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-softplus-basic-5e-5
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-softplus-basic-5e-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-softplus-basic-5e-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/xu25ngo6)
|
| 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": 71.97097947862413,
|
| 5 |
+
"train_runtime": 8116.2448,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.495,
|
| 8 |
+
"train_steps_per_second": 0.039
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.9765925925925925,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 71.97097947862413,
|
| 5 |
+
"train_runtime": 8116.2448,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.495,
|
| 8 |
+
"train_steps_per_second": 0.039
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.9765925925925925,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 315,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.047407407407407405,
|
| 13 |
+
"grad_norm": 899.4839477539062,
|
| 14 |
+
"learning_rate": 7.8125e-06,
|
| 15 |
+
"log_odds_chosen": -3.253441572189331,
|
| 16 |
+
"log_odds_ratio": -11.290441513061523,
|
| 17 |
+
"logits/chosen": 151.96023559570312,
|
| 18 |
+
"logits/rejected": 125.16461181640625,
|
| 19 |
+
"logps/chosen": -21.246450424194336,
|
| 20 |
+
"logps/rejected": -17.994016647338867,
|
| 21 |
+
"loss": 456.0568,
|
| 22 |
+
"nll_loss": 9.107990264892578,
|
| 23 |
+
"rewards/accuracies": 0.512499988079071,
|
| 24 |
+
"rewards/chosen": -10.623225212097168,
|
| 25 |
+
"rewards/margins": -1.6262180805206299,
|
| 26 |
+
"rewards/rejected": -8.997008323669434,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 665.6010131835938,
|
| 32 |
+
"learning_rate": 1.5625e-05,
|
| 33 |
+
"log_odds_chosen": -0.5072727203369141,
|
| 34 |
+
"log_odds_ratio": -9.659940719604492,
|
| 35 |
+
"logits/chosen": 151.84805297851562,
|
| 36 |
+
"logits/rejected": 114.96217346191406,
|
| 37 |
+
"logps/chosen": -19.77961540222168,
|
| 38 |
+
"logps/rejected": -19.273290634155273,
|
| 39 |
+
"loss": 394.0516,
|
| 40 |
+
"nll_loss": 8.324145317077637,
|
| 41 |
+
"rewards/accuracies": 0.40625,
|
| 42 |
+
"rewards/chosen": -9.88980770111084,
|
| 43 |
+
"rewards/margins": -0.25316303968429565,
|
| 44 |
+
"rewards/rejected": -9.636645317077637,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 584.8187255859375,
|
| 50 |
+
"learning_rate": 2.34375e-05,
|
| 51 |
+
"log_odds_chosen": -3.0495519638061523,
|
| 52 |
+
"log_odds_ratio": -12.564888000488281,
|
| 53 |
+
"logits/chosen": 165.22348022460938,
|
| 54 |
+
"logits/rejected": 135.7709197998047,
|
| 55 |
+
"logps/chosen": -23.73749351501465,
|
| 56 |
+
"logps/rejected": -20.685997009277344,
|
| 57 |
+
"loss": 398.6285,
|
| 58 |
+
"nll_loss": 8.013145446777344,
|
| 59 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 60 |
+
"rewards/chosen": -11.868746757507324,
|
| 61 |
+
"rewards/margins": -1.5257480144500732,
|
| 62 |
+
"rewards/rejected": -10.342998504638672,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 2222.182861328125,
|
| 68 |
+
"learning_rate": 3.125e-05,
|
| 69 |
+
"log_odds_chosen": -1.6242220401763916,
|
| 70 |
+
"log_odds_ratio": -6.402337551116943,
|
| 71 |
+
"logits/chosen": 166.9097137451172,
|
| 72 |
+
"logits/rejected": 148.36062622070312,
|
| 73 |
+
"logps/chosen": -13.310510635375977,
|
| 74 |
+
"logps/rejected": -11.688575744628906,
|
| 75 |
+
"loss": 263.4254,
|
| 76 |
+
"nll_loss": 5.480126857757568,
|
| 77 |
+
"rewards/accuracies": 0.4312500059604645,
|
| 78 |
+
"rewards/chosen": -6.655255317687988,
|
| 79 |
+
"rewards/margins": -0.8109671473503113,
|
| 80 |
+
"rewards/rejected": -5.844287872314453,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 239.38790893554688,
|
| 86 |
+
"learning_rate": 3.90625e-05,
|
| 87 |
+
"log_odds_chosen": -0.3463691771030426,
|
| 88 |
+
"log_odds_ratio": -1.4706324338912964,
|
| 89 |
+
"logits/chosen": 178.37814331054688,
|
| 90 |
+
"logits/rejected": 188.9853515625,
|
| 91 |
+
"logps/chosen": -2.9337403774261475,
|
| 92 |
+
"logps/rejected": -2.58659291267395,
|
| 93 |
+
"loss": 101.069,
|
| 94 |
+
"nll_loss": 2.465604305267334,
|
| 95 |
+
"rewards/accuracies": 0.512499988079071,
|
| 96 |
+
"rewards/chosen": -1.4668701887130737,
|
| 97 |
+
"rewards/margins": -0.17357376217842102,
|
| 98 |
+
"rewards/rejected": -1.293296456336975,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 168.3093719482422,
|
| 104 |
+
"learning_rate": 4.6875e-05,
|
| 105 |
+
"log_odds_chosen": -0.08818509429693222,
|
| 106 |
+
"log_odds_ratio": -0.9202126264572144,
|
| 107 |
+
"logits/chosen": 208.46914672851562,
|
| 108 |
+
"logits/rejected": 211.0690155029297,
|
| 109 |
+
"logps/chosen": -1.8011804819107056,
|
| 110 |
+
"logps/rejected": -1.7284847497940063,
|
| 111 |
+
"loss": 76.8201,
|
| 112 |
+
"nll_loss": 1.9414145946502686,
|
| 113 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 114 |
+
"rewards/chosen": -0.9005902409553528,
|
| 115 |
+
"rewards/margins": -0.036347877234220505,
|
| 116 |
+
"rewards/rejected": -0.8642423748970032,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 344.3992919921875,
|
| 122 |
+
"learning_rate": 4.998613757348784e-05,
|
| 123 |
+
"log_odds_chosen": 0.2355690896511078,
|
| 124 |
+
"log_odds_ratio": -0.7713075876235962,
|
| 125 |
+
"logits/chosen": 221.82272338867188,
|
| 126 |
+
"logits/rejected": 217.92764282226562,
|
| 127 |
+
"logps/chosen": -1.5275888442993164,
|
| 128 |
+
"logps/rejected": -1.7404054403305054,
|
| 129 |
+
"loss": 73.3205,
|
| 130 |
+
"nll_loss": 1.7808992862701416,
|
| 131 |
+
"rewards/accuracies": 0.5,
|
| 132 |
+
"rewards/chosen": -0.7637944221496582,
|
| 133 |
+
"rewards/margins": 0.1064082607626915,
|
| 134 |
+
"rewards/rejected": -0.8702027201652527,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 205.59912109375,
|
| 140 |
+
"learning_rate": 4.990147841143462e-05,
|
| 141 |
+
"log_odds_chosen": 0.3138998746871948,
|
| 142 |
+
"log_odds_ratio": -0.7420972585678101,
|
| 143 |
+
"logits/chosen": 225.82821655273438,
|
| 144 |
+
"logits/rejected": 226.96774291992188,
|
| 145 |
+
"logps/chosen": -1.5016671419143677,
|
| 146 |
+
"logps/rejected": -1.7676680088043213,
|
| 147 |
+
"loss": 71.1531,
|
| 148 |
+
"nll_loss": 1.7955719232559204,
|
| 149 |
+
"rewards/accuracies": 0.574999988079071,
|
| 150 |
+
"rewards/chosen": -0.7508335709571838,
|
| 151 |
+
"rewards/margins": 0.1330004334449768,
|
| 152 |
+
"rewards/rejected": -0.8838340044021606,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 139.50970458984375,
|
| 158 |
+
"learning_rate": 4.97401218720448e-05,
|
| 159 |
+
"log_odds_chosen": 0.1954536736011505,
|
| 160 |
+
"log_odds_ratio": -0.72107994556427,
|
| 161 |
+
"logits/chosen": 208.2395782470703,
|
| 162 |
+
"logits/rejected": 205.417236328125,
|
| 163 |
+
"logps/chosen": -1.3215817213058472,
|
| 164 |
+
"logps/rejected": -1.5023378133773804,
|
| 165 |
+
"loss": 66.4276,
|
| 166 |
+
"nll_loss": 1.647528886795044,
|
| 167 |
+
"rewards/accuracies": 0.550000011920929,
|
| 168 |
+
"rewards/chosen": -0.6607908606529236,
|
| 169 |
+
"rewards/margins": 0.0903780609369278,
|
| 170 |
+
"rewards/rejected": -0.7511689066886902,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 77.09749603271484,
|
| 176 |
+
"learning_rate": 4.9502564938797946e-05,
|
| 177 |
+
"log_odds_chosen": 0.19976872205734253,
|
| 178 |
+
"log_odds_ratio": -0.7074006199836731,
|
| 179 |
+
"logits/chosen": 202.8539276123047,
|
| 180 |
+
"logits/rejected": 195.70155334472656,
|
| 181 |
+
"logps/chosen": -1.2350599765777588,
|
| 182 |
+
"logps/rejected": -1.4153530597686768,
|
| 183 |
+
"loss": 63.0281,
|
| 184 |
+
"nll_loss": 1.5506360530853271,
|
| 185 |
+
"rewards/accuracies": 0.574999988079071,
|
| 186 |
+
"rewards/chosen": -0.6175299882888794,
|
| 187 |
+
"rewards/margins": 0.0901465192437172,
|
| 188 |
+
"rewards/rejected": -0.7076765298843384,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 59.40414810180664,
|
| 194 |
+
"learning_rate": 4.918953929490768e-05,
|
| 195 |
+
"log_odds_chosen": 0.14897426962852478,
|
| 196 |
+
"log_odds_ratio": -0.7380977869033813,
|
| 197 |
+
"logits/chosen": 190.48110961914062,
|
| 198 |
+
"logits/rejected": 195.70455932617188,
|
| 199 |
+
"logps/chosen": -1.221083641052246,
|
| 200 |
+
"logps/rejected": -1.3429195880889893,
|
| 201 |
+
"loss": 62.9068,
|
| 202 |
+
"nll_loss": 1.543068528175354,
|
| 203 |
+
"rewards/accuracies": 0.518750011920929,
|
| 204 |
+
"rewards/chosen": -0.610541820526123,
|
| 205 |
+
"rewards/margins": 0.0609179362654686,
|
| 206 |
+
"rewards/rejected": -0.6714597940444946,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 126.07240295410156,
|
| 212 |
+
"learning_rate": 4.88020090697132e-05,
|
| 213 |
+
"log_odds_chosen": 0.15872475504875183,
|
| 214 |
+
"log_odds_ratio": -0.7120335102081299,
|
| 215 |
+
"logits/chosen": 186.23619079589844,
|
| 216 |
+
"logits/rejected": 179.56393432617188,
|
| 217 |
+
"logps/chosen": -1.2440166473388672,
|
| 218 |
+
"logps/rejected": -1.3791967630386353,
|
| 219 |
+
"loss": 63.2646,
|
| 220 |
+
"nll_loss": 1.5736169815063477,
|
| 221 |
+
"rewards/accuracies": 0.543749988079071,
|
| 222 |
+
"rewards/chosen": -0.6220083236694336,
|
| 223 |
+
"rewards/margins": 0.06759000569581985,
|
| 224 |
+
"rewards/rejected": -0.6895983815193176,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 139.16448974609375,
|
| 230 |
+
"learning_rate": 4.834116786912897e-05,
|
| 231 |
+
"log_odds_chosen": 0.18847045302391052,
|
| 232 |
+
"log_odds_ratio": -0.7004148364067078,
|
| 233 |
+
"logits/chosen": 180.8818359375,
|
| 234 |
+
"logits/rejected": 185.00570678710938,
|
| 235 |
+
"logps/chosen": -1.2013404369354248,
|
| 236 |
+
"logps/rejected": -1.3533903360366821,
|
| 237 |
+
"loss": 60.8734,
|
| 238 |
+
"nll_loss": 1.485313057899475,
|
| 239 |
+
"rewards/accuracies": 0.5625,
|
| 240 |
+
"rewards/chosen": -0.6006702184677124,
|
| 241 |
+
"rewards/margins": 0.07602496445178986,
|
| 242 |
+
"rewards/rejected": -0.6766951680183411,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 251.40016174316406,
|
| 248 |
+
"learning_rate": 4.7808435099299045e-05,
|
| 249 |
+
"log_odds_chosen": 0.18260684609413147,
|
| 250 |
+
"log_odds_ratio": -0.7129999995231628,
|
| 251 |
+
"logits/chosen": 182.959716796875,
|
| 252 |
+
"logits/rejected": 183.2157745361328,
|
| 253 |
+
"logps/chosen": -1.1487746238708496,
|
| 254 |
+
"logps/rejected": -1.2979029417037964,
|
| 255 |
+
"loss": 61.4203,
|
| 256 |
+
"nll_loss": 1.4665228128433228,
|
| 257 |
+
"rewards/accuracies": 0.543749988079071,
|
| 258 |
+
"rewards/chosen": -0.5743873119354248,
|
| 259 |
+
"rewards/margins": 0.07456417381763458,
|
| 260 |
+
"rewards/rejected": -0.6489514708518982,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 124.29930877685547,
|
| 266 |
+
"learning_rate": 4.720545159477922e-05,
|
| 267 |
+
"log_odds_chosen": 0.17588524520397186,
|
| 268 |
+
"log_odds_ratio": -0.7309717535972595,
|
| 269 |
+
"logits/chosen": 175.79225158691406,
|
| 270 |
+
"logits/rejected": 185.4251708984375,
|
| 271 |
+
"logps/chosen": -1.171048641204834,
|
| 272 |
+
"logps/rejected": -1.3039839267730713,
|
| 273 |
+
"loss": 60.8133,
|
| 274 |
+
"nll_loss": 1.5102540254592896,
|
| 275 |
+
"rewards/accuracies": 0.574999988079071,
|
| 276 |
+
"rewards/chosen": -0.585524320602417,
|
| 277 |
+
"rewards/margins": 0.06646756827831268,
|
| 278 |
+
"rewards/rejected": -0.6519919633865356,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 358.4803771972656,
|
| 284 |
+
"learning_rate": 4.653407456471222e-05,
|
| 285 |
+
"log_odds_chosen": 0.1115400418639183,
|
| 286 |
+
"log_odds_ratio": -0.7338749170303345,
|
| 287 |
+
"logits/chosen": 191.45970153808594,
|
| 288 |
+
"logits/rejected": 181.09698486328125,
|
| 289 |
+
"logps/chosen": -1.1183122396469116,
|
| 290 |
+
"logps/rejected": -1.2054791450500488,
|
| 291 |
+
"loss": 57.9734,
|
| 292 |
+
"nll_loss": 1.4206385612487793,
|
| 293 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 294 |
+
"rewards/chosen": -0.5591561198234558,
|
| 295 |
+
"rewards/margins": 0.04358343034982681,
|
| 296 |
+
"rewards/rejected": -0.6027395725250244,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 136.4271697998047,
|
| 302 |
+
"learning_rate": 4.579637187256222e-05,
|
| 303 |
+
"log_odds_chosen": 0.04175865277647972,
|
| 304 |
+
"log_odds_ratio": -0.7415879964828491,
|
| 305 |
+
"logits/chosen": 192.60121154785156,
|
| 306 |
+
"logits/rejected": 192.03916931152344,
|
| 307 |
+
"logps/chosen": -1.0868916511535645,
|
| 308 |
+
"logps/rejected": -1.1372497081756592,
|
| 309 |
+
"loss": 59.0801,
|
| 310 |
+
"nll_loss": 1.4550647735595703,
|
| 311 |
+
"rewards/accuracies": 0.574999988079071,
|
| 312 |
+
"rewards/chosen": -0.5434458255767822,
|
| 313 |
+
"rewards/margins": 0.025179097428917885,
|
| 314 |
+
"rewards/rejected": -0.5686248540878296,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 56.09324645996094,
|
| 320 |
+
"learning_rate": 4.499461566702685e-05,
|
| 321 |
+
"log_odds_chosen": 0.12758071720600128,
|
| 322 |
+
"log_odds_ratio": -0.7168221473693848,
|
| 323 |
+
"logits/chosen": 182.20901489257812,
|
| 324 |
+
"logits/rejected": 187.005126953125,
|
| 325 |
+
"logps/chosen": -1.0412800312042236,
|
| 326 |
+
"logps/rejected": -1.140512466430664,
|
| 327 |
+
"loss": 56.929,
|
| 328 |
+
"nll_loss": 1.3427871465682983,
|
| 329 |
+
"rewards/accuracies": 0.5625,
|
| 330 |
+
"rewards/chosen": -0.5206400156021118,
|
| 331 |
+
"rewards/margins": 0.04961626976728439,
|
| 332 |
+
"rewards/rejected": -0.570256233215332,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 101.89226531982422,
|
| 338 |
+
"learning_rate": 4.413127538374411e-05,
|
| 339 |
+
"log_odds_chosen": 0.03433248773217201,
|
| 340 |
+
"log_odds_ratio": -0.7621655464172363,
|
| 341 |
+
"logits/chosen": 183.79061889648438,
|
| 342 |
+
"logits/rejected": 181.02096557617188,
|
| 343 |
+
"logps/chosen": -1.1271755695343018,
|
| 344 |
+
"logps/rejected": -1.155602216720581,
|
| 345 |
+
"loss": 57.6873,
|
| 346 |
+
"nll_loss": 1.4238755702972412,
|
| 347 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 348 |
+
"rewards/chosen": -0.5635877847671509,
|
| 349 |
+
"rewards/margins": 0.014213320799171925,
|
| 350 |
+
"rewards/rejected": -0.5778011083602905,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 56.19839859008789,
|
| 356 |
+
"learning_rate": 4.320901013934887e-05,
|
| 357 |
+
"log_odds_chosen": 0.19043493270874023,
|
| 358 |
+
"log_odds_ratio": -0.6898019313812256,
|
| 359 |
+
"logits/chosen": 187.52651977539062,
|
| 360 |
+
"logits/rejected": 188.4263458251953,
|
| 361 |
+
"logps/chosen": -1.064992904663086,
|
| 362 |
+
"logps/rejected": -1.2005788087844849,
|
| 363 |
+
"loss": 56.784,
|
| 364 |
+
"nll_loss": 1.3723514080047607,
|
| 365 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 366 |
+
"rewards/chosen": -0.532496452331543,
|
| 367 |
+
"rewards/margins": 0.06779297441244125,
|
| 368 |
+
"rewards/rejected": -0.6002894043922424,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 134.25718688964844,
|
| 374 |
+
"learning_rate": 4.223066054130568e-05,
|
| 375 |
+
"log_odds_chosen": 0.19098126888275146,
|
| 376 |
+
"log_odds_ratio": -0.708988606929779,
|
| 377 |
+
"logits/chosen": 186.099365234375,
|
| 378 |
+
"logits/rejected": 191.89564514160156,
|
| 379 |
+
"logps/chosen": -1.0854551792144775,
|
| 380 |
+
"logps/rejected": -1.2490203380584717,
|
| 381 |
+
"loss": 59.6264,
|
| 382 |
+
"nll_loss": 1.4756276607513428,
|
| 383 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 384 |
+
"rewards/chosen": -0.5427275896072388,
|
| 385 |
+
"rewards/margins": 0.0817825049161911,
|
| 386 |
+
"rewards/rejected": -0.6245101690292358,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.037925925925926,
|
| 391 |
+
"grad_norm": 48.896156311035156,
|
| 392 |
+
"learning_rate": 4.1199239938743797e-05,
|
| 393 |
+
"log_odds_chosen": 0.09143175929784775,
|
| 394 |
+
"log_odds_ratio": -0.7199470400810242,
|
| 395 |
+
"logits/chosen": 192.02879333496094,
|
| 396 |
+
"logits/rejected": 189.01097106933594,
|
| 397 |
+
"logps/chosen": -1.0210695266723633,
|
| 398 |
+
"logps/rejected": -1.0847587585449219,
|
| 399 |
+
"loss": 46.1645,
|
| 400 |
+
"nll_loss": 1.2595568895339966,
|
| 401 |
+
"rewards/accuracies": 0.5384615659713745,
|
| 402 |
+
"rewards/chosen": -0.5105347633361816,
|
| 403 |
+
"rewards/margins": 0.031844645738601685,
|
| 404 |
+
"rewards/rejected": -0.5423793792724609,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0853333333333333,
|
| 409 |
+
"grad_norm": 44.063316345214844,
|
| 410 |
+
"learning_rate": 4.0117925141242174e-05,
|
| 411 |
+
"log_odds_chosen": 0.09873426705598831,
|
| 412 |
+
"log_odds_ratio": -0.741472601890564,
|
| 413 |
+
"logits/chosen": 185.77395629882812,
|
| 414 |
+
"logits/rejected": 194.8834686279297,
|
| 415 |
+
"logps/chosen": -0.9893190264701843,
|
| 416 |
+
"logps/rejected": -1.0386199951171875,
|
| 417 |
+
"loss": 50.7825,
|
| 418 |
+
"nll_loss": 1.2114965915679932,
|
| 419 |
+
"rewards/accuracies": 0.550000011920929,
|
| 420 |
+
"rewards/chosen": -0.49465951323509216,
|
| 421 |
+
"rewards/margins": 0.024650435894727707,
|
| 422 |
+
"rewards/rejected": -0.5193099975585938,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1327407407407408,
|
| 427 |
+
"grad_norm": 42.30266571044922,
|
| 428 |
+
"learning_rate": 3.899004663415084e-05,
|
| 429 |
+
"log_odds_chosen": 0.03399781137704849,
|
| 430 |
+
"log_odds_ratio": -0.7742182612419128,
|
| 431 |
+
"logits/chosen": 182.58143615722656,
|
| 432 |
+
"logits/rejected": 186.83212280273438,
|
| 433 |
+
"logps/chosen": -0.9977661967277527,
|
| 434 |
+
"logps/rejected": -1.0048539638519287,
|
| 435 |
+
"loss": 51.3458,
|
| 436 |
+
"nll_loss": 1.2454302310943604,
|
| 437 |
+
"rewards/accuracies": 0.512499988079071,
|
| 438 |
+
"rewards/chosen": -0.49888309836387634,
|
| 439 |
+
"rewards/margins": 0.0035438979975879192,
|
| 440 |
+
"rewards/rejected": -0.5024269819259644,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1801481481481482,
|
| 445 |
+
"grad_norm": 125.85904693603516,
|
| 446 |
+
"learning_rate": 3.781907832058587e-05,
|
| 447 |
+
"log_odds_chosen": 0.09150873124599457,
|
| 448 |
+
"log_odds_ratio": -0.7233962416648865,
|
| 449 |
+
"logits/chosen": 181.49447631835938,
|
| 450 |
+
"logits/rejected": 180.0559844970703,
|
| 451 |
+
"logps/chosen": -0.9917734861373901,
|
| 452 |
+
"logps/rejected": -1.051242470741272,
|
| 453 |
+
"loss": 49.8693,
|
| 454 |
+
"nll_loss": 1.1629549264907837,
|
| 455 |
+
"rewards/accuracies": 0.59375,
|
| 456 |
+
"rewards/chosen": -0.49588674306869507,
|
| 457 |
+
"rewards/margins": 0.029734421521425247,
|
| 458 |
+
"rewards/rejected": -0.525621235370636,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2275555555555555,
|
| 463 |
+
"grad_norm": 39.95571517944336,
|
| 464 |
+
"learning_rate": 3.660862682169282e-05,
|
| 465 |
+
"log_odds_chosen": 0.09172184765338898,
|
| 466 |
+
"log_odds_ratio": -0.7451839447021484,
|
| 467 |
+
"logits/chosen": 179.13302612304688,
|
| 468 |
+
"logits/rejected": 183.4129638671875,
|
| 469 |
+
"logps/chosen": -1.0290215015411377,
|
| 470 |
+
"logps/rejected": -1.0855556726455688,
|
| 471 |
+
"loss": 51.0956,
|
| 472 |
+
"nll_loss": 1.1969176530838013,
|
| 473 |
+
"rewards/accuracies": 0.518750011920929,
|
| 474 |
+
"rewards/chosen": -0.5145107507705688,
|
| 475 |
+
"rewards/margins": 0.028267022222280502,
|
| 476 |
+
"rewards/rejected": -0.5427778363227844,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.274962962962963,
|
| 481 |
+
"grad_norm": 115.57888793945312,
|
| 482 |
+
"learning_rate": 3.5362420368134356e-05,
|
| 483 |
+
"log_odds_chosen": 0.07470552623271942,
|
| 484 |
+
"log_odds_ratio": -0.7265952825546265,
|
| 485 |
+
"logits/chosen": 183.11488342285156,
|
| 486 |
+
"logits/rejected": 177.27346801757812,
|
| 487 |
+
"logps/chosen": -0.9192112684249878,
|
| 488 |
+
"logps/rejected": -0.9482501745223999,
|
| 489 |
+
"loss": 49.3732,
|
| 490 |
+
"nll_loss": 1.1410008668899536,
|
| 491 |
+
"rewards/accuracies": 0.574999988079071,
|
| 492 |
+
"rewards/chosen": -0.4596056342124939,
|
| 493 |
+
"rewards/margins": 0.0145194623619318,
|
| 494 |
+
"rewards/rejected": -0.47412508726119995,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3223703703703704,
|
| 499 |
+
"grad_norm": 40.40840148925781,
|
| 500 |
+
"learning_rate": 3.408429731701635e-05,
|
| 501 |
+
"log_odds_chosen": 0.15658657252788544,
|
| 502 |
+
"log_odds_ratio": -0.7292534112930298,
|
| 503 |
+
"logits/chosen": 183.2898712158203,
|
| 504 |
+
"logits/rejected": 183.63088989257812,
|
| 505 |
+
"logps/chosen": -0.9870889782905579,
|
| 506 |
+
"logps/rejected": -1.0798033475875854,
|
| 507 |
+
"loss": 51.391,
|
| 508 |
+
"nll_loss": 1.1972719430923462,
|
| 509 |
+
"rewards/accuracies": 0.5625,
|
| 510 |
+
"rewards/chosen": -0.49354448914527893,
|
| 511 |
+
"rewards/margins": 0.046357229351997375,
|
| 512 |
+
"rewards/rejected": -0.5399016737937927,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.3697777777777778,
|
| 517 |
+
"grad_norm": 43.5965461730957,
|
| 518 |
+
"learning_rate": 3.2778194329621104e-05,
|
| 519 |
+
"log_odds_chosen": 0.0713091716170311,
|
| 520 |
+
"log_odds_ratio": -0.7587572932243347,
|
| 521 |
+
"logits/chosen": 175.42311096191406,
|
| 522 |
+
"logits/rejected": 179.28359985351562,
|
| 523 |
+
"logps/chosen": -0.9829910397529602,
|
| 524 |
+
"logps/rejected": -1.0261934995651245,
|
| 525 |
+
"loss": 51.113,
|
| 526 |
+
"nll_loss": 1.2084944248199463,
|
| 527 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 528 |
+
"rewards/chosen": -0.4914955198764801,
|
| 529 |
+
"rewards/margins": 0.021601270884275436,
|
| 530 |
+
"rewards/rejected": -0.5130967497825623,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.417185185185185,
|
| 535 |
+
"grad_norm": 68.55988311767578,
|
| 536 |
+
"learning_rate": 3.144813424636031e-05,
|
| 537 |
+
"log_odds_chosen": 0.03821509703993797,
|
| 538 |
+
"log_odds_ratio": -0.7714879512786865,
|
| 539 |
+
"logits/chosen": 173.73973083496094,
|
| 540 |
+
"logits/rejected": 177.93600463867188,
|
| 541 |
+
"logps/chosen": -1.0191147327423096,
|
| 542 |
+
"logps/rejected": -1.0414360761642456,
|
| 543 |
+
"loss": 49.2398,
|
| 544 |
+
"nll_loss": 1.162936806678772,
|
| 545 |
+
"rewards/accuracies": 0.46875,
|
| 546 |
+
"rewards/chosen": -0.5095573663711548,
|
| 547 |
+
"rewards/margins": 0.011160662397742271,
|
| 548 |
+
"rewards/rejected": -0.5207180380821228,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4645925925925927,
|
| 553 |
+
"grad_norm": 87.03449249267578,
|
| 554 |
+
"learning_rate": 3.0098213696293542e-05,
|
| 555 |
+
"log_odds_chosen": 0.10836371034383774,
|
| 556 |
+
"log_odds_ratio": -0.7314402461051941,
|
| 557 |
+
"logits/chosen": 175.88619995117188,
|
| 558 |
+
"logits/rejected": 182.52371215820312,
|
| 559 |
+
"logps/chosen": -0.9865490794181824,
|
| 560 |
+
"logps/rejected": -1.0499975681304932,
|
| 561 |
+
"loss": 50.2605,
|
| 562 |
+
"nll_loss": 1.1926219463348389,
|
| 563 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 564 |
+
"rewards/chosen": -0.4932745397090912,
|
| 565 |
+
"rewards/margins": 0.03172413632273674,
|
| 566 |
+
"rewards/rejected": -0.5249987840652466,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.512,
|
| 571 |
+
"grad_norm": 58.0168342590332,
|
| 572 |
+
"learning_rate": 2.8732590479375165e-05,
|
| 573 |
+
"log_odds_chosen": 0.12078354507684708,
|
| 574 |
+
"log_odds_ratio": -0.7301235795021057,
|
| 575 |
+
"logits/chosen": 173.8167266845703,
|
| 576 |
+
"logits/rejected": 181.9156494140625,
|
| 577 |
+
"logps/chosen": -0.9404324293136597,
|
| 578 |
+
"logps/rejected": -1.0260807275772095,
|
| 579 |
+
"loss": 48.9856,
|
| 580 |
+
"nll_loss": 1.1453100442886353,
|
| 581 |
+
"rewards/accuracies": 0.512499988079071,
|
| 582 |
+
"rewards/chosen": -0.47021621465682983,
|
| 583 |
+
"rewards/margins": 0.0428241565823555,
|
| 584 |
+
"rewards/rejected": -0.5130403637886047,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5594074074074074,
|
| 589 |
+
"grad_norm": 47.982913970947266,
|
| 590 |
+
"learning_rate": 2.7355470760292956e-05,
|
| 591 |
+
"log_odds_chosen": 0.0946766808629036,
|
| 592 |
+
"log_odds_ratio": -0.7334403395652771,
|
| 593 |
+
"logits/chosen": 166.52523803710938,
|
| 594 |
+
"logits/rejected": 171.69485473632812,
|
| 595 |
+
"logps/chosen": -0.9203466176986694,
|
| 596 |
+
"logps/rejected": -0.9718505144119263,
|
| 597 |
+
"loss": 50.1007,
|
| 598 |
+
"nll_loss": 1.1756643056869507,
|
| 599 |
+
"rewards/accuracies": 0.543749988079071,
|
| 600 |
+
"rewards/chosen": -0.4601733088493347,
|
| 601 |
+
"rewards/margins": 0.025751952081918716,
|
| 602 |
+
"rewards/rejected": -0.48592525720596313,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6068148148148147,
|
| 607 |
+
"grad_norm": 50.041561126708984,
|
| 608 |
+
"learning_rate": 2.597109611334169e-05,
|
| 609 |
+
"log_odds_chosen": 0.17343485355377197,
|
| 610 |
+
"log_odds_ratio": -0.6794081926345825,
|
| 611 |
+
"logits/chosen": 171.59066772460938,
|
| 612 |
+
"logits/rejected": 179.27670288085938,
|
| 613 |
+
"logps/chosen": -0.936198353767395,
|
| 614 |
+
"logps/rejected": -1.0650147199630737,
|
| 615 |
+
"loss": 49.7789,
|
| 616 |
+
"nll_loss": 1.186452031135559,
|
| 617 |
+
"rewards/accuracies": 0.574999988079071,
|
| 618 |
+
"rewards/chosen": -0.4680991768836975,
|
| 619 |
+
"rewards/margins": 0.06440822780132294,
|
| 620 |
+
"rewards/rejected": -0.5325073599815369,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6542222222222223,
|
| 625 |
+
"grad_norm": 47.45258331298828,
|
| 626 |
+
"learning_rate": 2.458373045823404e-05,
|
| 627 |
+
"log_odds_chosen": 0.07308103144168854,
|
| 628 |
+
"log_odds_ratio": -0.7454826235771179,
|
| 629 |
+
"logits/chosen": 172.97149658203125,
|
| 630 |
+
"logits/rejected": 172.43182373046875,
|
| 631 |
+
"logps/chosen": -0.9455814361572266,
|
| 632 |
+
"logps/rejected": -1.0050225257873535,
|
| 633 |
+
"loss": 49.1083,
|
| 634 |
+
"nll_loss": 1.1290260553359985,
|
| 635 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 636 |
+
"rewards/chosen": -0.4727907180786133,
|
| 637 |
+
"rewards/margins": 0.029720569029450417,
|
| 638 |
+
"rewards/rejected": -0.5025112628936768,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7016296296296296,
|
| 643 |
+
"grad_norm": 58.00626754760742,
|
| 644 |
+
"learning_rate": 2.3197646927086697e-05,
|
| 645 |
+
"log_odds_chosen": 0.1577189862728119,
|
| 646 |
+
"log_odds_ratio": -0.7023075819015503,
|
| 647 |
+
"logits/chosen": 168.512451171875,
|
| 648 |
+
"logits/rejected": 175.3289794921875,
|
| 649 |
+
"logps/chosen": -0.9105680584907532,
|
| 650 |
+
"logps/rejected": -1.0289887189865112,
|
| 651 |
+
"loss": 49.2958,
|
| 652 |
+
"nll_loss": 1.1708818674087524,
|
| 653 |
+
"rewards/accuracies": 0.53125,
|
| 654 |
+
"rewards/chosen": -0.4552840292453766,
|
| 655 |
+
"rewards/margins": 0.059210360050201416,
|
| 656 |
+
"rewards/rejected": -0.5144943594932556,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.749037037037037,
|
| 661 |
+
"grad_norm": 34.98508071899414,
|
| 662 |
+
"learning_rate": 2.1817114703032176e-05,
|
| 663 |
+
"log_odds_chosen": 0.12351224571466446,
|
| 664 |
+
"log_odds_ratio": -0.7270097732543945,
|
| 665 |
+
"logits/chosen": 173.38623046875,
|
| 666 |
+
"logits/rejected": 165.6890869140625,
|
| 667 |
+
"logps/chosen": -0.9919120669364929,
|
| 668 |
+
"logps/rejected": -1.0763611793518066,
|
| 669 |
+
"loss": 48.9183,
|
| 670 |
+
"nll_loss": 1.1456924676895142,
|
| 671 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 672 |
+
"rewards/chosen": -0.49595603346824646,
|
| 673 |
+
"rewards/margins": 0.04222451522946358,
|
| 674 |
+
"rewards/rejected": -0.5381805896759033,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.7964444444444445,
|
| 679 |
+
"grad_norm": 35.57364273071289,
|
| 680 |
+
"learning_rate": 2.0446385870993467e-05,
|
| 681 |
+
"log_odds_chosen": 0.046650804579257965,
|
| 682 |
+
"log_odds_ratio": -0.7537095546722412,
|
| 683 |
+
"logits/chosen": 161.9040069580078,
|
| 684 |
+
"logits/rejected": 164.87686157226562,
|
| 685 |
+
"logps/chosen": -0.9197586178779602,
|
| 686 |
+
"logps/rejected": -0.9383406639099121,
|
| 687 |
+
"loss": 49.163,
|
| 688 |
+
"nll_loss": 1.1618720293045044,
|
| 689 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 690 |
+
"rewards/chosen": -0.4598793089389801,
|
| 691 |
+
"rewards/margins": 0.009291025809943676,
|
| 692 |
+
"rewards/rejected": -0.46917033195495605,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8438518518518519,
|
| 697 |
+
"grad_norm": 57.462890625,
|
| 698 |
+
"learning_rate": 1.9089682321121834e-05,
|
| 699 |
+
"log_odds_chosen": 0.10921354591846466,
|
| 700 |
+
"log_odds_ratio": -0.7022596001625061,
|
| 701 |
+
"logits/chosen": 164.3706817626953,
|
| 702 |
+
"logits/rejected": 167.53907775878906,
|
| 703 |
+
"logps/chosen": -0.9306834936141968,
|
| 704 |
+
"logps/rejected": -1.0057896375656128,
|
| 705 |
+
"loss": 48.8493,
|
| 706 |
+
"nll_loss": 1.1373628377914429,
|
| 707 |
+
"rewards/accuracies": 0.550000011920929,
|
| 708 |
+
"rewards/chosen": -0.4653417468070984,
|
| 709 |
+
"rewards/margins": 0.03755306452512741,
|
| 710 |
+
"rewards/rejected": -0.5028948187828064,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8912592592592592,
|
| 715 |
+
"grad_norm": 51.58010482788086,
|
| 716 |
+
"learning_rate": 1.775118274523545e-05,
|
| 717 |
+
"log_odds_chosen": 0.0803312212228775,
|
| 718 |
+
"log_odds_ratio": -0.7448180913925171,
|
| 719 |
+
"logits/chosen": 177.42755126953125,
|
| 720 |
+
"logits/rejected": 166.17420959472656,
|
| 721 |
+
"logps/chosen": -0.971086323261261,
|
| 722 |
+
"logps/rejected": -1.025727391242981,
|
| 723 |
+
"loss": 50.0697,
|
| 724 |
+
"nll_loss": 1.1380794048309326,
|
| 725 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 726 |
+
"rewards/chosen": -0.4855431616306305,
|
| 727 |
+
"rewards/margins": 0.027320479974150658,
|
| 728 |
+
"rewards/rejected": -0.5128636956214905,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9386666666666668,
|
| 733 |
+
"grad_norm": 40.688987731933594,
|
| 734 |
+
"learning_rate": 1.643500976631037e-05,
|
| 735 |
+
"log_odds_chosen": 0.10912259668111801,
|
| 736 |
+
"log_odds_ratio": -0.7380350232124329,
|
| 737 |
+
"logits/chosen": 173.36325073242188,
|
| 738 |
+
"logits/rejected": 171.0248565673828,
|
| 739 |
+
"logps/chosen": -0.9703338742256165,
|
| 740 |
+
"logps/rejected": -1.0370721817016602,
|
| 741 |
+
"loss": 48.6517,
|
| 742 |
+
"nll_loss": 1.1243274211883545,
|
| 743 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 744 |
+
"rewards/chosen": -0.4851669371128082,
|
| 745 |
+
"rewards/margins": 0.03336922079324722,
|
| 746 |
+
"rewards/rejected": -0.5185360908508301,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.986074074074074,
|
| 751 |
+
"grad_norm": 47.03554153442383,
|
| 752 |
+
"learning_rate": 1.514521724066537e-05,
|
| 753 |
+
"log_odds_chosen": 0.07311935722827911,
|
| 754 |
+
"log_odds_ratio": -0.7410269975662231,
|
| 755 |
+
"logits/chosen": 161.46792602539062,
|
| 756 |
+
"logits/rejected": 176.0153045654297,
|
| 757 |
+
"logps/chosen": -0.9304733276367188,
|
| 758 |
+
"logps/rejected": -0.9752525091171265,
|
| 759 |
+
"loss": 49.3022,
|
| 760 |
+
"nll_loss": 1.163869857788086,
|
| 761 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 762 |
+
"rewards/chosen": -0.4652366638183594,
|
| 763 |
+
"rewards/margins": 0.02238963544368744,
|
| 764 |
+
"rewards/rejected": -0.48762625455856323,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0284444444444443,
|
| 769 |
+
"grad_norm": 46.73811721801758,
|
| 770 |
+
"learning_rate": 1.3885777771950348e-05,
|
| 771 |
+
"log_odds_chosen": 0.04928523674607277,
|
| 772 |
+
"log_odds_ratio": -0.7592554688453674,
|
| 773 |
+
"logits/chosen": 154.12142944335938,
|
| 774 |
+
"logits/rejected": 166.54530334472656,
|
| 775 |
+
"logps/chosen": -0.8669578433036804,
|
| 776 |
+
"logps/rejected": -0.8768206238746643,
|
| 777 |
+
"loss": 38.0486,
|
| 778 |
+
"nll_loss": 0.969521701335907,
|
| 779 |
+
"rewards/accuracies": 0.5314685106277466,
|
| 780 |
+
"rewards/chosen": -0.4334789216518402,
|
| 781 |
+
"rewards/margins": 0.004931411240249872,
|
| 782 |
+
"rewards/rejected": -0.43841031193733215,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.075851851851852,
|
| 787 |
+
"grad_norm": 66.5871353149414,
|
| 788 |
+
"learning_rate": 1.2660570475395683e-05,
|
| 789 |
+
"log_odds_chosen": 0.07944320142269135,
|
| 790 |
+
"log_odds_ratio": -0.7612829804420471,
|
| 791 |
+
"logits/chosen": 158.1134033203125,
|
| 792 |
+
"logits/rejected": 145.2221221923828,
|
| 793 |
+
"logps/chosen": -0.8715459704399109,
|
| 794 |
+
"logps/rejected": -0.8965581059455872,
|
| 795 |
+
"loss": 42.355,
|
| 796 |
+
"nll_loss": 0.9041290283203125,
|
| 797 |
+
"rewards/accuracies": 0.606249988079071,
|
| 798 |
+
"rewards/chosen": -0.43577298521995544,
|
| 799 |
+
"rewards/margins": 0.01250611525028944,
|
| 800 |
+
"rewards/rejected": -0.4482790529727936,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1232592592592594,
|
| 805 |
+
"grad_norm": 43.993412017822266,
|
| 806 |
+
"learning_rate": 1.1473369030008974e-05,
|
| 807 |
+
"log_odds_chosen": 0.07922948896884918,
|
| 808 |
+
"log_odds_ratio": -0.748726487159729,
|
| 809 |
+
"logits/chosen": 150.79869079589844,
|
| 810 |
+
"logits/rejected": 159.51519775390625,
|
| 811 |
+
"logps/chosen": -0.9018028974533081,
|
| 812 |
+
"logps/rejected": -0.9198867082595825,
|
| 813 |
+
"loss": 41.5811,
|
| 814 |
+
"nll_loss": 0.9309173822402954,
|
| 815 |
+
"rewards/accuracies": 0.550000011920929,
|
| 816 |
+
"rewards/chosen": -0.45090144872665405,
|
| 817 |
+
"rewards/margins": 0.009041833691298962,
|
| 818 |
+
"rewards/rejected": -0.45994335412979126,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.1706666666666665,
|
| 823 |
+
"grad_norm": 42.56783676147461,
|
| 824 |
+
"learning_rate": 1.0327830055518842e-05,
|
| 825 |
+
"log_odds_chosen": 0.1404309719800949,
|
| 826 |
+
"log_odds_ratio": -0.7541959285736084,
|
| 827 |
+
"logits/chosen": 149.96189880371094,
|
| 828 |
+
"logits/rejected": 155.0712127685547,
|
| 829 |
+
"logps/chosen": -0.8312407732009888,
|
| 830 |
+
"logps/rejected": -0.9046605825424194,
|
| 831 |
+
"loss": 41.5919,
|
| 832 |
+
"nll_loss": 0.9145228266716003,
|
| 833 |
+
"rewards/accuracies": 0.59375,
|
| 834 |
+
"rewards/chosen": -0.4156203866004944,
|
| 835 |
+
"rewards/margins": 0.03670995682477951,
|
| 836 |
+
"rewards/rejected": -0.4523302912712097,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.218074074074074,
|
| 841 |
+
"grad_norm": 46.45758819580078,
|
| 842 |
+
"learning_rate": 9.227481849865235e-06,
|
| 843 |
+
"log_odds_chosen": -0.03481599688529968,
|
| 844 |
+
"log_odds_ratio": -0.8210490942001343,
|
| 845 |
+
"logits/chosen": 156.96890258789062,
|
| 846 |
+
"logits/rejected": 156.92953491210938,
|
| 847 |
+
"logps/chosen": -0.899451732635498,
|
| 848 |
+
"logps/rejected": -0.8785243034362793,
|
| 849 |
+
"loss": 42.667,
|
| 850 |
+
"nll_loss": 0.988889217376709,
|
| 851 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 852 |
+
"rewards/chosen": -0.449725866317749,
|
| 853 |
+
"rewards/margins": -0.010463694110512733,
|
| 854 |
+
"rewards/rejected": -0.43926215171813965,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2654814814814817,
|
| 859 |
+
"grad_norm": 41.77062225341797,
|
| 860 |
+
"learning_rate": 8.175713521924978e-06,
|
| 861 |
+
"log_odds_chosen": 0.19800688326358795,
|
| 862 |
+
"log_odds_ratio": -0.7289772033691406,
|
| 863 |
+
"logits/chosen": 161.128173828125,
|
| 864 |
+
"logits/rejected": 159.4349365234375,
|
| 865 |
+
"logps/chosen": -0.8593847155570984,
|
| 866 |
+
"logps/rejected": -0.9744852185249329,
|
| 867 |
+
"loss": 41.5045,
|
| 868 |
+
"nll_loss": 0.9035336375236511,
|
| 869 |
+
"rewards/accuracies": 0.612500011920929,
|
| 870 |
+
"rewards/chosen": -0.4296923577785492,
|
| 871 |
+
"rewards/margins": 0.05755022168159485,
|
| 872 |
+
"rewards/rejected": -0.48724260926246643,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.3128888888888888,
|
| 877 |
+
"grad_norm": 41.01064682006836,
|
| 878 |
+
"learning_rate": 7.1757645529443665e-06,
|
| 879 |
+
"log_odds_chosen": -0.0014010012382641435,
|
| 880 |
+
"log_odds_ratio": -0.7947742342948914,
|
| 881 |
+
"logits/chosen": 155.06546020507812,
|
| 882 |
+
"logits/rejected": 155.10775756835938,
|
| 883 |
+
"logps/chosen": -0.878911018371582,
|
| 884 |
+
"logps/rejected": -0.8687770962715149,
|
| 885 |
+
"loss": 41.6816,
|
| 886 |
+
"nll_loss": 0.9432564973831177,
|
| 887 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 888 |
+
"rewards/chosen": -0.439455509185791,
|
| 889 |
+
"rewards/margins": -0.0050669461488723755,
|
| 890 |
+
"rewards/rejected": -0.43438854813575745,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3602962962962963,
|
| 895 |
+
"grad_norm": 41.34367752075195,
|
| 896 |
+
"learning_rate": 6.230714818829733e-06,
|
| 897 |
+
"log_odds_chosen": 0.0013508498668670654,
|
| 898 |
+
"log_odds_ratio": -0.7936294674873352,
|
| 899 |
+
"logits/chosen": 153.34498596191406,
|
| 900 |
+
"logits/rejected": 158.2041015625,
|
| 901 |
+
"logps/chosen": -0.8574392199516296,
|
| 902 |
+
"logps/rejected": -0.8317993879318237,
|
| 903 |
+
"loss": 42.0429,
|
| 904 |
+
"nll_loss": 0.932290256023407,
|
| 905 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 906 |
+
"rewards/chosen": -0.4287196099758148,
|
| 907 |
+
"rewards/margins": -0.0128199253231287,
|
| 908 |
+
"rewards/rejected": -0.41589969396591187,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.407703703703704,
|
| 913 |
+
"grad_norm": 37.83774185180664,
|
| 914 |
+
"learning_rate": 5.343475104027743e-06,
|
| 915 |
+
"log_odds_chosen": 0.17812205851078033,
|
| 916 |
+
"log_odds_ratio": -0.7299228310585022,
|
| 917 |
+
"logits/chosen": 155.40672302246094,
|
| 918 |
+
"logits/rejected": 160.87484741210938,
|
| 919 |
+
"logps/chosen": -0.8623374104499817,
|
| 920 |
+
"logps/rejected": -0.9512872695922852,
|
| 921 |
+
"loss": 42.1891,
|
| 922 |
+
"nll_loss": 0.9383618235588074,
|
| 923 |
+
"rewards/accuracies": 0.5625,
|
| 924 |
+
"rewards/chosen": -0.43116870522499084,
|
| 925 |
+
"rewards/margins": 0.044474903494119644,
|
| 926 |
+
"rewards/rejected": -0.4756436347961426,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.455111111111111,
|
| 931 |
+
"grad_norm": 34.86994934082031,
|
| 932 |
+
"learning_rate": 4.516778136213037e-06,
|
| 933 |
+
"log_odds_chosen": 0.06448222696781158,
|
| 934 |
+
"log_odds_ratio": -0.7623555660247803,
|
| 935 |
+
"logits/chosen": 159.42433166503906,
|
| 936 |
+
"logits/rejected": 155.6562957763672,
|
| 937 |
+
"logps/chosen": -0.853624701499939,
|
| 938 |
+
"logps/rejected": -0.8752864003181458,
|
| 939 |
+
"loss": 41.3101,
|
| 940 |
+
"nll_loss": 0.8972175717353821,
|
| 941 |
+
"rewards/accuracies": 0.5625,
|
| 942 |
+
"rewards/chosen": -0.4268123507499695,
|
| 943 |
+
"rewards/margins": 0.010830795392394066,
|
| 944 |
+
"rewards/rejected": -0.4376432001590729,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5025185185185186,
|
| 949 |
+
"grad_norm": 40.81254577636719,
|
| 950 |
+
"learning_rate": 3.7531701693965554e-06,
|
| 951 |
+
"log_odds_chosen": 0.0020365030504763126,
|
| 952 |
+
"log_odds_ratio": -0.7836844325065613,
|
| 953 |
+
"logits/chosen": 152.5310516357422,
|
| 954 |
+
"logits/rejected": 162.28890991210938,
|
| 955 |
+
"logps/chosen": -0.904189944267273,
|
| 956 |
+
"logps/rejected": -0.8998059034347534,
|
| 957 |
+
"loss": 41.5474,
|
| 958 |
+
"nll_loss": 0.9223392605781555,
|
| 959 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 960 |
+
"rewards/chosen": -0.4520949721336365,
|
| 961 |
+
"rewards/margins": -0.002192039042711258,
|
| 962 |
+
"rewards/rejected": -0.4499029517173767,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.549925925925926,
|
| 967 |
+
"grad_norm": 35.19020080566406,
|
| 968 |
+
"learning_rate": 3.055003141378948e-06,
|
| 969 |
+
"log_odds_chosen": 0.08025786280632019,
|
| 970 |
+
"log_odds_ratio": -0.7740025520324707,
|
| 971 |
+
"logits/chosen": 152.10177612304688,
|
| 972 |
+
"logits/rejected": 160.26998901367188,
|
| 973 |
+
"logps/chosen": -0.8461346626281738,
|
| 974 |
+
"logps/rejected": -0.8950198888778687,
|
| 975 |
+
"loss": 41.2177,
|
| 976 |
+
"nll_loss": 0.8989561200141907,
|
| 977 |
+
"rewards/accuracies": 0.543749988079071,
|
| 978 |
+
"rewards/chosen": -0.4230673313140869,
|
| 979 |
+
"rewards/margins": 0.02444261871278286,
|
| 980 |
+
"rewards/rejected": -0.4475099444389343,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.5973333333333333,
|
| 985 |
+
"grad_norm": 39.02735900878906,
|
| 986 |
+
"learning_rate": 2.424427429704365e-06,
|
| 987 |
+
"log_odds_chosen": 0.061826206743717194,
|
| 988 |
+
"log_odds_ratio": -0.7568526268005371,
|
| 989 |
+
"logits/chosen": 161.21261596679688,
|
| 990 |
+
"logits/rejected": 161.19491577148438,
|
| 991 |
+
"logps/chosen": -0.8993644714355469,
|
| 992 |
+
"logps/rejected": -0.903190016746521,
|
| 993 |
+
"loss": 41.1732,
|
| 994 |
+
"nll_loss": 0.8910198211669922,
|
| 995 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 996 |
+
"rewards/chosen": -0.44968223571777344,
|
| 997 |
+
"rewards/margins": 0.0019127646228298545,
|
| 998 |
+
"rewards/rejected": -0.4515950083732605,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.644740740740741,
|
| 1003 |
+
"grad_norm": 39.4185676574707,
|
| 1004 |
+
"learning_rate": 1.8633852284264508e-06,
|
| 1005 |
+
"log_odds_chosen": 0.02886720933020115,
|
| 1006 |
+
"log_odds_ratio": -0.8128382563591003,
|
| 1007 |
+
"logits/chosen": 153.80528259277344,
|
| 1008 |
+
"logits/rejected": 150.25201416015625,
|
| 1009 |
+
"logps/chosen": -0.8430732488632202,
|
| 1010 |
+
"logps/rejected": -0.8419806361198425,
|
| 1011 |
+
"loss": 41.047,
|
| 1012 |
+
"nll_loss": 0.9150816202163696,
|
| 1013 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 1014 |
+
"rewards/chosen": -0.4215366244316101,
|
| 1015 |
+
"rewards/margins": -0.0005463186535052955,
|
| 1016 |
+
"rewards/rejected": -0.42099031805992126,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.6921481481481484,
|
| 1021 |
+
"grad_norm": 50.44974899291992,
|
| 1022 |
+
"learning_rate": 1.3736045660864034e-06,
|
| 1023 |
+
"log_odds_chosen": -0.17111532390117645,
|
| 1024 |
+
"log_odds_ratio": -0.9221655130386353,
|
| 1025 |
+
"logits/chosen": 152.35177612304688,
|
| 1026 |
+
"logits/rejected": 156.8704376220703,
|
| 1027 |
+
"logps/chosen": -1.0376865863800049,
|
| 1028 |
+
"logps/rejected": -0.9074798822402954,
|
| 1029 |
+
"loss": 42.497,
|
| 1030 |
+
"nll_loss": 0.9657168388366699,
|
| 1031 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 1032 |
+
"rewards/chosen": -0.5188432931900024,
|
| 1033 |
+
"rewards/margins": -0.06510341912508011,
|
| 1034 |
+
"rewards/rejected": -0.4537399411201477,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.7395555555555555,
|
| 1039 |
+
"grad_norm": 38.56003952026367,
|
| 1040 |
+
"learning_rate": 9.565939833279192e-07,
|
| 1041 |
+
"log_odds_chosen": -0.0652594044804573,
|
| 1042 |
+
"log_odds_ratio": -0.8579480051994324,
|
| 1043 |
+
"logits/chosen": 155.33377075195312,
|
| 1044 |
+
"logits/rejected": 156.094482421875,
|
| 1045 |
+
"logps/chosen": -0.9537650942802429,
|
| 1046 |
+
"logps/rejected": -0.8857321739196777,
|
| 1047 |
+
"loss": 41.9732,
|
| 1048 |
+
"nll_loss": 0.9463297724723816,
|
| 1049 |
+
"rewards/accuracies": 0.512499988079071,
|
| 1050 |
+
"rewards/chosen": -0.47688254714012146,
|
| 1051 |
+
"rewards/margins": -0.034016452729701996,
|
| 1052 |
+
"rewards/rejected": -0.44286608695983887,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.786962962962963,
|
| 1057 |
+
"grad_norm": 35.984092712402344,
|
| 1058 |
+
"learning_rate": 6.136378865420872e-07,
|
| 1059 |
+
"log_odds_chosen": -0.05488128587603569,
|
| 1060 |
+
"log_odds_ratio": -0.8374530076980591,
|
| 1061 |
+
"logits/chosen": 159.79251098632812,
|
| 1062 |
+
"logits/rejected": 153.42626953125,
|
| 1063 |
+
"logps/chosen": -0.8930980563163757,
|
| 1064 |
+
"logps/rejected": -0.848624587059021,
|
| 1065 |
+
"loss": 42.7359,
|
| 1066 |
+
"nll_loss": 0.9567394256591797,
|
| 1067 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 1068 |
+
"rewards/chosen": -0.44654902815818787,
|
| 1069 |
+
"rewards/margins": -0.02223675511777401,
|
| 1070 |
+
"rewards/rejected": -0.4243122935295105,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.83437037037037,
|
| 1075 |
+
"grad_norm": 40.65980911254883,
|
| 1076 |
+
"learning_rate": 3.45792591853214e-07,
|
| 1077 |
+
"log_odds_chosen": 0.03835665062069893,
|
| 1078 |
+
"log_odds_ratio": -0.7734814286231995,
|
| 1079 |
+
"logits/chosen": 157.85092163085938,
|
| 1080 |
+
"logits/rejected": 154.9134979248047,
|
| 1081 |
+
"logps/chosen": -0.8622520565986633,
|
| 1082 |
+
"logps/rejected": -0.8551470637321472,
|
| 1083 |
+
"loss": 41.1462,
|
| 1084 |
+
"nll_loss": 0.9184630513191223,
|
| 1085 |
+
"rewards/accuracies": 0.543749988079071,
|
| 1086 |
+
"rewards/chosen": -0.43112602829933167,
|
| 1087 |
+
"rewards/margins": -0.0035525336861610413,
|
| 1088 |
+
"rewards/rejected": -0.4275735318660736,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.8817777777777778,
|
| 1093 |
+
"grad_norm": 45.03302001953125,
|
| 1094 |
+
"learning_rate": 1.538830716302092e-07,
|
| 1095 |
+
"log_odds_chosen": 0.0824732631444931,
|
| 1096 |
+
"log_odds_ratio": -0.7868438959121704,
|
| 1097 |
+
"logits/chosen": 152.17709350585938,
|
| 1098 |
+
"logits/rejected": 160.5044403076172,
|
| 1099 |
+
"logps/chosen": -0.8794566988945007,
|
| 1100 |
+
"logps/rejected": -0.887550950050354,
|
| 1101 |
+
"loss": 40.8093,
|
| 1102 |
+
"nll_loss": 0.9352946281433105,
|
| 1103 |
+
"rewards/accuracies": 0.543749988079071,
|
| 1104 |
+
"rewards/chosen": -0.43972834944725037,
|
| 1105 |
+
"rewards/margins": 0.004047115799039602,
|
| 1106 |
+
"rewards/rejected": -0.443775475025177,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9291851851851853,
|
| 1111 |
+
"grad_norm": 41.13324737548828,
|
| 1112 |
+
"learning_rate": 3.8500413544415025e-08,
|
| 1113 |
+
"log_odds_chosen": 0.036232512444257736,
|
| 1114 |
+
"log_odds_ratio": -0.7697147727012634,
|
| 1115 |
+
"logits/chosen": 153.0935821533203,
|
| 1116 |
+
"logits/rejected": 152.86441040039062,
|
| 1117 |
+
"logps/chosen": -0.8729937672615051,
|
| 1118 |
+
"logps/rejected": -0.8775345087051392,
|
| 1119 |
+
"loss": 41.2786,
|
| 1120 |
+
"nll_loss": 0.9117900133132935,
|
| 1121 |
+
"rewards/accuracies": 0.59375,
|
| 1122 |
+
"rewards/chosen": -0.43649688363075256,
|
| 1123 |
+
"rewards/margins": 0.002270397264510393,
|
| 1124 |
+
"rewards/rejected": -0.4387672543525696,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.9765925925925925,
|
| 1129 |
+
"grad_norm": 116.33460235595703,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 0.13965412974357605,
|
| 1132 |
+
"log_odds_ratio": -0.712945282459259,
|
| 1133 |
+
"logits/chosen": 145.65011596679688,
|
| 1134 |
+
"logits/rejected": 152.03152465820312,
|
| 1135 |
+
"logps/chosen": -0.739302933216095,
|
| 1136 |
+
"logps/rejected": -0.7891243696212769,
|
| 1137 |
+
"loss": 39.5762,
|
| 1138 |
+
"nll_loss": 0.8406442403793335,
|
| 1139 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1140 |
+
"rewards/chosen": -0.3696514666080475,
|
| 1141 |
+
"rewards/margins": 0.024910712614655495,
|
| 1142 |
+
"rewards/rejected": -0.3945621848106384,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.9765925925925925,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 71.97097947862413,
|
| 1150 |
+
"train_runtime": 8116.2448,
|
| 1151 |
+
"train_samples_per_second": 2.495,
|
| 1152 |
+
"train_steps_per_second": 0.039
|
| 1153 |
+
}
|
| 1154 |
+
],
|
| 1155 |
+
"logging_steps": 5,
|
| 1156 |
+
"max_steps": 315,
|
| 1157 |
+
"num_input_tokens_seen": 0,
|
| 1158 |
+
"num_train_epochs": 3,
|
| 1159 |
+
"save_steps": 100000,
|
| 1160 |
+
"stateful_callbacks": {
|
| 1161 |
+
"TrainerControl": {
|
| 1162 |
+
"args": {
|
| 1163 |
+
"should_epoch_stop": false,
|
| 1164 |
+
"should_evaluate": false,
|
| 1165 |
+
"should_log": false,
|
| 1166 |
+
"should_save": true,
|
| 1167 |
+
"should_training_stop": true
|
| 1168 |
+
},
|
| 1169 |
+
"attributes": {}
|
| 1170 |
+
}
|
| 1171 |
+
},
|
| 1172 |
+
"total_flos": 0.0,
|
| 1173 |
+
"train_batch_size": 1,
|
| 1174 |
+
"trial_name": null,
|
| 1175 |
+
"trial_params": null
|
| 1176 |
+
}
|