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-simpo-noisy-5e-5
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-simpo-noisy-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-simpo-noisy-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/x11gsd3s)
|
| 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": 81.45431906079489,
|
| 5 |
+
"train_runtime": 8117.5608,
|
| 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": 81.45431906079489,
|
| 5 |
+
"train_runtime": 8117.5608,
|
| 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": 1084.8690185546875,
|
| 14 |
+
"learning_rate": 7.8125e-06,
|
| 15 |
+
"log_odds_chosen": 4.959512233734131,
|
| 16 |
+
"log_odds_ratio": -9.777482986450195,
|
| 17 |
+
"logits/chosen": 138.49520874023438,
|
| 18 |
+
"logits/rejected": 153.09359741210938,
|
| 19 |
+
"logps/chosen": -20.70966339111328,
|
| 20 |
+
"logps/rejected": -25.66940689086914,
|
| 21 |
+
"loss": 633.2478,
|
| 22 |
+
"nll_loss": 8.337706565856934,
|
| 23 |
+
"rewards/accuracies": 0.53125,
|
| 24 |
+
"rewards/chosen": -20.70966339111328,
|
| 25 |
+
"rewards/margins": 4.959742546081543,
|
| 26 |
+
"rewards/rejected": -25.66940689086914,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 765.169189453125,
|
| 32 |
+
"learning_rate": 1.5625e-05,
|
| 33 |
+
"log_odds_chosen": 2.3558285236358643,
|
| 34 |
+
"log_odds_ratio": -7.167288303375244,
|
| 35 |
+
"logits/chosen": 135.65057373046875,
|
| 36 |
+
"logits/rejected": 159.9241180419922,
|
| 37 |
+
"logps/chosen": -15.381937980651855,
|
| 38 |
+
"logps/rejected": -17.73668670654297,
|
| 39 |
+
"loss": 513.577,
|
| 40 |
+
"nll_loss": 7.278769493103027,
|
| 41 |
+
"rewards/accuracies": 0.543749988079071,
|
| 42 |
+
"rewards/chosen": -15.381937980651855,
|
| 43 |
+
"rewards/margins": 2.3547475337982178,
|
| 44 |
+
"rewards/rejected": -17.73668670654297,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 811.398681640625,
|
| 50 |
+
"learning_rate": 2.34375e-05,
|
| 51 |
+
"log_odds_chosen": 4.317142486572266,
|
| 52 |
+
"log_odds_ratio": -7.490413665771484,
|
| 53 |
+
"logits/chosen": 124.2877197265625,
|
| 54 |
+
"logits/rejected": 150.99069213867188,
|
| 55 |
+
"logps/chosen": -19.422260284423828,
|
| 56 |
+
"logps/rejected": -23.737707138061523,
|
| 57 |
+
"loss": 463.5671,
|
| 58 |
+
"nll_loss": 8.224129676818848,
|
| 59 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 60 |
+
"rewards/chosen": -19.422260284423828,
|
| 61 |
+
"rewards/margins": 4.315448760986328,
|
| 62 |
+
"rewards/rejected": -23.737707138061523,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 2864.72998046875,
|
| 68 |
+
"learning_rate": 3.125e-05,
|
| 69 |
+
"log_odds_chosen": -0.37825077772140503,
|
| 70 |
+
"log_odds_ratio": -5.3752336502075195,
|
| 71 |
+
"logits/chosen": 149.6305694580078,
|
| 72 |
+
"logits/rejected": 154.65283203125,
|
| 73 |
+
"logps/chosen": -12.753531455993652,
|
| 74 |
+
"logps/rejected": -12.376798629760742,
|
| 75 |
+
"loss": 382.9157,
|
| 76 |
+
"nll_loss": 6.1617231369018555,
|
| 77 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 78 |
+
"rewards/chosen": -12.753531455993652,
|
| 79 |
+
"rewards/margins": -0.37673258781433105,
|
| 80 |
+
"rewards/rejected": -12.376798629760742,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 270.8457336425781,
|
| 86 |
+
"learning_rate": 3.90625e-05,
|
| 87 |
+
"log_odds_chosen": 0.9792464375495911,
|
| 88 |
+
"log_odds_ratio": -1.264578104019165,
|
| 89 |
+
"logits/chosen": 165.81503295898438,
|
| 90 |
+
"logits/rejected": 173.917236328125,
|
| 91 |
+
"logps/chosen": -3.277179002761841,
|
| 92 |
+
"logps/rejected": -4.237425327301025,
|
| 93 |
+
"loss": 118.0276,
|
| 94 |
+
"nll_loss": 2.5571084022521973,
|
| 95 |
+
"rewards/accuracies": 0.581250011920929,
|
| 96 |
+
"rewards/chosen": -3.277179002761841,
|
| 97 |
+
"rewards/margins": 0.9602456092834473,
|
| 98 |
+
"rewards/rejected": -4.237425327301025,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 418.2054138183594,
|
| 104 |
+
"learning_rate": 4.6875e-05,
|
| 105 |
+
"log_odds_chosen": 0.16537247598171234,
|
| 106 |
+
"log_odds_ratio": -0.7795592546463013,
|
| 107 |
+
"logits/chosen": 191.50888061523438,
|
| 108 |
+
"logits/rejected": 213.9929962158203,
|
| 109 |
+
"logps/chosen": -1.7166259288787842,
|
| 110 |
+
"logps/rejected": -1.8560094833374023,
|
| 111 |
+
"loss": 91.4676,
|
| 112 |
+
"nll_loss": 2.072207450866699,
|
| 113 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 114 |
+
"rewards/chosen": -1.7166259288787842,
|
| 115 |
+
"rewards/margins": 0.13938350975513458,
|
| 116 |
+
"rewards/rejected": -1.8560094833374023,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 174.2537841796875,
|
| 122 |
+
"learning_rate": 4.998613757348784e-05,
|
| 123 |
+
"log_odds_chosen": 0.051605671644210815,
|
| 124 |
+
"log_odds_ratio": -0.8088222742080688,
|
| 125 |
+
"logits/chosen": 232.1632080078125,
|
| 126 |
+
"logits/rejected": 228.53805541992188,
|
| 127 |
+
"logps/chosen": -1.6349283456802368,
|
| 128 |
+
"logps/rejected": -1.6776412725448608,
|
| 129 |
+
"loss": 83.3412,
|
| 130 |
+
"nll_loss": 1.8695510625839233,
|
| 131 |
+
"rewards/accuracies": 0.543749988079071,
|
| 132 |
+
"rewards/chosen": -1.6349283456802368,
|
| 133 |
+
"rewards/margins": 0.04271280765533447,
|
| 134 |
+
"rewards/rejected": -1.6776412725448608,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 226.8765869140625,
|
| 140 |
+
"learning_rate": 4.990147841143462e-05,
|
| 141 |
+
"log_odds_chosen": 0.25601527094841003,
|
| 142 |
+
"log_odds_ratio": -0.6702470779418945,
|
| 143 |
+
"logits/chosen": 223.4049072265625,
|
| 144 |
+
"logits/rejected": 223.411865234375,
|
| 145 |
+
"logps/chosen": -1.4389511346817017,
|
| 146 |
+
"logps/rejected": -1.6484975814819336,
|
| 147 |
+
"loss": 78.3711,
|
| 148 |
+
"nll_loss": 1.7379310131072998,
|
| 149 |
+
"rewards/accuracies": 0.518750011920929,
|
| 150 |
+
"rewards/chosen": -1.4389511346817017,
|
| 151 |
+
"rewards/margins": 0.20954635739326477,
|
| 152 |
+
"rewards/rejected": -1.6484975814819336,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 218.253173828125,
|
| 158 |
+
"learning_rate": 4.97401218720448e-05,
|
| 159 |
+
"log_odds_chosen": 0.19037587940692902,
|
| 160 |
+
"log_odds_ratio": -0.7434884905815125,
|
| 161 |
+
"logits/chosen": 205.4334716796875,
|
| 162 |
+
"logits/rejected": 204.65597534179688,
|
| 163 |
+
"logps/chosen": -1.4705419540405273,
|
| 164 |
+
"logps/rejected": -1.6275854110717773,
|
| 165 |
+
"loss": 77.5272,
|
| 166 |
+
"nll_loss": 1.7267898321151733,
|
| 167 |
+
"rewards/accuracies": 0.53125,
|
| 168 |
+
"rewards/chosen": -1.4705419540405273,
|
| 169 |
+
"rewards/margins": 0.1570436656475067,
|
| 170 |
+
"rewards/rejected": -1.6275854110717773,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 198.9919891357422,
|
| 176 |
+
"learning_rate": 4.9502564938797946e-05,
|
| 177 |
+
"log_odds_chosen": 0.234850212931633,
|
| 178 |
+
"log_odds_ratio": -0.6888414621353149,
|
| 179 |
+
"logits/chosen": 188.22451782226562,
|
| 180 |
+
"logits/rejected": 189.4857635498047,
|
| 181 |
+
"logps/chosen": -1.421937346458435,
|
| 182 |
+
"logps/rejected": -1.612047553062439,
|
| 183 |
+
"loss": 77.551,
|
| 184 |
+
"nll_loss": 1.770594596862793,
|
| 185 |
+
"rewards/accuracies": 0.612500011920929,
|
| 186 |
+
"rewards/chosen": -1.421937346458435,
|
| 187 |
+
"rewards/margins": 0.19011005759239197,
|
| 188 |
+
"rewards/rejected": -1.612047553062439,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 135.9225311279297,
|
| 194 |
+
"learning_rate": 4.918953929490768e-05,
|
| 195 |
+
"log_odds_chosen": 0.16915282607078552,
|
| 196 |
+
"log_odds_ratio": -0.6923421621322632,
|
| 197 |
+
"logits/chosen": 176.1234130859375,
|
| 198 |
+
"logits/rejected": 175.10275268554688,
|
| 199 |
+
"logps/chosen": -1.357633352279663,
|
| 200 |
+
"logps/rejected": -1.491914987564087,
|
| 201 |
+
"loss": 73.8272,
|
| 202 |
+
"nll_loss": 1.6543657779693604,
|
| 203 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 204 |
+
"rewards/chosen": -1.357633352279663,
|
| 205 |
+
"rewards/margins": 0.134281724691391,
|
| 206 |
+
"rewards/rejected": -1.491914987564087,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 131.33181762695312,
|
| 212 |
+
"learning_rate": 4.88020090697132e-05,
|
| 213 |
+
"log_odds_chosen": 0.31041574478149414,
|
| 214 |
+
"log_odds_ratio": -0.6263538599014282,
|
| 215 |
+
"logits/chosen": 176.39370727539062,
|
| 216 |
+
"logits/rejected": 170.0980682373047,
|
| 217 |
+
"logps/chosen": -1.2713478803634644,
|
| 218 |
+
"logps/rejected": -1.5111758708953857,
|
| 219 |
+
"loss": 70.1645,
|
| 220 |
+
"nll_loss": 1.550362229347229,
|
| 221 |
+
"rewards/accuracies": 0.65625,
|
| 222 |
+
"rewards/chosen": -1.2713478803634644,
|
| 223 |
+
"rewards/margins": 0.23982813954353333,
|
| 224 |
+
"rewards/rejected": -1.5111758708953857,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 104.18936157226562,
|
| 230 |
+
"learning_rate": 4.834116786912897e-05,
|
| 231 |
+
"log_odds_chosen": 0.23683097958564758,
|
| 232 |
+
"log_odds_ratio": -0.6781455278396606,
|
| 233 |
+
"logits/chosen": 192.8118438720703,
|
| 234 |
+
"logits/rejected": 195.36318969726562,
|
| 235 |
+
"logps/chosen": -1.346442699432373,
|
| 236 |
+
"logps/rejected": -1.5250723361968994,
|
| 237 |
+
"loss": 71.1398,
|
| 238 |
+
"nll_loss": 1.585824966430664,
|
| 239 |
+
"rewards/accuracies": 0.574999988079071,
|
| 240 |
+
"rewards/chosen": -1.346442699432373,
|
| 241 |
+
"rewards/margins": 0.17862969636917114,
|
| 242 |
+
"rewards/rejected": -1.5250723361968994,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 123.73108673095703,
|
| 248 |
+
"learning_rate": 4.7808435099299045e-05,
|
| 249 |
+
"log_odds_chosen": 0.3379821181297302,
|
| 250 |
+
"log_odds_ratio": -0.6382076740264893,
|
| 251 |
+
"logits/chosen": 199.63497924804688,
|
| 252 |
+
"logits/rejected": 185.8668212890625,
|
| 253 |
+
"logps/chosen": -1.2565580606460571,
|
| 254 |
+
"logps/rejected": -1.5383936166763306,
|
| 255 |
+
"loss": 70.0385,
|
| 256 |
+
"nll_loss": 1.5619443655014038,
|
| 257 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 258 |
+
"rewards/chosen": -1.2565580606460571,
|
| 259 |
+
"rewards/margins": 0.2818354666233063,
|
| 260 |
+
"rewards/rejected": -1.5383936166763306,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 275.268310546875,
|
| 266 |
+
"learning_rate": 4.720545159477922e-05,
|
| 267 |
+
"log_odds_chosen": 0.3406980633735657,
|
| 268 |
+
"log_odds_ratio": -0.6167820692062378,
|
| 269 |
+
"logits/chosen": 188.97837829589844,
|
| 270 |
+
"logits/rejected": 196.31442260742188,
|
| 271 |
+
"logps/chosen": -1.2103688716888428,
|
| 272 |
+
"logps/rejected": -1.4666224718093872,
|
| 273 |
+
"loss": 68.3913,
|
| 274 |
+
"nll_loss": 1.5270273685455322,
|
| 275 |
+
"rewards/accuracies": 0.65625,
|
| 276 |
+
"rewards/chosen": -1.2103688716888428,
|
| 277 |
+
"rewards/margins": 0.2562536597251892,
|
| 278 |
+
"rewards/rejected": -1.4666224718093872,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 192.03659057617188,
|
| 284 |
+
"learning_rate": 4.653407456471222e-05,
|
| 285 |
+
"log_odds_chosen": 0.3067890405654907,
|
| 286 |
+
"log_odds_ratio": -0.6568073034286499,
|
| 287 |
+
"logits/chosen": 198.9715118408203,
|
| 288 |
+
"logits/rejected": 192.9124298095703,
|
| 289 |
+
"logps/chosen": -1.2415003776550293,
|
| 290 |
+
"logps/rejected": -1.4735556840896606,
|
| 291 |
+
"loss": 68.6961,
|
| 292 |
+
"nll_loss": 1.5345680713653564,
|
| 293 |
+
"rewards/accuracies": 0.643750011920929,
|
| 294 |
+
"rewards/chosen": -1.2415003776550293,
|
| 295 |
+
"rewards/margins": 0.23205523192882538,
|
| 296 |
+
"rewards/rejected": -1.4735556840896606,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 109.5743637084961,
|
| 302 |
+
"learning_rate": 4.579637187256222e-05,
|
| 303 |
+
"log_odds_chosen": 0.27371346950531006,
|
| 304 |
+
"log_odds_ratio": -0.6534064412117004,
|
| 305 |
+
"logits/chosen": 186.74472045898438,
|
| 306 |
+
"logits/rejected": 174.53298950195312,
|
| 307 |
+
"logps/chosen": -1.1753740310668945,
|
| 308 |
+
"logps/rejected": -1.384543776512146,
|
| 309 |
+
"loss": 66.8724,
|
| 310 |
+
"nll_loss": 1.45794677734375,
|
| 311 |
+
"rewards/accuracies": 0.625,
|
| 312 |
+
"rewards/chosen": -1.1753740310668945,
|
| 313 |
+
"rewards/margins": 0.20916974544525146,
|
| 314 |
+
"rewards/rejected": -1.384543776512146,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 147.4329833984375,
|
| 320 |
+
"learning_rate": 4.499461566702685e-05,
|
| 321 |
+
"log_odds_chosen": 0.33060330152511597,
|
| 322 |
+
"log_odds_ratio": -0.629867434501648,
|
| 323 |
+
"logits/chosen": 188.45364379882812,
|
| 324 |
+
"logits/rejected": 184.3141632080078,
|
| 325 |
+
"logps/chosen": -1.1476401090621948,
|
| 326 |
+
"logps/rejected": -1.3880122900009155,
|
| 327 |
+
"loss": 70.2813,
|
| 328 |
+
"nll_loss": 1.5526297092437744,
|
| 329 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 330 |
+
"rewards/chosen": -1.1476401090621948,
|
| 331 |
+
"rewards/margins": 0.2403722107410431,
|
| 332 |
+
"rewards/rejected": -1.3880122900009155,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 114.0191421508789,
|
| 338 |
+
"learning_rate": 4.413127538374411e-05,
|
| 339 |
+
"log_odds_chosen": 0.4007638990879059,
|
| 340 |
+
"log_odds_ratio": -0.5877656936645508,
|
| 341 |
+
"logits/chosen": 182.2388916015625,
|
| 342 |
+
"logits/rejected": 174.63726806640625,
|
| 343 |
+
"logps/chosen": -1.0714807510375977,
|
| 344 |
+
"logps/rejected": -1.3553247451782227,
|
| 345 |
+
"loss": 67.3621,
|
| 346 |
+
"nll_loss": 1.3920471668243408,
|
| 347 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 348 |
+
"rewards/chosen": -1.0714807510375977,
|
| 349 |
+
"rewards/margins": 0.2838440239429474,
|
| 350 |
+
"rewards/rejected": -1.3553247451782227,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 159.00311279296875,
|
| 356 |
+
"learning_rate": 4.320901013934887e-05,
|
| 357 |
+
"log_odds_chosen": 0.2362973690032959,
|
| 358 |
+
"log_odds_ratio": -0.6814771890640259,
|
| 359 |
+
"logits/chosen": 189.81747436523438,
|
| 360 |
+
"logits/rejected": 179.8380584716797,
|
| 361 |
+
"logps/chosen": -1.2289823293685913,
|
| 362 |
+
"logps/rejected": -1.4264824390411377,
|
| 363 |
+
"loss": 68.8832,
|
| 364 |
+
"nll_loss": 1.524703860282898,
|
| 365 |
+
"rewards/accuracies": 0.543749988079071,
|
| 366 |
+
"rewards/chosen": -1.2289823293685913,
|
| 367 |
+
"rewards/margins": 0.197500079870224,
|
| 368 |
+
"rewards/rejected": -1.4264824390411377,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 95.12399291992188,
|
| 374 |
+
"learning_rate": 4.223066054130568e-05,
|
| 375 |
+
"log_odds_chosen": 0.32050785422325134,
|
| 376 |
+
"log_odds_ratio": -0.6115350723266602,
|
| 377 |
+
"logits/chosen": 187.16676330566406,
|
| 378 |
+
"logits/rejected": 178.86520385742188,
|
| 379 |
+
"logps/chosen": -1.1217820644378662,
|
| 380 |
+
"logps/rejected": -1.3648459911346436,
|
| 381 |
+
"loss": 66.0352,
|
| 382 |
+
"nll_loss": 1.4325681924819946,
|
| 383 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 384 |
+
"rewards/chosen": -1.1217820644378662,
|
| 385 |
+
"rewards/margins": 0.24306419491767883,
|
| 386 |
+
"rewards/rejected": -1.3648459911346436,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.037925925925926,
|
| 391 |
+
"grad_norm": 105.41276550292969,
|
| 392 |
+
"learning_rate": 4.1199239938743797e-05,
|
| 393 |
+
"log_odds_chosen": 0.8203701376914978,
|
| 394 |
+
"log_odds_ratio": -0.4882337749004364,
|
| 395 |
+
"logits/chosen": 178.85992431640625,
|
| 396 |
+
"logits/rejected": 173.57992553710938,
|
| 397 |
+
"logps/chosen": -0.968072235584259,
|
| 398 |
+
"logps/rejected": -1.573646068572998,
|
| 399 |
+
"loss": 50.8327,
|
| 400 |
+
"nll_loss": 1.271679162979126,
|
| 401 |
+
"rewards/accuracies": 0.7902097702026367,
|
| 402 |
+
"rewards/chosen": -0.968072235584259,
|
| 403 |
+
"rewards/margins": 0.6055738925933838,
|
| 404 |
+
"rewards/rejected": -1.573646068572998,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0853333333333333,
|
| 409 |
+
"grad_norm": 100.29273223876953,
|
| 410 |
+
"learning_rate": 4.0117925141242174e-05,
|
| 411 |
+
"log_odds_chosen": 1.2380282878875732,
|
| 412 |
+
"log_odds_ratio": -0.35476842522621155,
|
| 413 |
+
"logits/chosen": 169.09445190429688,
|
| 414 |
+
"logits/rejected": 162.7259063720703,
|
| 415 |
+
"logps/chosen": -0.9231376647949219,
|
| 416 |
+
"logps/rejected": -1.8257957696914673,
|
| 417 |
+
"loss": 52.7199,
|
| 418 |
+
"nll_loss": 1.2373106479644775,
|
| 419 |
+
"rewards/accuracies": 0.875,
|
| 420 |
+
"rewards/chosen": -0.9231376647949219,
|
| 421 |
+
"rewards/margins": 0.9026579856872559,
|
| 422 |
+
"rewards/rejected": -1.8257957696914673,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1327407407407408,
|
| 427 |
+
"grad_norm": 80.36443328857422,
|
| 428 |
+
"learning_rate": 3.899004663415084e-05,
|
| 429 |
+
"log_odds_chosen": 1.1377732753753662,
|
| 430 |
+
"log_odds_ratio": -0.3693670630455017,
|
| 431 |
+
"logits/chosen": 165.61392211914062,
|
| 432 |
+
"logits/rejected": 165.60780334472656,
|
| 433 |
+
"logps/chosen": -0.9129512906074524,
|
| 434 |
+
"logps/rejected": -1.708299994468689,
|
| 435 |
+
"loss": 53.4303,
|
| 436 |
+
"nll_loss": 1.2692068815231323,
|
| 437 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 438 |
+
"rewards/chosen": -0.9129512906074524,
|
| 439 |
+
"rewards/margins": 0.7953487634658813,
|
| 440 |
+
"rewards/rejected": -1.708299994468689,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1801481481481482,
|
| 445 |
+
"grad_norm": 85.88944244384766,
|
| 446 |
+
"learning_rate": 3.781907832058587e-05,
|
| 447 |
+
"log_odds_chosen": 1.2076492309570312,
|
| 448 |
+
"log_odds_ratio": -0.3797214925289154,
|
| 449 |
+
"logits/chosen": 160.6963348388672,
|
| 450 |
+
"logits/rejected": 159.27244567871094,
|
| 451 |
+
"logps/chosen": -0.9866069555282593,
|
| 452 |
+
"logps/rejected": -1.8863252401351929,
|
| 453 |
+
"loss": 50.8548,
|
| 454 |
+
"nll_loss": 1.2557852268218994,
|
| 455 |
+
"rewards/accuracies": 0.862500011920929,
|
| 456 |
+
"rewards/chosen": -0.9866069555282593,
|
| 457 |
+
"rewards/margins": 0.8997182846069336,
|
| 458 |
+
"rewards/rejected": -1.8863252401351929,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2275555555555555,
|
| 463 |
+
"grad_norm": 103.79596710205078,
|
| 464 |
+
"learning_rate": 3.660862682169282e-05,
|
| 465 |
+
"log_odds_chosen": 1.0689417123794556,
|
| 466 |
+
"log_odds_ratio": -0.39478644728660583,
|
| 467 |
+
"logits/chosen": 164.0693359375,
|
| 468 |
+
"logits/rejected": 163.32144165039062,
|
| 469 |
+
"logps/chosen": -0.9268579483032227,
|
| 470 |
+
"logps/rejected": -1.66475510597229,
|
| 471 |
+
"loss": 51.5853,
|
| 472 |
+
"nll_loss": 1.2472236156463623,
|
| 473 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 474 |
+
"rewards/chosen": -0.9268579483032227,
|
| 475 |
+
"rewards/margins": 0.7378970384597778,
|
| 476 |
+
"rewards/rejected": -1.66475510597229,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.274962962962963,
|
| 481 |
+
"grad_norm": 101.42869567871094,
|
| 482 |
+
"learning_rate": 3.5362420368134356e-05,
|
| 483 |
+
"log_odds_chosen": 1.0775753259658813,
|
| 484 |
+
"log_odds_ratio": -0.3807566165924072,
|
| 485 |
+
"logits/chosen": 165.19088745117188,
|
| 486 |
+
"logits/rejected": 165.37791442871094,
|
| 487 |
+
"logps/chosen": -0.8853290677070618,
|
| 488 |
+
"logps/rejected": -1.6260745525360107,
|
| 489 |
+
"loss": 50.9928,
|
| 490 |
+
"nll_loss": 1.2169647216796875,
|
| 491 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 492 |
+
"rewards/chosen": -0.8853290677070618,
|
| 493 |
+
"rewards/margins": 0.7407453656196594,
|
| 494 |
+
"rewards/rejected": -1.6260745525360107,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3223703703703704,
|
| 499 |
+
"grad_norm": 97.27503967285156,
|
| 500 |
+
"learning_rate": 3.408429731701635e-05,
|
| 501 |
+
"log_odds_chosen": 1.1813185214996338,
|
| 502 |
+
"log_odds_ratio": -0.37981921434402466,
|
| 503 |
+
"logits/chosen": 161.71279907226562,
|
| 504 |
+
"logits/rejected": 166.4205780029297,
|
| 505 |
+
"logps/chosen": -0.9885748028755188,
|
| 506 |
+
"logps/rejected": -1.8441470861434937,
|
| 507 |
+
"loss": 53.5563,
|
| 508 |
+
"nll_loss": 1.2953031063079834,
|
| 509 |
+
"rewards/accuracies": 0.84375,
|
| 510 |
+
"rewards/chosen": -0.9885748028755188,
|
| 511 |
+
"rewards/margins": 0.855572521686554,
|
| 512 |
+
"rewards/rejected": -1.8441470861434937,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.3697777777777778,
|
| 517 |
+
"grad_norm": 92.24687194824219,
|
| 518 |
+
"learning_rate": 3.2778194329621104e-05,
|
| 519 |
+
"log_odds_chosen": 1.2082122564315796,
|
| 520 |
+
"log_odds_ratio": -0.3629351854324341,
|
| 521 |
+
"logits/chosen": 165.2741241455078,
|
| 522 |
+
"logits/rejected": 156.53671264648438,
|
| 523 |
+
"logps/chosen": -0.967271625995636,
|
| 524 |
+
"logps/rejected": -1.8560912609100342,
|
| 525 |
+
"loss": 52.2453,
|
| 526 |
+
"nll_loss": 1.2710577249526978,
|
| 527 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 528 |
+
"rewards/chosen": -0.967271625995636,
|
| 529 |
+
"rewards/margins": 0.888819694519043,
|
| 530 |
+
"rewards/rejected": -1.8560912609100342,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.417185185185185,
|
| 535 |
+
"grad_norm": 97.35833740234375,
|
| 536 |
+
"learning_rate": 3.144813424636031e-05,
|
| 537 |
+
"log_odds_chosen": 1.1898943185806274,
|
| 538 |
+
"log_odds_ratio": -0.357857882976532,
|
| 539 |
+
"logits/chosen": 153.11080932617188,
|
| 540 |
+
"logits/rejected": 157.1022491455078,
|
| 541 |
+
"logps/chosen": -0.9347141981124878,
|
| 542 |
+
"logps/rejected": -1.767988920211792,
|
| 543 |
+
"loss": 50.7791,
|
| 544 |
+
"nll_loss": 1.218804121017456,
|
| 545 |
+
"rewards/accuracies": 0.893750011920929,
|
| 546 |
+
"rewards/chosen": -0.9347141981124878,
|
| 547 |
+
"rewards/margins": 0.8332747220993042,
|
| 548 |
+
"rewards/rejected": -1.767988920211792,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4645925925925927,
|
| 553 |
+
"grad_norm": 105.37850952148438,
|
| 554 |
+
"learning_rate": 3.0098213696293542e-05,
|
| 555 |
+
"log_odds_chosen": 1.1642911434173584,
|
| 556 |
+
"log_odds_ratio": -0.3831285238265991,
|
| 557 |
+
"logits/chosen": 160.2550048828125,
|
| 558 |
+
"logits/rejected": 152.56373596191406,
|
| 559 |
+
"logps/chosen": -0.904646098613739,
|
| 560 |
+
"logps/rejected": -1.7168747186660767,
|
| 561 |
+
"loss": 52.4723,
|
| 562 |
+
"nll_loss": 1.257243037223816,
|
| 563 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 564 |
+
"rewards/chosen": -0.904646098613739,
|
| 565 |
+
"rewards/margins": 0.8122288584709167,
|
| 566 |
+
"rewards/rejected": -1.7168747186660767,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.512,
|
| 571 |
+
"grad_norm": 107.52549743652344,
|
| 572 |
+
"learning_rate": 2.8732590479375165e-05,
|
| 573 |
+
"log_odds_chosen": 1.1039049625396729,
|
| 574 |
+
"log_odds_ratio": -0.38589566946029663,
|
| 575 |
+
"logits/chosen": 156.07980346679688,
|
| 576 |
+
"logits/rejected": 157.7055206298828,
|
| 577 |
+
"logps/chosen": -1.0009373426437378,
|
| 578 |
+
"logps/rejected": -1.8057382106781006,
|
| 579 |
+
"loss": 51.9453,
|
| 580 |
+
"nll_loss": 1.290351152420044,
|
| 581 |
+
"rewards/accuracies": 0.856249988079071,
|
| 582 |
+
"rewards/chosen": -1.0009373426437378,
|
| 583 |
+
"rewards/margins": 0.8048008680343628,
|
| 584 |
+
"rewards/rejected": -1.8057382106781006,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5594074074074074,
|
| 589 |
+
"grad_norm": 75.2291030883789,
|
| 590 |
+
"learning_rate": 2.7355470760292956e-05,
|
| 591 |
+
"log_odds_chosen": 1.389765977859497,
|
| 592 |
+
"log_odds_ratio": -0.31658080220222473,
|
| 593 |
+
"logits/chosen": 146.20651245117188,
|
| 594 |
+
"logits/rejected": 156.01681518554688,
|
| 595 |
+
"logps/chosen": -0.939640998840332,
|
| 596 |
+
"logps/rejected": -1.9385979175567627,
|
| 597 |
+
"loss": 50.8517,
|
| 598 |
+
"nll_loss": 1.2414509057998657,
|
| 599 |
+
"rewards/accuracies": 0.90625,
|
| 600 |
+
"rewards/chosen": -0.939640998840332,
|
| 601 |
+
"rewards/margins": 0.9989569783210754,
|
| 602 |
+
"rewards/rejected": -1.9385979175567627,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6068148148148147,
|
| 607 |
+
"grad_norm": 119.51274871826172,
|
| 608 |
+
"learning_rate": 2.597109611334169e-05,
|
| 609 |
+
"log_odds_chosen": 1.3545691967010498,
|
| 610 |
+
"log_odds_ratio": -0.343548983335495,
|
| 611 |
+
"logits/chosen": 160.31771850585938,
|
| 612 |
+
"logits/rejected": 159.28652954101562,
|
| 613 |
+
"logps/chosen": -0.9329752922058105,
|
| 614 |
+
"logps/rejected": -1.9087892770767212,
|
| 615 |
+
"loss": 50.7377,
|
| 616 |
+
"nll_loss": 1.2252418994903564,
|
| 617 |
+
"rewards/accuracies": 0.887499988079071,
|
| 618 |
+
"rewards/chosen": -0.9329752922058105,
|
| 619 |
+
"rewards/margins": 0.9758139848709106,
|
| 620 |
+
"rewards/rejected": -1.9087892770767212,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6542222222222223,
|
| 625 |
+
"grad_norm": 134.74673461914062,
|
| 626 |
+
"learning_rate": 2.458373045823404e-05,
|
| 627 |
+
"log_odds_chosen": 1.3076380491256714,
|
| 628 |
+
"log_odds_ratio": -0.32780036330223083,
|
| 629 |
+
"logits/chosen": 162.3087921142578,
|
| 630 |
+
"logits/rejected": 153.71522521972656,
|
| 631 |
+
"logps/chosen": -0.9549840688705444,
|
| 632 |
+
"logps/rejected": -1.9126832485198975,
|
| 633 |
+
"loss": 50.8072,
|
| 634 |
+
"nll_loss": 1.28162682056427,
|
| 635 |
+
"rewards/accuracies": 0.918749988079071,
|
| 636 |
+
"rewards/chosen": -0.9549840688705444,
|
| 637 |
+
"rewards/margins": 0.9576990008354187,
|
| 638 |
+
"rewards/rejected": -1.9126832485198975,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7016296296296296,
|
| 643 |
+
"grad_norm": 86.47370147705078,
|
| 644 |
+
"learning_rate": 2.3197646927086697e-05,
|
| 645 |
+
"log_odds_chosen": 1.1750502586364746,
|
| 646 |
+
"log_odds_ratio": -0.39812663197517395,
|
| 647 |
+
"logits/chosen": 168.8065643310547,
|
| 648 |
+
"logits/rejected": 170.52169799804688,
|
| 649 |
+
"logps/chosen": -0.9777830243110657,
|
| 650 |
+
"logps/rejected": -1.8347434997558594,
|
| 651 |
+
"loss": 51.9848,
|
| 652 |
+
"nll_loss": 1.2955636978149414,
|
| 653 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 654 |
+
"rewards/chosen": -0.9777830243110657,
|
| 655 |
+
"rewards/margins": 0.8569604158401489,
|
| 656 |
+
"rewards/rejected": -1.8347434997558594,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.749037037037037,
|
| 661 |
+
"grad_norm": 102.95687866210938,
|
| 662 |
+
"learning_rate": 2.1817114703032176e-05,
|
| 663 |
+
"log_odds_chosen": 1.341557264328003,
|
| 664 |
+
"log_odds_ratio": -0.3403902053833008,
|
| 665 |
+
"logits/chosen": 163.19363403320312,
|
| 666 |
+
"logits/rejected": 160.15911865234375,
|
| 667 |
+
"logps/chosen": -0.99519282579422,
|
| 668 |
+
"logps/rejected": -2.007880449295044,
|
| 669 |
+
"loss": 51.291,
|
| 670 |
+
"nll_loss": 1.2735412120819092,
|
| 671 |
+
"rewards/accuracies": 0.887499988079071,
|
| 672 |
+
"rewards/chosen": -0.99519282579422,
|
| 673 |
+
"rewards/margins": 1.0126875638961792,
|
| 674 |
+
"rewards/rejected": -2.007880449295044,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.7964444444444445,
|
| 679 |
+
"grad_norm": 90.98896026611328,
|
| 680 |
+
"learning_rate": 2.0446385870993467e-05,
|
| 681 |
+
"log_odds_chosen": 1.407165288925171,
|
| 682 |
+
"log_odds_ratio": -0.3309064209461212,
|
| 683 |
+
"logits/chosen": 172.62367248535156,
|
| 684 |
+
"logits/rejected": 163.45089721679688,
|
| 685 |
+
"logps/chosen": -0.9951199293136597,
|
| 686 |
+
"logps/rejected": -2.050931930541992,
|
| 687 |
+
"loss": 53.2905,
|
| 688 |
+
"nll_loss": 1.2619965076446533,
|
| 689 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 690 |
+
"rewards/chosen": -0.9951199293136597,
|
| 691 |
+
"rewards/margins": 1.0558120012283325,
|
| 692 |
+
"rewards/rejected": -2.050931930541992,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8438518518518519,
|
| 697 |
+
"grad_norm": 112.30895233154297,
|
| 698 |
+
"learning_rate": 1.9089682321121834e-05,
|
| 699 |
+
"log_odds_chosen": 1.4201080799102783,
|
| 700 |
+
"log_odds_ratio": -0.32679349184036255,
|
| 701 |
+
"logits/chosen": 173.34461975097656,
|
| 702 |
+
"logits/rejected": 160.87173461914062,
|
| 703 |
+
"logps/chosen": -0.9603132009506226,
|
| 704 |
+
"logps/rejected": -2.0295138359069824,
|
| 705 |
+
"loss": 50.4567,
|
| 706 |
+
"nll_loss": 1.2545397281646729,
|
| 707 |
+
"rewards/accuracies": 0.887499988079071,
|
| 708 |
+
"rewards/chosen": -0.9603132009506226,
|
| 709 |
+
"rewards/margins": 1.0692007541656494,
|
| 710 |
+
"rewards/rejected": -2.0295138359069824,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8912592592592592,
|
| 715 |
+
"grad_norm": 117.05001068115234,
|
| 716 |
+
"learning_rate": 1.775118274523545e-05,
|
| 717 |
+
"log_odds_chosen": 1.2845535278320312,
|
| 718 |
+
"log_odds_ratio": -0.34441834688186646,
|
| 719 |
+
"logits/chosen": 163.27099609375,
|
| 720 |
+
"logits/rejected": 160.7901611328125,
|
| 721 |
+
"logps/chosen": -0.9582241177558899,
|
| 722 |
+
"logps/rejected": -1.888132095336914,
|
| 723 |
+
"loss": 52.6478,
|
| 724 |
+
"nll_loss": 1.2212011814117432,
|
| 725 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 726 |
+
"rewards/chosen": -0.9582241177558899,
|
| 727 |
+
"rewards/margins": 0.9299079775810242,
|
| 728 |
+
"rewards/rejected": -1.888132095336914,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9386666666666668,
|
| 733 |
+
"grad_norm": 81.6810302734375,
|
| 734 |
+
"learning_rate": 1.643500976631037e-05,
|
| 735 |
+
"log_odds_chosen": 1.142424464225769,
|
| 736 |
+
"log_odds_ratio": -0.3913617432117462,
|
| 737 |
+
"logits/chosen": 166.19735717773438,
|
| 738 |
+
"logits/rejected": 156.3197784423828,
|
| 739 |
+
"logps/chosen": -0.9541540145874023,
|
| 740 |
+
"logps/rejected": -1.7820985317230225,
|
| 741 |
+
"loss": 51.1929,
|
| 742 |
+
"nll_loss": 1.2262585163116455,
|
| 743 |
+
"rewards/accuracies": 0.856249988079071,
|
| 744 |
+
"rewards/chosen": -0.9541540145874023,
|
| 745 |
+
"rewards/margins": 0.8279445767402649,
|
| 746 |
+
"rewards/rejected": -1.7820985317230225,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.986074074074074,
|
| 751 |
+
"grad_norm": 148.79823303222656,
|
| 752 |
+
"learning_rate": 1.514521724066537e-05,
|
| 753 |
+
"log_odds_chosen": 1.1579644680023193,
|
| 754 |
+
"log_odds_ratio": -0.36968621611595154,
|
| 755 |
+
"logits/chosen": 152.0222625732422,
|
| 756 |
+
"logits/rejected": 154.07534790039062,
|
| 757 |
+
"logps/chosen": -0.9201663732528687,
|
| 758 |
+
"logps/rejected": -1.7177432775497437,
|
| 759 |
+
"loss": 49.3216,
|
| 760 |
+
"nll_loss": 1.1781848669052124,
|
| 761 |
+
"rewards/accuracies": 0.887499988079071,
|
| 762 |
+
"rewards/chosen": -0.9201663732528687,
|
| 763 |
+
"rewards/margins": 0.7975768446922302,
|
| 764 |
+
"rewards/rejected": -1.7177432775497437,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0284444444444443,
|
| 769 |
+
"grad_norm": 79.6141586303711,
|
| 770 |
+
"learning_rate": 1.3885777771950348e-05,
|
| 771 |
+
"log_odds_chosen": 2.226461887359619,
|
| 772 |
+
"log_odds_ratio": -0.2312411367893219,
|
| 773 |
+
"logits/chosen": 146.73081970214844,
|
| 774 |
+
"logits/rejected": 143.3143768310547,
|
| 775 |
+
"logps/chosen": -0.8414949774742126,
|
| 776 |
+
"logps/rejected": -2.482272148132324,
|
| 777 |
+
"loss": 38.807,
|
| 778 |
+
"nll_loss": 1.1267651319503784,
|
| 779 |
+
"rewards/accuracies": 0.9300699234008789,
|
| 780 |
+
"rewards/chosen": -0.8414949774742126,
|
| 781 |
+
"rewards/margins": 1.6407769918441772,
|
| 782 |
+
"rewards/rejected": -2.482272148132324,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.075851851851852,
|
| 787 |
+
"grad_norm": 65.1515884399414,
|
| 788 |
+
"learning_rate": 1.2660570475395683e-05,
|
| 789 |
+
"log_odds_chosen": 2.904186964035034,
|
| 790 |
+
"log_odds_ratio": -0.1275218427181244,
|
| 791 |
+
"logits/chosen": 122.88028717041016,
|
| 792 |
+
"logits/rejected": 118.4028091430664,
|
| 793 |
+
"logps/chosen": -0.7325994372367859,
|
| 794 |
+
"logps/rejected": -2.9179022312164307,
|
| 795 |
+
"loss": 37.2572,
|
| 796 |
+
"nll_loss": 1.0604547262191772,
|
| 797 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 798 |
+
"rewards/chosen": -0.7325994372367859,
|
| 799 |
+
"rewards/margins": 2.185302495956421,
|
| 800 |
+
"rewards/rejected": -2.9179022312164307,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1232592592592594,
|
| 805 |
+
"grad_norm": 72.1199951171875,
|
| 806 |
+
"learning_rate": 1.1473369030008974e-05,
|
| 807 |
+
"log_odds_chosen": 2.7867863178253174,
|
| 808 |
+
"log_odds_ratio": -0.12314349412918091,
|
| 809 |
+
"logits/chosen": 136.52151489257812,
|
| 810 |
+
"logits/rejected": 127.45872497558594,
|
| 811 |
+
"logps/chosen": -0.741514265537262,
|
| 812 |
+
"logps/rejected": -2.819490909576416,
|
| 813 |
+
"loss": 38.5583,
|
| 814 |
+
"nll_loss": 1.0891916751861572,
|
| 815 |
+
"rewards/accuracies": 0.981249988079071,
|
| 816 |
+
"rewards/chosen": -0.741514265537262,
|
| 817 |
+
"rewards/margins": 2.077976703643799,
|
| 818 |
+
"rewards/rejected": -2.819490909576416,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.1706666666666665,
|
| 823 |
+
"grad_norm": 80.40210723876953,
|
| 824 |
+
"learning_rate": 1.0327830055518842e-05,
|
| 825 |
+
"log_odds_chosen": 3.051785469055176,
|
| 826 |
+
"log_odds_ratio": -0.100181445479393,
|
| 827 |
+
"logits/chosen": 138.90798950195312,
|
| 828 |
+
"logits/rejected": 127.01904296875,
|
| 829 |
+
"logps/chosen": -0.7537847757339478,
|
| 830 |
+
"logps/rejected": -3.103349447250366,
|
| 831 |
+
"loss": 36.7512,
|
| 832 |
+
"nll_loss": 1.0241692066192627,
|
| 833 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 834 |
+
"rewards/chosen": -0.7537847757339478,
|
| 835 |
+
"rewards/margins": 2.349565029144287,
|
| 836 |
+
"rewards/rejected": -3.103349447250366,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.218074074074074,
|
| 841 |
+
"grad_norm": 64.71866607666016,
|
| 842 |
+
"learning_rate": 9.227481849865235e-06,
|
| 843 |
+
"log_odds_chosen": 2.7597384452819824,
|
| 844 |
+
"log_odds_ratio": -0.12918250262737274,
|
| 845 |
+
"logits/chosen": 139.7825469970703,
|
| 846 |
+
"logits/rejected": 128.8076629638672,
|
| 847 |
+
"logps/chosen": -0.74356609582901,
|
| 848 |
+
"logps/rejected": -2.7902283668518066,
|
| 849 |
+
"loss": 37.6805,
|
| 850 |
+
"nll_loss": 1.052249789237976,
|
| 851 |
+
"rewards/accuracies": 1.0,
|
| 852 |
+
"rewards/chosen": -0.74356609582901,
|
| 853 |
+
"rewards/margins": 2.0466625690460205,
|
| 854 |
+
"rewards/rejected": -2.7902283668518066,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2654814814814817,
|
| 859 |
+
"grad_norm": 73.16040802001953,
|
| 860 |
+
"learning_rate": 8.175713521924978e-06,
|
| 861 |
+
"log_odds_chosen": 2.906674861907959,
|
| 862 |
+
"log_odds_ratio": -0.1187477856874466,
|
| 863 |
+
"logits/chosen": 141.90513610839844,
|
| 864 |
+
"logits/rejected": 132.15574645996094,
|
| 865 |
+
"logps/chosen": -0.7376239895820618,
|
| 866 |
+
"logps/rejected": -2.9203648567199707,
|
| 867 |
+
"loss": 37.9035,
|
| 868 |
+
"nll_loss": 1.0850651264190674,
|
| 869 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 870 |
+
"rewards/chosen": -0.7376239895820618,
|
| 871 |
+
"rewards/margins": 2.1827406883239746,
|
| 872 |
+
"rewards/rejected": -2.9203648567199707,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.3128888888888888,
|
| 877 |
+
"grad_norm": 90.49305725097656,
|
| 878 |
+
"learning_rate": 7.1757645529443665e-06,
|
| 879 |
+
"log_odds_chosen": 3.180687427520752,
|
| 880 |
+
"log_odds_ratio": -0.09146104753017426,
|
| 881 |
+
"logits/chosen": 137.44186401367188,
|
| 882 |
+
"logits/rejected": 123.99406433105469,
|
| 883 |
+
"logps/chosen": -0.7279849052429199,
|
| 884 |
+
"logps/rejected": -3.173057794570923,
|
| 885 |
+
"loss": 35.677,
|
| 886 |
+
"nll_loss": 1.0552171468734741,
|
| 887 |
+
"rewards/accuracies": 1.0,
|
| 888 |
+
"rewards/chosen": -0.7279849052429199,
|
| 889 |
+
"rewards/margins": 2.445072650909424,
|
| 890 |
+
"rewards/rejected": -3.173057794570923,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3602962962962963,
|
| 895 |
+
"grad_norm": 64.9398422241211,
|
| 896 |
+
"learning_rate": 6.230714818829733e-06,
|
| 897 |
+
"log_odds_chosen": 3.0522115230560303,
|
| 898 |
+
"log_odds_ratio": -0.10979001224040985,
|
| 899 |
+
"logits/chosen": 139.00146484375,
|
| 900 |
+
"logits/rejected": 133.86178588867188,
|
| 901 |
+
"logps/chosen": -0.7283380031585693,
|
| 902 |
+
"logps/rejected": -3.035508394241333,
|
| 903 |
+
"loss": 36.5909,
|
| 904 |
+
"nll_loss": 1.041661024093628,
|
| 905 |
+
"rewards/accuracies": 0.987500011920929,
|
| 906 |
+
"rewards/chosen": -0.7283380031585693,
|
| 907 |
+
"rewards/margins": 2.3071703910827637,
|
| 908 |
+
"rewards/rejected": -3.035508394241333,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.407703703703704,
|
| 913 |
+
"grad_norm": 65.51322937011719,
|
| 914 |
+
"learning_rate": 5.343475104027743e-06,
|
| 915 |
+
"log_odds_chosen": 3.1422739028930664,
|
| 916 |
+
"log_odds_ratio": -0.10407264530658722,
|
| 917 |
+
"logits/chosen": 137.86683654785156,
|
| 918 |
+
"logits/rejected": 134.326416015625,
|
| 919 |
+
"logps/chosen": -0.6767088174819946,
|
| 920 |
+
"logps/rejected": -3.0573792457580566,
|
| 921 |
+
"loss": 36.0937,
|
| 922 |
+
"nll_loss": 1.0149527788162231,
|
| 923 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 924 |
+
"rewards/chosen": -0.6767088174819946,
|
| 925 |
+
"rewards/margins": 2.3806705474853516,
|
| 926 |
+
"rewards/rejected": -3.0573792457580566,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.455111111111111,
|
| 931 |
+
"grad_norm": 69.81231689453125,
|
| 932 |
+
"learning_rate": 4.516778136213037e-06,
|
| 933 |
+
"log_odds_chosen": 3.3670654296875,
|
| 934 |
+
"log_odds_ratio": -0.08844997733831406,
|
| 935 |
+
"logits/chosen": 141.07667541503906,
|
| 936 |
+
"logits/rejected": 131.75830078125,
|
| 937 |
+
"logps/chosen": -0.6812475323677063,
|
| 938 |
+
"logps/rejected": -3.250110149383545,
|
| 939 |
+
"loss": 35.9589,
|
| 940 |
+
"nll_loss": 1.0102312564849854,
|
| 941 |
+
"rewards/accuracies": 1.0,
|
| 942 |
+
"rewards/chosen": -0.6812475323677063,
|
| 943 |
+
"rewards/margins": 2.5688624382019043,
|
| 944 |
+
"rewards/rejected": -3.250110149383545,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5025185185185186,
|
| 949 |
+
"grad_norm": 85.25850677490234,
|
| 950 |
+
"learning_rate": 3.7531701693965554e-06,
|
| 951 |
+
"log_odds_chosen": 2.994309663772583,
|
| 952 |
+
"log_odds_ratio": -0.13702499866485596,
|
| 953 |
+
"logits/chosen": 143.72352600097656,
|
| 954 |
+
"logits/rejected": 137.90383911132812,
|
| 955 |
+
"logps/chosen": -0.7097981572151184,
|
| 956 |
+
"logps/rejected": -2.959254264831543,
|
| 957 |
+
"loss": 35.9198,
|
| 958 |
+
"nll_loss": 0.9953324198722839,
|
| 959 |
+
"rewards/accuracies": 0.981249988079071,
|
| 960 |
+
"rewards/chosen": -0.7097981572151184,
|
| 961 |
+
"rewards/margins": 2.2494559288024902,
|
| 962 |
+
"rewards/rejected": -2.959254264831543,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.549925925925926,
|
| 967 |
+
"grad_norm": 67.82206726074219,
|
| 968 |
+
"learning_rate": 3.055003141378948e-06,
|
| 969 |
+
"log_odds_chosen": 3.0670759677886963,
|
| 970 |
+
"log_odds_ratio": -0.11550645530223846,
|
| 971 |
+
"logits/chosen": 142.99095153808594,
|
| 972 |
+
"logits/rejected": 136.27975463867188,
|
| 973 |
+
"logps/chosen": -0.763370156288147,
|
| 974 |
+
"logps/rejected": -3.0597498416900635,
|
| 975 |
+
"loss": 35.5905,
|
| 976 |
+
"nll_loss": 1.0493717193603516,
|
| 977 |
+
"rewards/accuracies": 0.987500011920929,
|
| 978 |
+
"rewards/chosen": -0.763370156288147,
|
| 979 |
+
"rewards/margins": 2.296379566192627,
|
| 980 |
+
"rewards/rejected": -3.0597498416900635,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.5973333333333333,
|
| 985 |
+
"grad_norm": 60.3643913269043,
|
| 986 |
+
"learning_rate": 2.424427429704365e-06,
|
| 987 |
+
"log_odds_chosen": 3.3947010040283203,
|
| 988 |
+
"log_odds_ratio": -0.10236648470163345,
|
| 989 |
+
"logits/chosen": 141.64218139648438,
|
| 990 |
+
"logits/rejected": 128.118408203125,
|
| 991 |
+
"logps/chosen": -0.7243070006370544,
|
| 992 |
+
"logps/rejected": -3.3256897926330566,
|
| 993 |
+
"loss": 36.7892,
|
| 994 |
+
"nll_loss": 1.0436440706253052,
|
| 995 |
+
"rewards/accuracies": 0.987500011920929,
|
| 996 |
+
"rewards/chosen": -0.7243070006370544,
|
| 997 |
+
"rewards/margins": 2.6013827323913574,
|
| 998 |
+
"rewards/rejected": -3.3256897926330566,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.644740740740741,
|
| 1003 |
+
"grad_norm": 68.43870544433594,
|
| 1004 |
+
"learning_rate": 1.8633852284264508e-06,
|
| 1005 |
+
"log_odds_chosen": 3.1925907135009766,
|
| 1006 |
+
"log_odds_ratio": -0.10052428394556046,
|
| 1007 |
+
"logits/chosen": 137.6027374267578,
|
| 1008 |
+
"logits/rejected": 123.51395416259766,
|
| 1009 |
+
"logps/chosen": -0.7043062448501587,
|
| 1010 |
+
"logps/rejected": -3.119410991668701,
|
| 1011 |
+
"loss": 34.9491,
|
| 1012 |
+
"nll_loss": 0.9785323143005371,
|
| 1013 |
+
"rewards/accuracies": 1.0,
|
| 1014 |
+
"rewards/chosen": -0.7043062448501587,
|
| 1015 |
+
"rewards/margins": 2.415104389190674,
|
| 1016 |
+
"rewards/rejected": -3.119410991668701,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.6921481481481484,
|
| 1021 |
+
"grad_norm": 87.91110229492188,
|
| 1022 |
+
"learning_rate": 1.3736045660864034e-06,
|
| 1023 |
+
"log_odds_chosen": 3.2196686267852783,
|
| 1024 |
+
"log_odds_ratio": -0.10369835793972015,
|
| 1025 |
+
"logits/chosen": 144.8685302734375,
|
| 1026 |
+
"logits/rejected": 124.35963439941406,
|
| 1027 |
+
"logps/chosen": -0.7213758230209351,
|
| 1028 |
+
"logps/rejected": -3.1755340099334717,
|
| 1029 |
+
"loss": 36.0271,
|
| 1030 |
+
"nll_loss": 1.0363926887512207,
|
| 1031 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1032 |
+
"rewards/chosen": -0.7213758230209351,
|
| 1033 |
+
"rewards/margins": 2.454158306121826,
|
| 1034 |
+
"rewards/rejected": -3.1755340099334717,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.7395555555555555,
|
| 1039 |
+
"grad_norm": 64.4977798461914,
|
| 1040 |
+
"learning_rate": 9.565939833279192e-07,
|
| 1041 |
+
"log_odds_chosen": 3.287095546722412,
|
| 1042 |
+
"log_odds_ratio": -0.0926838293671608,
|
| 1043 |
+
"logits/chosen": 142.08680725097656,
|
| 1044 |
+
"logits/rejected": 134.31768798828125,
|
| 1045 |
+
"logps/chosen": -0.6762581467628479,
|
| 1046 |
+
"logps/rejected": -3.168743848800659,
|
| 1047 |
+
"loss": 36.5849,
|
| 1048 |
+
"nll_loss": 1.0401983261108398,
|
| 1049 |
+
"rewards/accuracies": 1.0,
|
| 1050 |
+
"rewards/chosen": -0.6762581467628479,
|
| 1051 |
+
"rewards/margins": 2.492486000061035,
|
| 1052 |
+
"rewards/rejected": -3.168743848800659,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.786962962962963,
|
| 1057 |
+
"grad_norm": 68.50914001464844,
|
| 1058 |
+
"learning_rate": 6.136378865420872e-07,
|
| 1059 |
+
"log_odds_chosen": 3.0792453289031982,
|
| 1060 |
+
"log_odds_ratio": -0.1326889842748642,
|
| 1061 |
+
"logits/chosen": 144.2269287109375,
|
| 1062 |
+
"logits/rejected": 135.65611267089844,
|
| 1063 |
+
"logps/chosen": -0.7598322629928589,
|
| 1064 |
+
"logps/rejected": -3.1091818809509277,
|
| 1065 |
+
"loss": 36.4593,
|
| 1066 |
+
"nll_loss": 1.0354132652282715,
|
| 1067 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1068 |
+
"rewards/chosen": -0.7598322629928589,
|
| 1069 |
+
"rewards/margins": 2.3493494987487793,
|
| 1070 |
+
"rewards/rejected": -3.1091818809509277,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.83437037037037,
|
| 1075 |
+
"grad_norm": 85.84087371826172,
|
| 1076 |
+
"learning_rate": 3.45792591853214e-07,
|
| 1077 |
+
"log_odds_chosen": 3.3315589427948,
|
| 1078 |
+
"log_odds_ratio": -0.09676862508058548,
|
| 1079 |
+
"logits/chosen": 154.02792358398438,
|
| 1080 |
+
"logits/rejected": 134.80799865722656,
|
| 1081 |
+
"logps/chosen": -0.7510647773742676,
|
| 1082 |
+
"logps/rejected": -3.340522050857544,
|
| 1083 |
+
"loss": 36.5466,
|
| 1084 |
+
"nll_loss": 1.073294997215271,
|
| 1085 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1086 |
+
"rewards/chosen": -0.7510647773742676,
|
| 1087 |
+
"rewards/margins": 2.5894570350646973,
|
| 1088 |
+
"rewards/rejected": -3.340522050857544,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.8817777777777778,
|
| 1093 |
+
"grad_norm": 70.5570297241211,
|
| 1094 |
+
"learning_rate": 1.538830716302092e-07,
|
| 1095 |
+
"log_odds_chosen": 3.1449217796325684,
|
| 1096 |
+
"log_odds_ratio": -0.1062285304069519,
|
| 1097 |
+
"logits/chosen": 143.44528198242188,
|
| 1098 |
+
"logits/rejected": 130.6827392578125,
|
| 1099 |
+
"logps/chosen": -0.7541910409927368,
|
| 1100 |
+
"logps/rejected": -3.144932270050049,
|
| 1101 |
+
"loss": 35.8455,
|
| 1102 |
+
"nll_loss": 1.0105626583099365,
|
| 1103 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1104 |
+
"rewards/chosen": -0.7541910409927368,
|
| 1105 |
+
"rewards/margins": 2.3907411098480225,
|
| 1106 |
+
"rewards/rejected": -3.144932270050049,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9291851851851853,
|
| 1111 |
+
"grad_norm": 105.31517791748047,
|
| 1112 |
+
"learning_rate": 3.8500413544415025e-08,
|
| 1113 |
+
"log_odds_chosen": 3.273428440093994,
|
| 1114 |
+
"log_odds_ratio": -0.09262979030609131,
|
| 1115 |
+
"logits/chosen": 142.3734588623047,
|
| 1116 |
+
"logits/rejected": 128.36416625976562,
|
| 1117 |
+
"logps/chosen": -0.704811692237854,
|
| 1118 |
+
"logps/rejected": -3.1651034355163574,
|
| 1119 |
+
"loss": 35.5275,
|
| 1120 |
+
"nll_loss": 1.0222610235214233,
|
| 1121 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1122 |
+
"rewards/chosen": -0.704811692237854,
|
| 1123 |
+
"rewards/margins": 2.460291624069214,
|
| 1124 |
+
"rewards/rejected": -3.1651034355163574,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.9765925925925925,
|
| 1129 |
+
"grad_norm": 73.82266998291016,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 3.2872276306152344,
|
| 1132 |
+
"log_odds_ratio": -0.09776648133993149,
|
| 1133 |
+
"logits/chosen": 140.93003845214844,
|
| 1134 |
+
"logits/rejected": 121.86959075927734,
|
| 1135 |
+
"logps/chosen": -0.7005801796913147,
|
| 1136 |
+
"logps/rejected": -3.1921160221099854,
|
| 1137 |
+
"loss": 34.8233,
|
| 1138 |
+
"nll_loss": 1.0123968124389648,
|
| 1139 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1140 |
+
"rewards/chosen": -0.7005801796913147,
|
| 1141 |
+
"rewards/margins": 2.4915356636047363,
|
| 1142 |
+
"rewards/rejected": -3.1921160221099854,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.9765925925925925,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 81.45431906079489,
|
| 1150 |
+
"train_runtime": 8117.5608,
|
| 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 |
+
}
|