080810547,
|
| 307 |
+
"logps/chosen": -1.09432852268219,
|
| 308 |
+
"logps/rejected": -1.2580267190933228,
|
| 309 |
+
"loss": 25.5057,
|
| 310 |
+
"nll_loss": 1.3930976390838623,
|
| 311 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 312 |
+
"rewards/chosen": -0.547164261341095,
|
| 313 |
+
"rewards/margins": 0.08184906840324402,
|
| 314 |
+
"rewards/rejected": -0.6290133595466614,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8530805687203792,
|
| 319 |
+
"grad_norm": 57.25,
|
| 320 |
+
"learning_rate": 5.399353880043222e-05,
|
| 321 |
+
"log_odds_chosen": 0.258540540933609,
|
| 322 |
+
"log_odds_ratio": -0.6605676412582397,
|
| 323 |
+
"logits/chosen": 212.9822540283203,
|
| 324 |
+
"logits/rejected": 210.5906219482422,
|
| 325 |
+
"logps/chosen": -1.0728873014450073,
|
| 326 |
+
"logps/rejected": -1.2394678592681885,
|
| 327 |
+
"loss": 27.487,
|
| 328 |
+
"nll_loss": 1.4236419200897217,
|
| 329 |
+
"rewards/accuracies": 0.612500011920929,
|
| 330 |
+
"rewards/chosen": -0.5364436507225037,
|
| 331 |
+
"rewards/margins": 0.0832902193069458,
|
| 332 |
+
"rewards/rejected": -0.6197339296340942,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9004739336492891,
|
| 337 |
+
"grad_norm": 44.0,
|
| 338 |
+
"learning_rate": 5.295753046049293e-05,
|
| 339 |
+
"log_odds_chosen": 0.33555328845977783,
|
| 340 |
+
"log_odds_ratio": -0.599485456943512,
|
| 341 |
+
"logits/chosen": 199.41171264648438,
|
| 342 |
+
"logits/rejected": 198.87872314453125,
|
| 343 |
+
"logps/chosen": -0.9943248629570007,
|
| 344 |
+
"logps/rejected": -1.2264639139175415,
|
| 345 |
+
"loss": 25.259,
|
| 346 |
+
"nll_loss": 1.2300159931182861,
|
| 347 |
+
"rewards/accuracies": 0.612500011920929,
|
| 348 |
+
"rewards/chosen": -0.49716243147850037,
|
| 349 |
+
"rewards/margins": 0.11606951057910919,
|
| 350 |
+
"rewards/rejected": -0.6132319569587708,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9478672985781991,
|
| 355 |
+
"grad_norm": 28.0,
|
| 356 |
+
"learning_rate": 5.1850812167218644e-05,
|
| 357 |
+
"log_odds_chosen": 0.05684388801455498,
|
| 358 |
+
"log_odds_ratio": -0.7587562799453735,
|
| 359 |
+
"logits/chosen": 202.80810546875,
|
| 360 |
+
"logits/rejected": 196.2851104736328,
|
| 361 |
+
"logps/chosen": -1.1174707412719727,
|
| 362 |
+
"logps/rejected": -1.1865062713623047,
|
| 363 |
+
"loss": 25.2803,
|
| 364 |
+
"nll_loss": 1.4526774883270264,
|
| 365 |
+
"rewards/accuracies": 0.5,
|
| 366 |
+
"rewards/chosen": -0.5587353706359863,
|
| 367 |
+
"rewards/margins": 0.03451773524284363,
|
| 368 |
+
"rewards/rejected": -0.5932531356811523,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.995260663507109,
|
| 373 |
+
"grad_norm": 40.75,
|
| 374 |
+
"learning_rate": 5.067679264956681e-05,
|
| 375 |
+
"log_odds_chosen": 0.40639758110046387,
|
| 376 |
+
"log_odds_ratio": -0.6050174236297607,
|
| 377 |
+
"logits/chosen": 204.6847381591797,
|
| 378 |
+
"logits/rejected": 201.60601806640625,
|
| 379 |
+
"logps/chosen": -1.0167808532714844,
|
| 380 |
+
"logps/rejected": -1.3128955364227295,
|
| 381 |
+
"loss": 24.7542,
|
| 382 |
+
"nll_loss": 1.3192346096038818,
|
| 383 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 384 |
+
"rewards/chosen": -0.5083904266357422,
|
| 385 |
+
"rewards/margins": 0.14805743098258972,
|
| 386 |
+
"rewards/rejected": -0.6564477682113647,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.042654028436019,
|
| 391 |
+
"grad_norm": 23.375,
|
| 392 |
+
"learning_rate": 4.943908792649255e-05,
|
| 393 |
+
"log_odds_chosen": 0.21281662583351135,
|
| 394 |
+
"log_odds_ratio": -0.6757606267929077,
|
| 395 |
+
"logits/chosen": 198.5106201171875,
|
| 396 |
+
"logits/rejected": 196.48208618164062,
|
| 397 |
+
"logps/chosen": -0.9237734079360962,
|
| 398 |
+
"logps/rejected": -1.0613982677459717,
|
| 399 |
+
"loss": 22.539,
|
| 400 |
+
"nll_loss": 1.1923763751983643,
|
| 401 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 402 |
+
"rewards/chosen": -0.4618867039680481,
|
| 403 |
+
"rewards/margins": 0.06881250441074371,
|
| 404 |
+
"rewards/rejected": -0.5306991338729858,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0900473933649288,
|
| 409 |
+
"grad_norm": 51.25,
|
| 410 |
+
"learning_rate": 4.814151016949061e-05,
|
| 411 |
+
"log_odds_chosen": 0.45136967301368713,
|
| 412 |
+
"log_odds_ratio": -0.574053168296814,
|
| 413 |
+
"logits/chosen": 194.5438232421875,
|
| 414 |
+
"logits/rejected": 195.3663787841797,
|
| 415 |
+
"logps/chosen": -0.8666488528251648,
|
| 416 |
+
"logps/rejected": -1.1445974111557007,
|
| 417 |
+
"loss": 22.1373,
|
| 418 |
+
"nll_loss": 1.1306638717651367,
|
| 419 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 420 |
+
"rewards/chosen": -0.4333244264125824,
|
| 421 |
+
"rewards/margins": 0.13897429406642914,
|
| 422 |
+
"rewards/rejected": -0.5722987055778503,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1374407582938388,
|
| 427 |
+
"grad_norm": 32.0,
|
| 428 |
+
"learning_rate": 4.6788055960981e-05,
|
| 429 |
+
"log_odds_chosen": 0.5978150367736816,
|
| 430 |
+
"log_odds_ratio": -0.5191441774368286,
|
| 431 |
+
"logits/chosen": 195.80685424804688,
|
| 432 |
+
"logits/rejected": 192.3872528076172,
|
| 433 |
+
"logps/chosen": -0.816036581993103,
|
| 434 |
+
"logps/rejected": -1.1533689498901367,
|
| 435 |
+
"loss": 21.9447,
|
| 436 |
+
"nll_loss": 1.137957215309143,
|
| 437 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 438 |
+
"rewards/chosen": -0.4080182909965515,
|
| 439 |
+
"rewards/margins": 0.16866618394851685,
|
| 440 |
+
"rewards/rejected": -0.5766844749450684,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1848341232227488,
|
| 445 |
+
"grad_norm": 21.0,
|
| 446 |
+
"learning_rate": 4.538289398470304e-05,
|
| 447 |
+
"log_odds_chosen": 0.44998010993003845,
|
| 448 |
+
"log_odds_ratio": -0.5995658040046692,
|
| 449 |
+
"logits/chosen": 196.32797241210938,
|
| 450 |
+
"logits/rejected": 197.1869659423828,
|
| 451 |
+
"logps/chosen": -0.9477843046188354,
|
| 452 |
+
"logps/rejected": -1.2593724727630615,
|
| 453 |
+
"loss": 21.1467,
|
| 454 |
+
"nll_loss": 1.1062265634536743,
|
| 455 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 456 |
+
"rewards/chosen": -0.4738921523094177,
|
| 457 |
+
"rewards/margins": 0.15579405426979065,
|
| 458 |
+
"rewards/rejected": -0.6296862363815308,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2322274881516588,
|
| 463 |
+
"grad_norm": 21.625,
|
| 464 |
+
"learning_rate": 4.393035218603139e-05,
|
| 465 |
+
"log_odds_chosen": 0.19958534836769104,
|
| 466 |
+
"log_odds_ratio": -0.6756640672683716,
|
| 467 |
+
"logits/chosen": 199.8778076171875,
|
| 468 |
+
"logits/rejected": 195.39340209960938,
|
| 469 |
+
"logps/chosen": -0.9140733480453491,
|
| 470 |
+
"logps/rejected": -1.0152888298034668,
|
| 471 |
+
"loss": 21.3142,
|
| 472 |
+
"nll_loss": 1.2156976461410522,
|
| 473 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 474 |
+
"rewards/chosen": -0.45703667402267456,
|
| 475 |
+
"rewards/margins": 0.05060772970318794,
|
| 476 |
+
"rewards/rejected": -0.5076444149017334,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.2796208530805688,
|
| 481 |
+
"grad_norm": 27.75,
|
| 482 |
+
"learning_rate": 4.243490444176123e-05,
|
| 483 |
+
"log_odds_chosen": 0.38076427578926086,
|
| 484 |
+
"log_odds_ratio": -0.6123644113540649,
|
| 485 |
+
"logits/chosen": 199.5050048828125,
|
| 486 |
+
"logits/rejected": 198.98667907714844,
|
| 487 |
+
"logps/chosen": -0.8708294630050659,
|
| 488 |
+
"logps/rejected": -1.09108304977417,
|
| 489 |
+
"loss": 21.4393,
|
| 490 |
+
"nll_loss": 1.1632344722747803,
|
| 491 |
+
"rewards/accuracies": 0.625,
|
| 492 |
+
"rewards/chosen": -0.43541473150253296,
|
| 493 |
+
"rewards/margins": 0.11012685298919678,
|
| 494 |
+
"rewards/rejected": -0.545541524887085,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3270142180094786,
|
| 499 |
+
"grad_norm": 30.75,
|
| 500 |
+
"learning_rate": 4.090115678041962e-05,
|
| 501 |
+
"log_odds_chosen": 0.45514464378356934,
|
| 502 |
+
"log_odds_ratio": -0.6075628399848938,
|
| 503 |
+
"logits/chosen": 194.0288543701172,
|
| 504 |
+
"logits/rejected": 193.20309448242188,
|
| 505 |
+
"logps/chosen": -0.8634368181228638,
|
| 506 |
+
"logps/rejected": -1.140328288078308,
|
| 507 |
+
"loss": 21.9818,
|
| 508 |
+
"nll_loss": 1.1953437328338623,
|
| 509 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 510 |
+
"rewards/chosen": -0.4317184090614319,
|
| 511 |
+
"rewards/margins": 0.13844572007656097,
|
| 512 |
+
"rewards/rejected": -0.570164144039154,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.3744075829383886,
|
| 517 |
+
"grad_norm": 25.25,
|
| 518 |
+
"learning_rate": 3.9333833195545325e-05,
|
| 519 |
+
"log_odds_chosen": 0.3756052553653717,
|
| 520 |
+
"log_odds_ratio": -0.5947796106338501,
|
| 521 |
+
"logits/chosen": 198.31832885742188,
|
| 522 |
+
"logits/rejected": 192.33865356445312,
|
| 523 |
+
"logps/chosen": -0.9395328760147095,
|
| 524 |
+
"logps/rejected": -1.2024142742156982,
|
| 525 |
+
"loss": 21.7616,
|
| 526 |
+
"nll_loss": 1.2738587856292725,
|
| 527 |
+
"rewards/accuracies": 0.675000011920929,
|
| 528 |
+
"rewards/chosen": -0.46976643800735474,
|
| 529 |
+
"rewards/margins": 0.13144069910049438,
|
| 530 |
+
"rewards/rejected": -0.6012071371078491,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.4218009478672986,
|
| 535 |
+
"grad_norm": 23.0,
|
| 536 |
+
"learning_rate": 3.7737761095632374e-05,
|
| 537 |
+
"log_odds_chosen": 0.3193782866001129,
|
| 538 |
+
"log_odds_ratio": -0.6514483690261841,
|
| 539 |
+
"logits/chosen": 196.14259338378906,
|
| 540 |
+
"logits/rejected": 195.2425537109375,
|
| 541 |
+
"logps/chosen": -0.8644716143608093,
|
| 542 |
+
"logps/rejected": -1.049574851989746,
|
| 543 |
+
"loss": 20.83,
|
| 544 |
+
"nll_loss": 1.167014479637146,
|
| 545 |
+
"rewards/accuracies": 0.637499988079071,
|
| 546 |
+
"rewards/chosen": -0.43223580718040466,
|
| 547 |
+
"rewards/margins": 0.092551589012146,
|
| 548 |
+
"rewards/rejected": -0.524787425994873,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4691943127962086,
|
| 553 |
+
"grad_norm": 20.75,
|
| 554 |
+
"learning_rate": 3.611785643555225e-05,
|
| 555 |
+
"log_odds_chosen": 0.303898423910141,
|
| 556 |
+
"log_odds_ratio": -0.648755669593811,
|
| 557 |
+
"logits/chosen": 200.64492797851562,
|
| 558 |
+
"logits/rejected": 200.30389404296875,
|
| 559 |
+
"logps/chosen": -0.8748540878295898,
|
| 560 |
+
"logps/rejected": -1.0394160747528076,
|
| 561 |
+
"loss": 21.6333,
|
| 562 |
+
"nll_loss": 1.1785424947738647,
|
| 563 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 564 |
+
"rewards/chosen": -0.4374270439147949,
|
| 565 |
+
"rewards/margins": 0.08228104561567307,
|
| 566 |
+
"rewards/rejected": -0.5197080373764038,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.5165876777251186,
|
| 571 |
+
"grad_norm": 21.875,
|
| 572 |
+
"learning_rate": 3.44791085752502e-05,
|
| 573 |
+
"log_odds_chosen": 0.31724172830581665,
|
| 574 |
+
"log_odds_ratio": -0.622181236743927,
|
| 575 |
+
"logits/chosen": 205.41311645507812,
|
| 576 |
+
"logits/rejected": 208.6095733642578,
|
| 577 |
+
"logps/chosen": -0.9168610572814941,
|
| 578 |
+
"logps/rejected": -1.1023683547973633,
|
| 579 |
+
"loss": 22.0025,
|
| 580 |
+
"nll_loss": 1.2696937322616577,
|
| 581 |
+
"rewards/accuracies": 0.612500011920929,
|
| 582 |
+
"rewards/chosen": -0.45843052864074707,
|
| 583 |
+
"rewards/margins": 0.09275360405445099,
|
| 584 |
+
"rewards/rejected": -0.5511841773986816,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5639810426540284,
|
| 589 |
+
"grad_norm": 30.125,
|
| 590 |
+
"learning_rate": 3.2826564912351544e-05,
|
| 591 |
+
"log_odds_chosen": 0.2731252908706665,
|
| 592 |
+
"log_odds_ratio": -0.6811183094978333,
|
| 593 |
+
"logits/chosen": 204.3468017578125,
|
| 594 |
+
"logits/rejected": 205.2547149658203,
|
| 595 |
+
"logps/chosen": -1.0431245565414429,
|
| 596 |
+
"logps/rejected": -1.184552550315857,
|
| 597 |
+
"loss": 21.4814,
|
| 598 |
+
"nll_loss": 1.184350609779358,
|
| 599 |
+
"rewards/accuracies": 0.625,
|
| 600 |
+
"rewards/chosen": -0.5215622782707214,
|
| 601 |
+
"rewards/margins": 0.07071395963430405,
|
| 602 |
+
"rewards/rejected": -0.5922762751579285,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6113744075829384,
|
| 607 |
+
"grad_norm": 30.0,
|
| 608 |
+
"learning_rate": 3.116531533601003e-05,
|
| 609 |
+
"log_odds_chosen": 0.4361351430416107,
|
| 610 |
+
"log_odds_ratio": -0.5953701138496399,
|
| 611 |
+
"logits/chosen": 194.65945434570312,
|
| 612 |
+
"logits/rejected": 192.39102172851562,
|
| 613 |
+
"logps/chosen": -0.8711638450622559,
|
| 614 |
+
"logps/rejected": -1.1436076164245605,
|
| 615 |
+
"loss": 21.1767,
|
| 616 |
+
"nll_loss": 1.1069728136062622,
|
| 617 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 618 |
+
"rewards/chosen": -0.43558192253112793,
|
| 619 |
+
"rewards/margins": 0.13622191548347473,
|
| 620 |
+
"rewards/rejected": -0.5718038082122803,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6587677725118484,
|
| 625 |
+
"grad_norm": 34.0,
|
| 626 |
+
"learning_rate": 2.9500476549880848e-05,
|
| 627 |
+
"log_odds_chosen": 0.3290528357028961,
|
| 628 |
+
"log_odds_ratio": -0.6428475975990295,
|
| 629 |
+
"logits/chosen": 200.77029418945312,
|
| 630 |
+
"logits/rejected": 195.89601135253906,
|
| 631 |
+
"logps/chosen": -0.8381175994873047,
|
| 632 |
+
"logps/rejected": -1.0445606708526611,
|
| 633 |
+
"loss": 20.8206,
|
| 634 |
+
"nll_loss": 1.1465178728103638,
|
| 635 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 636 |
+
"rewards/chosen": -0.41905879974365234,
|
| 637 |
+
"rewards/margins": 0.10322149097919464,
|
| 638 |
+
"rewards/rejected": -0.5222803354263306,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7061611374407581,
|
| 643 |
+
"grad_norm": 28.0,
|
| 644 |
+
"learning_rate": 2.7837176312504037e-05,
|
| 645 |
+
"log_odds_chosen": 0.03685625642538071,
|
| 646 |
+
"log_odds_ratio": -0.766934335231781,
|
| 647 |
+
"logits/chosen": 198.74905395507812,
|
| 648 |
+
"logits/rejected": 196.25143432617188,
|
| 649 |
+
"logps/chosen": -0.9278505444526672,
|
| 650 |
+
"logps/rejected": -0.9253548383712769,
|
| 651 |
+
"loss": 21.729,
|
| 652 |
+
"nll_loss": 1.2215286493301392,
|
| 653 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 654 |
+
"rewards/chosen": -0.4639252722263336,
|
| 655 |
+
"rewards/margins": -0.0012478366261348128,
|
| 656 |
+
"rewards/rejected": -0.4626774191856384,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.7535545023696684,
|
| 661 |
+
"grad_norm": 18.0,
|
| 662 |
+
"learning_rate": 2.618053764363861e-05,
|
| 663 |
+
"log_odds_chosen": 0.3314729630947113,
|
| 664 |
+
"log_odds_ratio": -0.6066881418228149,
|
| 665 |
+
"logits/chosen": 201.25289916992188,
|
| 666 |
+
"logits/rejected": 198.02322387695312,
|
| 667 |
+
"logps/chosen": -0.8792837858200073,
|
| 668 |
+
"logps/rejected": -1.0925233364105225,
|
| 669 |
+
"loss": 21.2035,
|
| 670 |
+
"nll_loss": 1.1164947748184204,
|
| 671 |
+
"rewards/accuracies": 0.675000011920929,
|
| 672 |
+
"rewards/chosen": -0.43964189291000366,
|
| 673 |
+
"rewards/margins": 0.10661973804235458,
|
| 674 |
+
"rewards/rejected": -0.5462616682052612,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.8009478672985781,
|
| 679 |
+
"grad_norm": 38.25,
|
| 680 |
+
"learning_rate": 2.453566304519216e-05,
|
| 681 |
+
"log_odds_chosen": 0.4536499083042145,
|
| 682 |
+
"log_odds_ratio": -0.5942190885543823,
|
| 683 |
+
"logits/chosen": 203.0521697998047,
|
| 684 |
+
"logits/rejected": 202.32650756835938,
|
| 685 |
+
"logps/chosen": -0.9582914113998413,
|
| 686 |
+
"logps/rejected": -1.2642791271209717,
|
| 687 |
+
"loss": 21.7423,
|
| 688 |
+
"nll_loss": 1.1580461263656616,
|
| 689 |
+
"rewards/accuracies": 0.699999988079071,
|
| 690 |
+
"rewards/chosen": -0.47914570569992065,
|
| 691 |
+
"rewards/margins": 0.1529938280582428,
|
| 692 |
+
"rewards/rejected": -0.6321395635604858,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8483412322274881,
|
| 697 |
+
"grad_norm": 22.375,
|
| 698 |
+
"learning_rate": 2.29076187853462e-05,
|
| 699 |
+
"log_odds_chosen": 0.4630239009857178,
|
| 700 |
+
"log_odds_ratio": -0.5749759078025818,
|
| 701 |
+
"logits/chosen": 196.7127685546875,
|
| 702 |
+
"logits/rejected": 196.4191131591797,
|
| 703 |
+
"logps/chosen": -0.8674151301383972,
|
| 704 |
+
"logps/rejected": -1.1494576930999756,
|
| 705 |
+
"loss": 20.9195,
|
| 706 |
+
"nll_loss": 1.1604869365692139,
|
| 707 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 708 |
+
"rewards/chosen": -0.4337075650691986,
|
| 709 |
+
"rewards/margins": 0.14102117717266083,
|
| 710 |
+
"rewards/rejected": -0.5747288465499878,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8957345971563981,
|
| 715 |
+
"grad_norm": 28.625,
|
| 716 |
+
"learning_rate": 2.130141929428254e-05,
|
| 717 |
+
"log_odds_chosen": 0.35148704051971436,
|
| 718 |
+
"log_odds_ratio": -0.66729336977005,
|
| 719 |
+
"logits/chosen": 197.56497192382812,
|
| 720 |
+
"logits/rejected": 196.6879425048828,
|
| 721 |
+
"logps/chosen": -0.8802660703659058,
|
| 722 |
+
"logps/rejected": -1.102311134338379,
|
| 723 |
+
"loss": 22.0774,
|
| 724 |
+
"nll_loss": 1.1984275579452515,
|
| 725 |
+
"rewards/accuracies": 0.625,
|
| 726 |
+
"rewards/chosen": -0.4401330351829529,
|
| 727 |
+
"rewards/margins": 0.11102245002985,
|
| 728 |
+
"rewards/rejected": -0.5511555671691895,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.943127962085308,
|
| 733 |
+
"grad_norm": 25.375,
|
| 734 |
+
"learning_rate": 1.9722011719572444e-05,
|
| 735 |
+
"log_odds_chosen": 0.21564432978630066,
|
| 736 |
+
"log_odds_ratio": -0.6583319902420044,
|
| 737 |
+
"logits/chosen": 202.26856994628906,
|
| 738 |
+
"logits/rejected": 193.0558624267578,
|
| 739 |
+
"logps/chosen": -0.9100298881530762,
|
| 740 |
+
"logps/rejected": -1.0561821460723877,
|
| 741 |
+
"loss": 20.1611,
|
| 742 |
+
"nll_loss": 1.0852024555206299,
|
| 743 |
+
"rewards/accuracies": 0.574999988079071,
|
| 744 |
+
"rewards/chosen": -0.4550149440765381,
|
| 745 |
+
"rewards/margins": 0.07307618111371994,
|
| 746 |
+
"rewards/rejected": -0.5280910730361938,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.9905213270142181,
|
| 751 |
+
"grad_norm": 23.625,
|
| 752 |
+
"learning_rate": 1.8174260688798445e-05,
|
| 753 |
+
"log_odds_chosen": 0.3166791498661041,
|
| 754 |
+
"log_odds_ratio": -0.630929172039032,
|
| 755 |
+
"logits/chosen": 197.60903930664062,
|
| 756 |
+
"logits/rejected": 196.84121704101562,
|
| 757 |
+
"logps/chosen": -0.821063220500946,
|
| 758 |
+
"logps/rejected": -0.9948121905326843,
|
| 759 |
+
"loss": 19.9686,
|
| 760 |
+
"nll_loss": 1.0750689506530762,
|
| 761 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 762 |
+
"rewards/chosen": -0.410531610250473,
|
| 763 |
+
"rewards/margins": 0.08687452226877213,
|
| 764 |
+
"rewards/rejected": -0.49740609526634216,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.037914691943128,
|
| 769 |
+
"grad_norm": 22.75,
|
| 770 |
+
"learning_rate": 1.666293332634042e-05,
|
| 771 |
+
"log_odds_chosen": 0.6822348833084106,
|
| 772 |
+
"log_odds_ratio": -0.5266743898391724,
|
| 773 |
+
"logits/chosen": 191.23080444335938,
|
| 774 |
+
"logits/rejected": 194.97836303710938,
|
| 775 |
+
"logps/chosen": -0.7306900024414062,
|
| 776 |
+
"logps/rejected": -1.060121774673462,
|
| 777 |
+
"loss": 18.0893,
|
| 778 |
+
"nll_loss": 0.9460033178329468,
|
| 779 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 780 |
+
"rewards/chosen": -0.3653450012207031,
|
| 781 |
+
"rewards/margins": 0.16471591591835022,
|
| 782 |
+
"rewards/rejected": -0.530060887336731,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.085308056872038,
|
| 787 |
+
"grad_norm": 23.875,
|
| 788 |
+
"learning_rate": 1.519268457047482e-05,
|
| 789 |
+
"log_odds_chosen": 0.8683069944381714,
|
| 790 |
+
"log_odds_ratio": -0.4624325633049011,
|
| 791 |
+
"logits/chosen": 185.07095336914062,
|
| 792 |
+
"logits/rejected": 190.39279174804688,
|
| 793 |
+
"logps/chosen": -0.6318475008010864,
|
| 794 |
+
"logps/rejected": -1.0846574306488037,
|
| 795 |
+
"loss": 16.6833,
|
| 796 |
+
"nll_loss": 0.8812177777290344,
|
| 797 |
+
"rewards/accuracies": 0.8125,
|
| 798 |
+
"rewards/chosen": -0.3159237504005432,
|
| 799 |
+
"rewards/margins": 0.22640495002269745,
|
| 800 |
+
"rewards/rejected": -0.5423287153244019,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.132701421800948,
|
| 805 |
+
"grad_norm": 18.75,
|
| 806 |
+
"learning_rate": 1.3768042836010768e-05,
|
| 807 |
+
"log_odds_chosen": 0.3730294704437256,
|
| 808 |
+
"log_odds_ratio": -0.6350643038749695,
|
| 809 |
+
"logits/chosen": 194.38063049316406,
|
| 810 |
+
"logits/rejected": 189.1841583251953,
|
| 811 |
+
"logps/chosen": -0.7411255836486816,
|
| 812 |
+
"logps/rejected": -0.9265958070755005,
|
| 813 |
+
"loss": 17.0913,
|
| 814 |
+
"nll_loss": 1.006074070930481,
|
| 815 |
+
"rewards/accuracies": 0.675000011920929,
|
| 816 |
+
"rewards/chosen": -0.3705627918243408,
|
| 817 |
+
"rewards/margins": 0.09273514896631241,
|
| 818 |
+
"rewards/rejected": -0.46329790353775024,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.1800947867298577,
|
| 823 |
+
"grad_norm": 23.25,
|
| 824 |
+
"learning_rate": 1.239339606662261e-05,
|
| 825 |
+
"log_odds_chosen": 0.6575037240982056,
|
| 826 |
+
"log_odds_ratio": -0.4991639256477356,
|
| 827 |
+
"logits/chosen": 183.24179077148438,
|
| 828 |
+
"logits/rejected": 185.40365600585938,
|
| 829 |
+
"logps/chosen": -0.6491117477416992,
|
| 830 |
+
"logps/rejected": -1.0063084363937378,
|
| 831 |
+
"loss": 16.5076,
|
| 832 |
+
"nll_loss": 0.8716222643852234,
|
| 833 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 834 |
+
"rewards/chosen": -0.3245558738708496,
|
| 835 |
+
"rewards/margins": 0.17859837412834167,
|
| 836 |
+
"rewards/rejected": -0.5031542181968689,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.227488151658768,
|
| 841 |
+
"grad_norm": 24.875,
|
| 842 |
+
"learning_rate": 1.1072978219838283e-05,
|
| 843 |
+
"log_odds_chosen": 0.4254986345767975,
|
| 844 |
+
"log_odds_ratio": -0.5929109454154968,
|
| 845 |
+
"logits/chosen": 181.78013610839844,
|
| 846 |
+
"logits/rejected": 184.6556854248047,
|
| 847 |
+
"logps/chosen": -0.707780122756958,
|
| 848 |
+
"logps/rejected": -0.9049354791641235,
|
| 849 |
+
"loss": 16.9862,
|
| 850 |
+
"nll_loss": 0.9195895195007324,
|
| 851 |
+
"rewards/accuracies": 0.699999988079071,
|
| 852 |
+
"rewards/chosen": -0.353890061378479,
|
| 853 |
+
"rewards/margins": 0.09857770055532455,
|
| 854 |
+
"rewards/rejected": -0.45246773958206177,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2748815165876777,
|
| 859 |
+
"grad_norm": 21.0,
|
| 860 |
+
"learning_rate": 9.810856226309972e-06,
|
| 861 |
+
"log_odds_chosen": 0.8151445388793945,
|
| 862 |
+
"log_odds_ratio": -0.45585957169532776,
|
| 863 |
+
"logits/chosen": 182.42929077148438,
|
| 864 |
+
"logits/rejected": 186.09323120117188,
|
| 865 |
+
"logps/chosen": -0.6263293027877808,
|
| 866 |
+
"logps/rejected": -1.0641155242919922,
|
| 867 |
+
"loss": 16.7978,
|
| 868 |
+
"nll_loss": 0.9048817753791809,
|
| 869 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 870 |
+
"rewards/chosen": -0.3131646513938904,
|
| 871 |
+
"rewards/margins": 0.21889305114746094,
|
| 872 |
+
"rewards/rejected": -0.5320577621459961,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.322274881516588,
|
| 877 |
+
"grad_norm": 23.375,
|
| 878 |
+
"learning_rate": 8.61091746353324e-06,
|
| 879 |
+
"log_odds_chosen": 0.6102806925773621,
|
| 880 |
+
"log_odds_ratio": -0.5228442549705505,
|
| 881 |
+
"logits/chosen": 184.25186157226562,
|
| 882 |
+
"logits/rejected": 188.93673706054688,
|
| 883 |
+
"logps/chosen": -0.6725679636001587,
|
| 884 |
+
"logps/rejected": -0.954127311706543,
|
| 885 |
+
"loss": 16.4777,
|
| 886 |
+
"nll_loss": 0.9074475169181824,
|
| 887 |
+
"rewards/accuracies": 0.75,
|
| 888 |
+
"rewards/chosen": -0.33628398180007935,
|
| 889 |
+
"rewards/margins": 0.14077970385551453,
|
| 890 |
+
"rewards/rejected": -0.4770636558532715,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3696682464454977,
|
| 895 |
+
"grad_norm": 18.125,
|
| 896 |
+
"learning_rate": 7.47685778259568e-06,
|
| 897 |
+
"log_odds_chosen": 0.8383617401123047,
|
| 898 |
+
"log_odds_ratio": -0.45046114921569824,
|
| 899 |
+
"logits/chosen": 183.37762451171875,
|
| 900 |
+
"logits/rejected": 189.5059356689453,
|
| 901 |
+
"logps/chosen": -0.6437116861343384,
|
| 902 |
+
"logps/rejected": -1.0930787324905396,
|
| 903 |
+
"loss": 16.4396,
|
| 904 |
+
"nll_loss": 0.9055509567260742,
|
| 905 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 906 |
+
"rewards/chosen": -0.3218558430671692,
|
| 907 |
+
"rewards/margins": 0.2246834933757782,
|
| 908 |
+
"rewards/rejected": -0.5465393662452698,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.4170616113744074,
|
| 913 |
+
"grad_norm": 18.125,
|
| 914 |
+
"learning_rate": 6.4121701248332905e-06,
|
| 915 |
+
"log_odds_chosen": 0.6893147230148315,
|
| 916 |
+
"log_odds_ratio": -0.5377334356307983,
|
| 917 |
+
"logits/chosen": 179.131591796875,
|
| 918 |
+
"logits/rejected": 181.28529357910156,
|
| 919 |
+
"logps/chosen": -0.6199325323104858,
|
| 920 |
+
"logps/rejected": -0.9626436233520508,
|
| 921 |
+
"loss": 16.2759,
|
| 922 |
+
"nll_loss": 0.854143500328064,
|
| 923 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 924 |
+
"rewards/chosen": -0.3099662661552429,
|
| 925 |
+
"rewards/margins": 0.17135553061962128,
|
| 926 |
+
"rewards/rejected": -0.4813218116760254,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.4644549763033177,
|
| 931 |
+
"grad_norm": 19.5,
|
| 932 |
+
"learning_rate": 5.420133763455645e-06,
|
| 933 |
+
"log_odds_chosen": 0.6930850148200989,
|
| 934 |
+
"log_odds_ratio": -0.5194807648658752,
|
| 935 |
+
"logits/chosen": 179.89645385742188,
|
| 936 |
+
"logits/rejected": 182.66842651367188,
|
| 937 |
+
"logps/chosen": -0.6123950481414795,
|
| 938 |
+
"logps/rejected": -0.9452868700027466,
|
| 939 |
+
"loss": 16.4239,
|
| 940 |
+
"nll_loss": 0.902696430683136,
|
| 941 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 942 |
+
"rewards/chosen": -0.30619752407073975,
|
| 943 |
+
"rewards/margins": 0.16644588112831116,
|
| 944 |
+
"rewards/rejected": -0.4726434350013733,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5118483412322274,
|
| 949 |
+
"grad_norm": 20.5,
|
| 950 |
+
"learning_rate": 4.503804203275866e-06,
|
| 951 |
+
"log_odds_chosen": 0.7105423212051392,
|
| 952 |
+
"log_odds_ratio": -0.5525649189949036,
|
| 953 |
+
"logits/chosen": 177.3004608154297,
|
| 954 |
+
"logits/rejected": 179.8584747314453,
|
| 955 |
+
"logps/chosen": -0.6415736079216003,
|
| 956 |
+
"logps/rejected": -1.0120224952697754,
|
| 957 |
+
"loss": 16.3394,
|
| 958 |
+
"nll_loss": 0.8138397336006165,
|
| 959 |
+
"rewards/accuracies": 0.824999988079071,
|
| 960 |
+
"rewards/chosen": -0.32078680396080017,
|
| 961 |
+
"rewards/margins": 0.18522436916828156,
|
| 962 |
+
"rewards/rejected": -0.5060112476348877,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.5592417061611377,
|
| 967 |
+
"grad_norm": 18.5,
|
| 968 |
+
"learning_rate": 3.6660037696547376e-06,
|
| 969 |
+
"log_odds_chosen": 0.725407063961029,
|
| 970 |
+
"log_odds_ratio": -0.483724445104599,
|
| 971 |
+
"logits/chosen": 181.6314239501953,
|
| 972 |
+
"logits/rejected": 184.50576782226562,
|
| 973 |
+
"logps/chosen": -0.6412969827651978,
|
| 974 |
+
"logps/rejected": -0.9860894083976746,
|
| 975 |
+
"loss": 16.5899,
|
| 976 |
+
"nll_loss": 0.893083393573761,
|
| 977 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 978 |
+
"rewards/chosen": -0.3206484913825989,
|
| 979 |
+
"rewards/margins": 0.1723962128162384,
|
| 980 |
+
"rewards/rejected": -0.4930447041988373,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.6066350710900474,
|
| 985 |
+
"grad_norm": 19.625,
|
| 986 |
+
"learning_rate": 2.909312915645238e-06,
|
| 987 |
+
"log_odds_chosen": 0.6999877095222473,
|
| 988 |
+
"log_odds_ratio": -0.4896921217441559,
|
| 989 |
+
"logits/chosen": 179.7862548828125,
|
| 990 |
+
"logits/rejected": 178.49549865722656,
|
| 991 |
+
"logps/chosen": -0.6378815770149231,
|
| 992 |
+
"logps/rejected": -0.9467176198959351,
|
| 993 |
+
"loss": 16.7756,
|
| 994 |
+
"nll_loss": 0.8352192640304565,
|
| 995 |
+
"rewards/accuracies": 0.75,
|
| 996 |
+
"rewards/chosen": -0.31894078850746155,
|
| 997 |
+
"rewards/margins": 0.15441803634166718,
|
| 998 |
+
"rewards/rejected": -0.47335880994796753,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.654028436018957,
|
| 1003 |
+
"grad_norm": 23.25,
|
| 1004 |
+
"learning_rate": 2.236062274111741e-06,
|
| 1005 |
+
"log_odds_chosen": 0.7541594505310059,
|
| 1006 |
+
"log_odds_ratio": -0.5146032571792603,
|
| 1007 |
+
"logits/chosen": 178.07884216308594,
|
| 1008 |
+
"logits/rejected": 179.99327087402344,
|
| 1009 |
+
"logps/chosen": -0.6102009415626526,
|
| 1010 |
+
"logps/rejected": -1.0283238887786865,
|
| 1011 |
+
"loss": 15.7903,
|
| 1012 |
+
"nll_loss": 0.8353471755981445,
|
| 1013 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1014 |
+
"rewards/chosen": -0.3051004707813263,
|
| 1015 |
+
"rewards/margins": 0.20906153321266174,
|
| 1016 |
+
"rewards/rejected": -0.5141619443893433,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.7014218009478674,
|
| 1021 |
+
"grad_norm": 22.5,
|
| 1022 |
+
"learning_rate": 1.648325479303684e-06,
|
| 1023 |
+
"log_odds_chosen": 0.6386028528213501,
|
| 1024 |
+
"log_odds_ratio": -0.5239256024360657,
|
| 1025 |
+
"logits/chosen": 181.93246459960938,
|
| 1026 |
+
"logits/rejected": 183.00357055664062,
|
| 1027 |
+
"logps/chosen": -0.5961137413978577,
|
| 1028 |
+
"logps/rejected": -0.9210435748100281,
|
| 1029 |
+
"loss": 16.5912,
|
| 1030 |
+
"nll_loss": 0.8747022747993469,
|
| 1031 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1032 |
+
"rewards/chosen": -0.29805687069892883,
|
| 1033 |
+
"rewards/margins": 0.162464901804924,
|
| 1034 |
+
"rewards/rejected": -0.46052178740501404,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.748815165876777,
|
| 1039 |
+
"grad_norm": 23.0,
|
| 1040 |
+
"learning_rate": 1.1479127799935029e-06,
|
| 1041 |
+
"log_odds_chosen": 0.6820887327194214,
|
| 1042 |
+
"log_odds_ratio": -0.5130306482315063,
|
| 1043 |
+
"logits/chosen": 180.50137329101562,
|
| 1044 |
+
"logits/rejected": 187.8414764404297,
|
| 1045 |
+
"logps/chosen": -0.6403064727783203,
|
| 1046 |
+
"logps/rejected": -0.9878012537956238,
|
| 1047 |
+
"loss": 16.6567,
|
| 1048 |
+
"nll_loss": 0.873367190361023,
|
| 1049 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1050 |
+
"rewards/chosen": -0.32015323638916016,
|
| 1051 |
+
"rewards/margins": 0.17374737560749054,
|
| 1052 |
+
"rewards/rejected": -0.4939006268978119,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.7962085308056874,
|
| 1057 |
+
"grad_norm": 33.5,
|
| 1058 |
+
"learning_rate": 7.363654638505046e-07,
|
| 1059 |
+
"log_odds_chosen": 0.8129827380180359,
|
| 1060 |
+
"log_odds_ratio": -0.45822620391845703,
|
| 1061 |
+
"logits/chosen": 181.46929931640625,
|
| 1062 |
+
"logits/rejected": 186.03634643554688,
|
| 1063 |
+
"logps/chosen": -0.6228169202804565,
|
| 1064 |
+
"logps/rejected": -1.0206798315048218,
|
| 1065 |
+
"loss": 16.5789,
|
| 1066 |
+
"nll_loss": 0.825291633605957,
|
| 1067 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1068 |
+
"rewards/chosen": -0.31140846014022827,
|
| 1069 |
+
"rewards/margins": 0.1989315003156662,
|
| 1070 |
+
"rewards/rejected": -0.5103399157524109,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.843601895734597,
|
| 1075 |
+
"grad_norm": 20.25,
|
| 1076 |
+
"learning_rate": 4.149511102238568e-07,
|
| 1077 |
+
"log_odds_chosen": 0.6022200584411621,
|
| 1078 |
+
"log_odds_ratio": -0.5112254023551941,
|
| 1079 |
+
"logits/chosen": 186.76828002929688,
|
| 1080 |
+
"logits/rejected": 184.95945739746094,
|
| 1081 |
+
"logps/chosen": -0.656291127204895,
|
| 1082 |
+
"logps/rejected": -0.963117241859436,
|
| 1083 |
+
"loss": 16.9259,
|
| 1084 |
+
"nll_loss": 0.9450982809066772,
|
| 1085 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1086 |
+
"rewards/chosen": -0.3281455636024475,
|
| 1087 |
+
"rewards/margins": 0.1534130871295929,
|
| 1088 |
+
"rewards/rejected": -0.481558620929718,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.890995260663507,
|
| 1093 |
+
"grad_norm": 18.75,
|
| 1094 |
+
"learning_rate": 1.8465968595625105e-07,
|
| 1095 |
+
"log_odds_chosen": 0.7331669926643372,
|
| 1096 |
+
"log_odds_ratio": -0.4934759736061096,
|
| 1097 |
+
"logits/chosen": 180.53512573242188,
|
| 1098 |
+
"logits/rejected": 180.0854949951172,
|
| 1099 |
+
"logps/chosen": -0.6695073843002319,
|
| 1100 |
+
"logps/rejected": -1.0153210163116455,
|
| 1101 |
+
"loss": 16.1073,
|
| 1102 |
+
"nll_loss": 0.8310354948043823,
|
| 1103 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1104 |
+
"rewards/chosen": -0.33475369215011597,
|
| 1105 |
+
"rewards/margins": 0.17290683090686798,
|
| 1106 |
+
"rewards/rejected": -0.5076605081558228,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.938388625592417,
|
| 1111 |
+
"grad_norm": 19.875,
|
| 1112 |
+
"learning_rate": 4.620049625329803e-08,
|
| 1113 |
+
"log_odds_chosen": 0.8787549138069153,
|
| 1114 |
+
"log_odds_ratio": -0.4447788596153259,
|
| 1115 |
+
"logits/chosen": 182.63246154785156,
|
| 1116 |
+
"logits/rejected": 181.45892333984375,
|
| 1117 |
+
"logps/chosen": -0.6264249682426453,
|
| 1118 |
+
"logps/rejected": -1.0081883668899536,
|
| 1119 |
+
"loss": 16.5378,
|
| 1120 |
+
"nll_loss": 0.8261914253234863,
|
| 1121 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1122 |
+
"rewards/chosen": -0.31321248412132263,
|
| 1123 |
+
"rewards/margins": 0.19088168442249298,
|
| 1124 |
+
"rewards/rejected": -0.5040941834449768,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.985781990521327,
|
| 1129 |
+
"grad_norm": 21.75,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 0.7384462952613831,
|
| 1132 |
+
"log_odds_ratio": -0.4752270579338074,
|
| 1133 |
+
"logits/chosen": 184.30104064941406,
|
| 1134 |
+
"logits/rejected": 181.8874053955078,
|
| 1135 |
+
"logps/chosen": -0.6386845707893372,
|
| 1136 |
+
"logps/rejected": -1.0048197507858276,
|
| 1137 |
+
"loss": 16.1077,
|
| 1138 |
+
"nll_loss": 0.8921818733215332,
|
| 1139 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1140 |
+
"rewards/chosen": -0.3193422853946686,
|
| 1141 |
+
"rewards/margins": 0.18306761980056763,
|
| 1142 |
+
"rewards/rejected": -0.5024098753929138,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.985781990521327,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 127.4701649257115,
|
| 1150 |
+
"train_runtime": 3752.5983,
|
| 1151 |
+
"train_samples_per_second": 5.396,
|
| 1152 |
+
"train_steps_per_second": 0.084
|
| 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 |
+
}
|