Model save
Browse files- README.md +67 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1176 -0
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-7b
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: gemma-7b-borpo-noisy-1e-5-norm
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-borpo-noisy-1e-5-norm
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="silviasapora/gemma-7b-borpo-noisy-1e-5-norm", 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/7kgum8xr)
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
This model was trained with ORPO, a method introduced in [ORPO: Monolithic Preference Optimization without Reference Model](https://huggingface.co/papers/2403.07691).
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.13.0
|
| 38 |
+
- Transformers: 4.46.1
|
| 39 |
+
- Pytorch: 2.4.0
|
| 40 |
+
- Datasets: 3.1.0
|
| 41 |
+
- Tokenizers: 0.20.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite ORPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{hong2024orpo,
|
| 49 |
+
title = {{ORPO: Monolithic Preference Optimization without Reference Model}},
|
| 50 |
+
author = {Jiwoo Hong and Noah Lee and James Thorne},
|
| 51 |
+
year = 2024,
|
| 52 |
+
eprint = {arXiv:2403.07691}
|
| 53 |
+
}
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
Cite TRL as:
|
| 57 |
+
|
| 58 |
+
```bibtex
|
| 59 |
+
@misc{vonwerra2022trl,
|
| 60 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 61 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 62 |
+
year = 2020,
|
| 63 |
+
journal = {GitHub repository},
|
| 64 |
+
publisher = {GitHub},
|
| 65 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 66 |
+
}
|
| 67 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.985781990521327,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 31.780521017407615,
|
| 5 |
+
"train_runtime": 7128.5338,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.841,
|
| 8 |
+
"train_steps_per_second": 0.044
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.985781990521327,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 31.780521017407615,
|
| 5 |
+
"train_runtime": 7128.5338,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.841,
|
| 8 |
+
"train_steps_per_second": 0.044
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.985781990521327,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 315,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.04739336492890995,
|
| 13 |
+
"grad_norm": 720.0,
|
| 14 |
+
"learning_rate": 1.5625e-06,
|
| 15 |
+
"log_odds_chosen": 3.653895616531372,
|
| 16 |
+
"log_odds_ratio": -11.719749450683594,
|
| 17 |
+
"logits/chosen": 278.66485595703125,
|
| 18 |
+
"logits/rejected": 294.8902282714844,
|
| 19 |
+
"logps/chosen": -22.453229904174805,
|
| 20 |
+
"logps/rejected": -26.106348037719727,
|
| 21 |
+
"loss": 103.8003,
|
| 22 |
+
"nll_loss": 6.5329999923706055,
|
| 23 |
+
"rewards/accuracies": 0.543749988079071,
|
| 24 |
+
"rewards/chosen": -11.226614952087402,
|
| 25 |
+
"rewards/margins": 1.8265600204467773,
|
| 26 |
+
"rewards/rejected": -13.053174018859863,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.0947867298578199,
|
| 31 |
+
"grad_norm": 304.0,
|
| 32 |
+
"learning_rate": 3.125e-06,
|
| 33 |
+
"log_odds_chosen": 3.6431682109832764,
|
| 34 |
+
"log_odds_ratio": -7.441667079925537,
|
| 35 |
+
"logits/chosen": 269.0749206542969,
|
| 36 |
+
"logits/rejected": 282.71282958984375,
|
| 37 |
+
"logps/chosen": -18.20242691040039,
|
| 38 |
+
"logps/rejected": -21.8449649810791,
|
| 39 |
+
"loss": 99.1968,
|
| 40 |
+
"nll_loss": 5.850668907165527,
|
| 41 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 42 |
+
"rewards/chosen": -9.101213455200195,
|
| 43 |
+
"rewards/margins": 1.8212683200836182,
|
| 44 |
+
"rewards/rejected": -10.92248249053955,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14218009478672985,
|
| 49 |
+
"grad_norm": 139.0,
|
| 50 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 51 |
+
"log_odds_chosen": 5.687544345855713,
|
| 52 |
+
"log_odds_ratio": -7.769516944885254,
|
| 53 |
+
"logits/chosen": 264.29461669921875,
|
| 54 |
+
"logits/rejected": 291.05987548828125,
|
| 55 |
+
"logps/chosen": -18.975749969482422,
|
| 56 |
+
"logps/rejected": -24.661663055419922,
|
| 57 |
+
"loss": 95.6825,
|
| 58 |
+
"nll_loss": 6.186770439147949,
|
| 59 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 60 |
+
"rewards/chosen": -9.487874984741211,
|
| 61 |
+
"rewards/margins": 2.8429577350616455,
|
| 62 |
+
"rewards/rejected": -12.330831527709961,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.1895734597156398,
|
| 67 |
+
"grad_norm": 153.0,
|
| 68 |
+
"learning_rate": 6.25e-06,
|
| 69 |
+
"log_odds_chosen": -1.2323758602142334,
|
| 70 |
+
"log_odds_ratio": -10.211746215820312,
|
| 71 |
+
"logits/chosen": 292.7019958496094,
|
| 72 |
+
"logits/rejected": 290.5914001464844,
|
| 73 |
+
"logps/chosen": -20.584659576416016,
|
| 74 |
+
"logps/rejected": -19.353229522705078,
|
| 75 |
+
"loss": 99.4907,
|
| 76 |
+
"nll_loss": 6.084651470184326,
|
| 77 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 78 |
+
"rewards/chosen": -10.292329788208008,
|
| 79 |
+
"rewards/margins": -0.6157160997390747,
|
| 80 |
+
"rewards/rejected": -9.676614761352539,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23696682464454977,
|
| 85 |
+
"grad_norm": 81.5,
|
| 86 |
+
"learning_rate": 7.8125e-06,
|
| 87 |
+
"log_odds_chosen": 0.21437835693359375,
|
| 88 |
+
"log_odds_ratio": -9.132922172546387,
|
| 89 |
+
"logits/chosen": 278.7820129394531,
|
| 90 |
+
"logits/rejected": 288.18951416015625,
|
| 91 |
+
"logps/chosen": -18.19257354736328,
|
| 92 |
+
"logps/rejected": -18.40748405456543,
|
| 93 |
+
"loss": 88.5475,
|
| 94 |
+
"nll_loss": 4.977797508239746,
|
| 95 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 96 |
+
"rewards/chosen": -9.09628677368164,
|
| 97 |
+
"rewards/margins": 0.10745634883642197,
|
| 98 |
+
"rewards/rejected": -9.203742027282715,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.2843601895734597,
|
| 103 |
+
"grad_norm": 124.0,
|
| 104 |
+
"learning_rate": 9.375000000000001e-06,
|
| 105 |
+
"log_odds_chosen": 1.75958251953125,
|
| 106 |
+
"log_odds_ratio": -9.482071876525879,
|
| 107 |
+
"logits/chosen": 280.55633544921875,
|
| 108 |
+
"logits/rejected": 301.93743896484375,
|
| 109 |
+
"logps/chosen": -19.764301300048828,
|
| 110 |
+
"logps/rejected": -21.52188491821289,
|
| 111 |
+
"loss": 98.2752,
|
| 112 |
+
"nll_loss": 6.01733922958374,
|
| 113 |
+
"rewards/accuracies": 0.53125,
|
| 114 |
+
"rewards/chosen": -9.882150650024414,
|
| 115 |
+
"rewards/margins": 0.8787924647331238,
|
| 116 |
+
"rewards/rejected": -10.760942459106445,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33175355450236965,
|
| 121 |
+
"grad_norm": 239.0,
|
| 122 |
+
"learning_rate": 9.997227514697568e-06,
|
| 123 |
+
"log_odds_chosen": -0.15938568115234375,
|
| 124 |
+
"log_odds_ratio": -7.877626895904541,
|
| 125 |
+
"logits/chosen": 297.74163818359375,
|
| 126 |
+
"logits/rejected": 288.8265075683594,
|
| 127 |
+
"logps/chosen": -17.699235916137695,
|
| 128 |
+
"logps/rejected": -17.539152145385742,
|
| 129 |
+
"loss": 83.3604,
|
| 130 |
+
"nll_loss": 5.2883453369140625,
|
| 131 |
+
"rewards/accuracies": 0.5,
|
| 132 |
+
"rewards/chosen": -8.849617958068848,
|
| 133 |
+
"rewards/margins": -0.08004142343997955,
|
| 134 |
+
"rewards/rejected": -8.769576072692871,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3791469194312796,
|
| 139 |
+
"grad_norm": 410.0,
|
| 140 |
+
"learning_rate": 9.980295682286924e-06,
|
| 141 |
+
"log_odds_chosen": 2.3873629570007324,
|
| 142 |
+
"log_odds_ratio": -3.3676114082336426,
|
| 143 |
+
"logits/chosen": 300.3001708984375,
|
| 144 |
+
"logits/rejected": 310.92047119140625,
|
| 145 |
+
"logps/chosen": -10.030499458312988,
|
| 146 |
+
"logps/rejected": -12.416150093078613,
|
| 147 |
+
"loss": 73.3518,
|
| 148 |
+
"nll_loss": 4.223599433898926,
|
| 149 |
+
"rewards/accuracies": 0.512499988079071,
|
| 150 |
+
"rewards/chosen": -5.015249729156494,
|
| 151 |
+
"rewards/margins": 1.1928250789642334,
|
| 152 |
+
"rewards/rejected": -6.208075046539307,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4265402843601896,
|
| 157 |
+
"grad_norm": 308.0,
|
| 158 |
+
"learning_rate": 9.94802437440896e-06,
|
| 159 |
+
"log_odds_chosen": 1.004201889038086,
|
| 160 |
+
"log_odds_ratio": -3.536766767501831,
|
| 161 |
+
"logits/chosen": 309.2627258300781,
|
| 162 |
+
"logits/rejected": 307.8229675292969,
|
| 163 |
+
"logps/chosen": -8.605690002441406,
|
| 164 |
+
"logps/rejected": -9.609885215759277,
|
| 165 |
+
"loss": 52.2283,
|
| 166 |
+
"nll_loss": 3.156938076019287,
|
| 167 |
+
"rewards/accuracies": 0.518750011920929,
|
| 168 |
+
"rewards/chosen": -4.302845001220703,
|
| 169 |
+
"rewards/margins": 0.5020972490310669,
|
| 170 |
+
"rewards/rejected": -4.804942607879639,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.47393364928909953,
|
| 175 |
+
"grad_norm": 112.0,
|
| 176 |
+
"learning_rate": 9.90051298775959e-06,
|
| 177 |
+
"log_odds_chosen": 0.10521616786718369,
|
| 178 |
+
"log_odds_ratio": -1.6550161838531494,
|
| 179 |
+
"logits/chosen": 334.4619445800781,
|
| 180 |
+
"logits/rejected": 343.36077880859375,
|
| 181 |
+
"logps/chosen": -3.9240214824676514,
|
| 182 |
+
"logps/rejected": -4.02620792388916,
|
| 183 |
+
"loss": 36.9834,
|
| 184 |
+
"nll_loss": 2.3400511741638184,
|
| 185 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 186 |
+
"rewards/chosen": -1.9620107412338257,
|
| 187 |
+
"rewards/margins": 0.05109361559152603,
|
| 188 |
+
"rewards/rejected": -2.01310396194458,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5213270142180095,
|
| 193 |
+
"grad_norm": 50.25,
|
| 194 |
+
"learning_rate": 9.837907858981536e-06,
|
| 195 |
+
"log_odds_chosen": 0.29174837470054626,
|
| 196 |
+
"log_odds_ratio": -0.8331424593925476,
|
| 197 |
+
"logits/chosen": 334.2044677734375,
|
| 198 |
+
"logits/rejected": 351.3958435058594,
|
| 199 |
+
"logps/chosen": -1.8196113109588623,
|
| 200 |
+
"logps/rejected": -2.076477527618408,
|
| 201 |
+
"loss": 28.4647,
|
| 202 |
+
"nll_loss": 1.7829310894012451,
|
| 203 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 204 |
+
"rewards/chosen": -0.9098056554794312,
|
| 205 |
+
"rewards/margins": 0.1284329891204834,
|
| 206 |
+
"rewards/rejected": -1.038238763809204,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5687203791469194,
|
| 211 |
+
"grad_norm": 47.75,
|
| 212 |
+
"learning_rate": 9.760401813942641e-06,
|
| 213 |
+
"log_odds_chosen": 0.35578861832618713,
|
| 214 |
+
"log_odds_ratio": -0.7586129903793335,
|
| 215 |
+
"logits/chosen": 327.1716613769531,
|
| 216 |
+
"logits/rejected": 352.7845764160156,
|
| 217 |
+
"logps/chosen": -1.6458431482315063,
|
| 218 |
+
"logps/rejected": -1.9578958749771118,
|
| 219 |
+
"loss": 26.7026,
|
| 220 |
+
"nll_loss": 1.681125283241272,
|
| 221 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 222 |
+
"rewards/chosen": -0.8229215741157532,
|
| 223 |
+
"rewards/margins": 0.156026229262352,
|
| 224 |
+
"rewards/rejected": -0.9789479374885559,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6161137440758294,
|
| 229 |
+
"grad_norm": 34.5,
|
| 230 |
+
"learning_rate": 9.668233573825794e-06,
|
| 231 |
+
"log_odds_chosen": 0.31834372878074646,
|
| 232 |
+
"log_odds_ratio": -0.7771207094192505,
|
| 233 |
+
"logits/chosen": 322.9273681640625,
|
| 234 |
+
"logits/rejected": 341.0210266113281,
|
| 235 |
+
"logps/chosen": -1.5688014030456543,
|
| 236 |
+
"logps/rejected": -1.8273004293441772,
|
| 237 |
+
"loss": 26.4611,
|
| 238 |
+
"nll_loss": 1.6627897024154663,
|
| 239 |
+
"rewards/accuracies": 0.606249988079071,
|
| 240 |
+
"rewards/chosen": -0.7844007015228271,
|
| 241 |
+
"rewards/margins": 0.12924957275390625,
|
| 242 |
+
"rewards/rejected": -0.9136502146720886,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6635071090047393,
|
| 247 |
+
"grad_norm": 47.25,
|
| 248 |
+
"learning_rate": 9.56168701985981e-06,
|
| 249 |
+
"log_odds_chosen": 0.31223994493484497,
|
| 250 |
+
"log_odds_ratio": -0.7746738791465759,
|
| 251 |
+
"logits/chosen": 353.25103759765625,
|
| 252 |
+
"logits/rejected": 372.4833984375,
|
| 253 |
+
"logps/chosen": -1.557328701019287,
|
| 254 |
+
"logps/rejected": -1.8247381448745728,
|
| 255 |
+
"loss": 25.781,
|
| 256 |
+
"nll_loss": 1.557205319404602,
|
| 257 |
+
"rewards/accuracies": 0.581250011920929,
|
| 258 |
+
"rewards/chosen": -0.7786643505096436,
|
| 259 |
+
"rewards/margins": 0.13370472192764282,
|
| 260 |
+
"rewards/rejected": -0.9123690724372864,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7109004739336493,
|
| 265 |
+
"grad_norm": 36.0,
|
| 266 |
+
"learning_rate": 9.441090318955843e-06,
|
| 267 |
+
"log_odds_chosen": 0.3869365155696869,
|
| 268 |
+
"log_odds_ratio": -0.778424084186554,
|
| 269 |
+
"logits/chosen": 346.4082336425781,
|
| 270 |
+
"logits/rejected": 360.0977783203125,
|
| 271 |
+
"logps/chosen": -1.5176082849502563,
|
| 272 |
+
"logps/rejected": -1.8383760452270508,
|
| 273 |
+
"loss": 25.2293,
|
| 274 |
+
"nll_loss": 1.5534436702728271,
|
| 275 |
+
"rewards/accuracies": 0.637499988079071,
|
| 276 |
+
"rewards/chosen": -0.7588041424751282,
|
| 277 |
+
"rewards/margins": 0.1603839099407196,
|
| 278 |
+
"rewards/rejected": -0.9191880226135254,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7582938388625592,
|
| 283 |
+
"grad_norm": 83.0,
|
| 284 |
+
"learning_rate": 9.306814912942445e-06,
|
| 285 |
+
"log_odds_chosen": 0.3693597912788391,
|
| 286 |
+
"log_odds_ratio": -0.7350634336471558,
|
| 287 |
+
"logits/chosen": 346.6322937011719,
|
| 288 |
+
"logits/rejected": 358.0809326171875,
|
| 289 |
+
"logps/chosen": -1.4770724773406982,
|
| 290 |
+
"logps/rejected": -1.7809489965438843,
|
| 291 |
+
"loss": 24.6783,
|
| 292 |
+
"nll_loss": 1.5355550050735474,
|
| 293 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 294 |
+
"rewards/chosen": -0.7385362386703491,
|
| 295 |
+
"rewards/margins": 0.1519382894039154,
|
| 296 |
+
"rewards/rejected": -0.8904744982719421,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8056872037914692,
|
| 301 |
+
"grad_norm": 28.5,
|
| 302 |
+
"learning_rate": 9.159274374512444e-06,
|
| 303 |
+
"log_odds_chosen": 0.38480785489082336,
|
| 304 |
+
"log_odds_ratio": -0.7371683120727539,
|
| 305 |
+
"logits/chosen": 356.5403137207031,
|
| 306 |
+
"logits/rejected": 372.90631103515625,
|
| 307 |
+
"logps/chosen": -1.5246539115905762,
|
| 308 |
+
"logps/rejected": -1.854161024093628,
|
| 309 |
+
"loss": 24.3913,
|
| 310 |
+
"nll_loss": 1.5450783967971802,
|
| 311 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 312 |
+
"rewards/chosen": -0.7623269557952881,
|
| 313 |
+
"rewards/margins": 0.16475361585617065,
|
| 314 |
+
"rewards/rejected": -0.927080512046814,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8530805687203792,
|
| 319 |
+
"grad_norm": 44.75,
|
| 320 |
+
"learning_rate": 8.99892313340537e-06,
|
| 321 |
+
"log_odds_chosen": 0.2174159735441208,
|
| 322 |
+
"log_odds_ratio": -0.8837092518806458,
|
| 323 |
+
"logits/chosen": 354.29132080078125,
|
| 324 |
+
"logits/rejected": 361.8697814941406,
|
| 325 |
+
"logps/chosen": -1.6351244449615479,
|
| 326 |
+
"logps/rejected": -1.8084831237792969,
|
| 327 |
+
"loss": 25.1868,
|
| 328 |
+
"nll_loss": 1.6281230449676514,
|
| 329 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 330 |
+
"rewards/chosen": -0.8175622224807739,
|
| 331 |
+
"rewards/margins": 0.08667941391468048,
|
| 332 |
+
"rewards/rejected": -0.9042415618896484,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9004739336492891,
|
| 337 |
+
"grad_norm": 41.0,
|
| 338 |
+
"learning_rate": 8.826255076748823e-06,
|
| 339 |
+
"log_odds_chosen": 0.14011089503765106,
|
| 340 |
+
"log_odds_ratio": -0.7930618524551392,
|
| 341 |
+
"logits/chosen": 362.77874755859375,
|
| 342 |
+
"logits/rejected": 373.4800720214844,
|
| 343 |
+
"logps/chosen": -1.4461233615875244,
|
| 344 |
+
"logps/rejected": -1.5512562990188599,
|
| 345 |
+
"loss": 24.0704,
|
| 346 |
+
"nll_loss": 1.4283101558685303,
|
| 347 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 348 |
+
"rewards/chosen": -0.7230616807937622,
|
| 349 |
+
"rewards/margins": 0.05256646126508713,
|
| 350 |
+
"rewards/rejected": -0.7756281495094299,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9478672985781991,
|
| 355 |
+
"grad_norm": 70.5,
|
| 356 |
+
"learning_rate": 8.641802027869774e-06,
|
| 357 |
+
"log_odds_chosen": 0.2718258798122406,
|
| 358 |
+
"log_odds_ratio": -0.7795127630233765,
|
| 359 |
+
"logits/chosen": 367.4668273925781,
|
| 360 |
+
"logits/rejected": 367.6637878417969,
|
| 361 |
+
"logps/chosen": -1.4626038074493408,
|
| 362 |
+
"logps/rejected": -1.709529161453247,
|
| 363 |
+
"loss": 24.1424,
|
| 364 |
+
"nll_loss": 1.538220763206482,
|
| 365 |
+
"rewards/accuracies": 0.574999988079071,
|
| 366 |
+
"rewards/chosen": -0.7313019037246704,
|
| 367 |
+
"rewards/margins": 0.12346267700195312,
|
| 368 |
+
"rewards/rejected": -0.8547645807266235,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.995260663507109,
|
| 373 |
+
"grad_norm": 31.75,
|
| 374 |
+
"learning_rate": 8.446132108261136e-06,
|
| 375 |
+
"log_odds_chosen": 0.2905386984348297,
|
| 376 |
+
"log_odds_ratio": -0.6851338744163513,
|
| 377 |
+
"logits/chosen": 360.0987548828125,
|
| 378 |
+
"logits/rejected": 378.06414794921875,
|
| 379 |
+
"logps/chosen": -1.3334792852401733,
|
| 380 |
+
"logps/rejected": -1.5764299631118774,
|
| 381 |
+
"loss": 23.5124,
|
| 382 |
+
"nll_loss": 1.4705913066864014,
|
| 383 |
+
"rewards/accuracies": 0.606249988079071,
|
| 384 |
+
"rewards/chosen": -0.6667396426200867,
|
| 385 |
+
"rewards/margins": 0.12147532403469086,
|
| 386 |
+
"rewards/rejected": -0.7882149815559387,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.042654028436019,
|
| 391 |
+
"grad_norm": 30.125,
|
| 392 |
+
"learning_rate": 8.23984798774876e-06,
|
| 393 |
+
"log_odds_chosen": 0.2926952838897705,
|
| 394 |
+
"log_odds_ratio": -0.7383573055267334,
|
| 395 |
+
"logits/chosen": 356.43695068359375,
|
| 396 |
+
"logits/rejected": 380.3526611328125,
|
| 397 |
+
"logps/chosen": -1.3705774545669556,
|
| 398 |
+
"logps/rejected": -1.615526556968689,
|
| 399 |
+
"loss": 22.8403,
|
| 400 |
+
"nll_loss": 1.443694829940796,
|
| 401 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 402 |
+
"rewards/chosen": -0.6852887272834778,
|
| 403 |
+
"rewards/margins": 0.12247464805841446,
|
| 404 |
+
"rewards/rejected": -0.8077632784843445,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0900473933649288,
|
| 409 |
+
"grad_norm": 36.25,
|
| 410 |
+
"learning_rate": 8.023585028248435e-06,
|
| 411 |
+
"log_odds_chosen": 0.456248939037323,
|
| 412 |
+
"log_odds_ratio": -0.6926681399345398,
|
| 413 |
+
"logits/chosen": 356.6784973144531,
|
| 414 |
+
"logits/rejected": 381.8028259277344,
|
| 415 |
+
"logps/chosen": -1.3445045948028564,
|
| 416 |
+
"logps/rejected": -1.689343810081482,
|
| 417 |
+
"loss": 22.9507,
|
| 418 |
+
"nll_loss": 1.427918553352356,
|
| 419 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 420 |
+
"rewards/chosen": -0.6722522974014282,
|
| 421 |
+
"rewards/margins": 0.17241965234279633,
|
| 422 |
+
"rewards/rejected": -0.844671905040741,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1374407582938388,
|
| 427 |
+
"grad_norm": 32.0,
|
| 428 |
+
"learning_rate": 7.798009326830167e-06,
|
| 429 |
+
"log_odds_chosen": 0.44028574228286743,
|
| 430 |
+
"log_odds_ratio": -0.6239734888076782,
|
| 431 |
+
"logits/chosen": 351.1431579589844,
|
| 432 |
+
"logits/rejected": 363.55499267578125,
|
| 433 |
+
"logps/chosen": -1.2665112018585205,
|
| 434 |
+
"logps/rejected": -1.6185169219970703,
|
| 435 |
+
"loss": 22.7249,
|
| 436 |
+
"nll_loss": 1.449618935585022,
|
| 437 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 438 |
+
"rewards/chosen": -0.6332556009292603,
|
| 439 |
+
"rewards/margins": 0.1760028749704361,
|
| 440 |
+
"rewards/rejected": -0.8092584609985352,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1848341232227488,
|
| 445 |
+
"grad_norm": 30.375,
|
| 446 |
+
"learning_rate": 7.563815664117173e-06,
|
| 447 |
+
"log_odds_chosen": 0.3645080029964447,
|
| 448 |
+
"log_odds_ratio": -0.6768008470535278,
|
| 449 |
+
"logits/chosen": 369.084716796875,
|
| 450 |
+
"logits/rejected": 371.45977783203125,
|
| 451 |
+
"logps/chosen": -1.303069829940796,
|
| 452 |
+
"logps/rejected": -1.5789194107055664,
|
| 453 |
+
"loss": 21.767,
|
| 454 |
+
"nll_loss": 1.3582165241241455,
|
| 455 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 456 |
+
"rewards/chosen": -0.651534914970398,
|
| 457 |
+
"rewards/margins": 0.13792480528354645,
|
| 458 |
+
"rewards/rejected": -0.7894597053527832,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2322274881516588,
|
| 463 |
+
"grad_norm": 31.125,
|
| 464 |
+
"learning_rate": 7.321725364338566e-06,
|
| 465 |
+
"log_odds_chosen": 0.15869663655757904,
|
| 466 |
+
"log_odds_ratio": -0.7532464861869812,
|
| 467 |
+
"logits/chosen": 353.90032958984375,
|
| 468 |
+
"logits/rejected": 370.54364013671875,
|
| 469 |
+
"logps/chosen": -1.2809131145477295,
|
| 470 |
+
"logps/rejected": -1.387822151184082,
|
| 471 |
+
"loss": 21.6046,
|
| 472 |
+
"nll_loss": 1.3660101890563965,
|
| 473 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 474 |
+
"rewards/chosen": -0.6404565572738647,
|
| 475 |
+
"rewards/margins": 0.053454458713531494,
|
| 476 |
+
"rewards/rejected": -0.693911075592041,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.2796208530805688,
|
| 481 |
+
"grad_norm": 51.25,
|
| 482 |
+
"learning_rate": 7.072484073626872e-06,
|
| 483 |
+
"log_odds_chosen": 0.4641496241092682,
|
| 484 |
+
"log_odds_ratio": -0.6374386548995972,
|
| 485 |
+
"logits/chosen": 360.2814636230469,
|
| 486 |
+
"logits/rejected": 370.50323486328125,
|
| 487 |
+
"logps/chosen": -1.2486190795898438,
|
| 488 |
+
"logps/rejected": -1.6152652502059937,
|
| 489 |
+
"loss": 22.2528,
|
| 490 |
+
"nll_loss": 1.3934818506240845,
|
| 491 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 492 |
+
"rewards/chosen": -0.6243095397949219,
|
| 493 |
+
"rewards/margins": 0.18332314491271973,
|
| 494 |
+
"rewards/rejected": -0.8076326251029968,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3270142180094786,
|
| 499 |
+
"grad_norm": 50.25,
|
| 500 |
+
"learning_rate": 6.816859463403271e-06,
|
| 501 |
+
"log_odds_chosen": 0.2742091715335846,
|
| 502 |
+
"log_odds_ratio": -0.7456444501876831,
|
| 503 |
+
"logits/chosen": 355.84222412109375,
|
| 504 |
+
"logits/rejected": 354.6243591308594,
|
| 505 |
+
"logps/chosen": -1.2655051946640015,
|
| 506 |
+
"logps/rejected": -1.4896043539047241,
|
| 507 |
+
"loss": 22.9608,
|
| 508 |
+
"nll_loss": 1.401302695274353,
|
| 509 |
+
"rewards/accuracies": 0.5625,
|
| 510 |
+
"rewards/chosen": -0.6327525973320007,
|
| 511 |
+
"rewards/margins": 0.11204960197210312,
|
| 512 |
+
"rewards/rejected": -0.7448021769523621,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.3744075829383886,
|
| 517 |
+
"grad_norm": 42.75,
|
| 518 |
+
"learning_rate": 6.555638865924221e-06,
|
| 519 |
+
"log_odds_chosen": 0.5057348012924194,
|
| 520 |
+
"log_odds_ratio": -0.5890580415725708,
|
| 521 |
+
"logits/chosen": 345.1250915527344,
|
| 522 |
+
"logits/rejected": 369.5255432128906,
|
| 523 |
+
"logps/chosen": -1.169845461845398,
|
| 524 |
+
"logps/rejected": -1.5732532739639282,
|
| 525 |
+
"loss": 22.323,
|
| 526 |
+
"nll_loss": 1.3889384269714355,
|
| 527 |
+
"rewards/accuracies": 0.706250011920929,
|
| 528 |
+
"rewards/chosen": -0.584922730922699,
|
| 529 |
+
"rewards/margins": 0.20170390605926514,
|
| 530 |
+
"rewards/rejected": -0.7866266369819641,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.4218009478672986,
|
| 535 |
+
"grad_norm": 61.25,
|
| 536 |
+
"learning_rate": 6.289626849272062e-06,
|
| 537 |
+
"log_odds_chosen": 0.21882423758506775,
|
| 538 |
+
"log_odds_ratio": -0.7196656465530396,
|
| 539 |
+
"logits/chosen": 353.15582275390625,
|
| 540 |
+
"logits/rejected": 339.25982666015625,
|
| 541 |
+
"logps/chosen": -1.1948503255844116,
|
| 542 |
+
"logps/rejected": -1.3711706399917603,
|
| 543 |
+
"loss": 22.3161,
|
| 544 |
+
"nll_loss": 1.4074368476867676,
|
| 545 |
+
"rewards/accuracies": 0.574999988079071,
|
| 546 |
+
"rewards/chosen": -0.5974251627922058,
|
| 547 |
+
"rewards/margins": 0.08816017955541611,
|
| 548 |
+
"rewards/rejected": -0.6855853199958801,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4691943127962086,
|
| 553 |
+
"grad_norm": 36.25,
|
| 554 |
+
"learning_rate": 6.0196427392587085e-06,
|
| 555 |
+
"log_odds_chosen": 0.34388962388038635,
|
| 556 |
+
"log_odds_ratio": -0.6600432395935059,
|
| 557 |
+
"logits/chosen": 352.5672302246094,
|
| 558 |
+
"logits/rejected": 373.4999084472656,
|
| 559 |
+
"logps/chosen": -1.137880563735962,
|
| 560 |
+
"logps/rejected": -1.3981521129608154,
|
| 561 |
+
"loss": 21.9751,
|
| 562 |
+
"nll_loss": 1.3336918354034424,
|
| 563 |
+
"rewards/accuracies": 0.625,
|
| 564 |
+
"rewards/chosen": -0.568940281867981,
|
| 565 |
+
"rewards/margins": 0.13013575971126556,
|
| 566 |
+
"rewards/rejected": -0.6990760564804077,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.5165876777251186,
|
| 571 |
+
"grad_norm": 42.0,
|
| 572 |
+
"learning_rate": 5.746518095875033e-06,
|
| 573 |
+
"log_odds_chosen": 0.24596929550170898,
|
| 574 |
+
"log_odds_ratio": -0.7056238651275635,
|
| 575 |
+
"logits/chosen": 355.90606689453125,
|
| 576 |
+
"logits/rejected": 360.6635437011719,
|
| 577 |
+
"logps/chosen": -1.2390286922454834,
|
| 578 |
+
"logps/rejected": -1.4282642602920532,
|
| 579 |
+
"loss": 22.5868,
|
| 580 |
+
"nll_loss": 1.4504239559173584,
|
| 581 |
+
"rewards/accuracies": 0.59375,
|
| 582 |
+
"rewards/chosen": -0.6195143461227417,
|
| 583 |
+
"rewards/margins": 0.09461767971515656,
|
| 584 |
+
"rewards/rejected": -0.7141321301460266,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5639810426540284,
|
| 589 |
+
"grad_norm": 34.75,
|
| 590 |
+
"learning_rate": 5.471094152058592e-06,
|
| 591 |
+
"log_odds_chosen": 0.29543933272361755,
|
| 592 |
+
"log_odds_ratio": -0.6770726442337036,
|
| 593 |
+
"logits/chosen": 362.8612060546875,
|
| 594 |
+
"logits/rejected": 357.1744384765625,
|
| 595 |
+
"logps/chosen": -1.2012829780578613,
|
| 596 |
+
"logps/rejected": -1.4275243282318115,
|
| 597 |
+
"loss": 21.8669,
|
| 598 |
+
"nll_loss": 1.353208303451538,
|
| 599 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 600 |
+
"rewards/chosen": -0.6006414890289307,
|
| 601 |
+
"rewards/margins": 0.11312057822942734,
|
| 602 |
+
"rewards/rejected": -0.7137621641159058,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6113744075829384,
|
| 607 |
+
"grad_norm": 30.625,
|
| 608 |
+
"learning_rate": 5.1942192226683385e-06,
|
| 609 |
+
"log_odds_chosen": 0.359115868806839,
|
| 610 |
+
"log_odds_ratio": -0.6642639636993408,
|
| 611 |
+
"logits/chosen": 354.8602294921875,
|
| 612 |
+
"logits/rejected": 367.0655212402344,
|
| 613 |
+
"logps/chosen": -1.2128181457519531,
|
| 614 |
+
"logps/rejected": -1.4822323322296143,
|
| 615 |
+
"loss": 21.8466,
|
| 616 |
+
"nll_loss": 1.3282759189605713,
|
| 617 |
+
"rewards/accuracies": 0.637499988079071,
|
| 618 |
+
"rewards/chosen": -0.6064090728759766,
|
| 619 |
+
"rewards/margins": 0.13470709323883057,
|
| 620 |
+
"rewards/rejected": -0.7411161661148071,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6587677725118484,
|
| 625 |
+
"grad_norm": 38.75,
|
| 626 |
+
"learning_rate": 4.916746091646808e-06,
|
| 627 |
+
"log_odds_chosen": 0.20625083148479462,
|
| 628 |
+
"log_odds_ratio": -0.7107352018356323,
|
| 629 |
+
"logits/chosen": 358.31890869140625,
|
| 630 |
+
"logits/rejected": 366.37408447265625,
|
| 631 |
+
"logps/chosen": -1.1632264852523804,
|
| 632 |
+
"logps/rejected": -1.3307292461395264,
|
| 633 |
+
"loss": 21.8714,
|
| 634 |
+
"nll_loss": 1.3562901020050049,
|
| 635 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 636 |
+
"rewards/chosen": -0.5816132426261902,
|
| 637 |
+
"rewards/margins": 0.08375142514705658,
|
| 638 |
+
"rewards/rejected": -0.6653646230697632,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7061611374407581,
|
| 643 |
+
"grad_norm": 63.25,
|
| 644 |
+
"learning_rate": 4.6395293854173395e-06,
|
| 645 |
+
"log_odds_chosen": 0.19339993596076965,
|
| 646 |
+
"log_odds_ratio": -0.7509890198707581,
|
| 647 |
+
"logits/chosen": 347.43951416015625,
|
| 648 |
+
"logits/rejected": 358.0630798339844,
|
| 649 |
+
"logps/chosen": -1.2565109729766846,
|
| 650 |
+
"logps/rejected": -1.391413688659668,
|
| 651 |
+
"loss": 22.1212,
|
| 652 |
+
"nll_loss": 1.416812539100647,
|
| 653 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 654 |
+
"rewards/chosen": -0.6282554864883423,
|
| 655 |
+
"rewards/margins": 0.06745139509439468,
|
| 656 |
+
"rewards/rejected": -0.695706844329834,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.7535545023696684,
|
| 661 |
+
"grad_norm": 27.375,
|
| 662 |
+
"learning_rate": 4.363422940606435e-06,
|
| 663 |
+
"log_odds_chosen": 0.4669272005558014,
|
| 664 |
+
"log_odds_ratio": -0.6223559975624084,
|
| 665 |
+
"logits/chosen": 350.216796875,
|
| 666 |
+
"logits/rejected": 363.3144836425781,
|
| 667 |
+
"logps/chosen": -1.1427295207977295,
|
| 668 |
+
"logps/rejected": -1.5226044654846191,
|
| 669 |
+
"loss": 21.106,
|
| 670 |
+
"nll_loss": 1.2993601560592651,
|
| 671 |
+
"rewards/accuracies": 0.668749988079071,
|
| 672 |
+
"rewards/chosen": -0.5713647603988647,
|
| 673 |
+
"rewards/margins": 0.189937561750412,
|
| 674 |
+
"rewards/rejected": -0.7613022327423096,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.8009478672985781,
|
| 679 |
+
"grad_norm": 31.0,
|
| 680 |
+
"learning_rate": 4.089277174198694e-06,
|
| 681 |
+
"log_odds_chosen": 0.23544028401374817,
|
| 682 |
+
"log_odds_ratio": -0.722697913646698,
|
| 683 |
+
"logits/chosen": 353.28558349609375,
|
| 684 |
+
"logits/rejected": 369.5932922363281,
|
| 685 |
+
"logps/chosen": -1.1963495016098022,
|
| 686 |
+
"logps/rejected": -1.3738784790039062,
|
| 687 |
+
"loss": 21.9431,
|
| 688 |
+
"nll_loss": 1.3263076543807983,
|
| 689 |
+
"rewards/accuracies": 0.606249988079071,
|
| 690 |
+
"rewards/chosen": -0.5981747508049011,
|
| 691 |
+
"rewards/margins": 0.08876445889472961,
|
| 692 |
+
"rewards/rejected": -0.6869392395019531,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8483412322274881,
|
| 697 |
+
"grad_norm": 37.5,
|
| 698 |
+
"learning_rate": 3.817936464224367e-06,
|
| 699 |
+
"log_odds_chosen": 0.41035833954811096,
|
| 700 |
+
"log_odds_ratio": -0.6357004642486572,
|
| 701 |
+
"logits/chosen": 342.2323303222656,
|
| 702 |
+
"logits/rejected": 361.4287414550781,
|
| 703 |
+
"logps/chosen": -1.1266499757766724,
|
| 704 |
+
"logps/rejected": -1.4432952404022217,
|
| 705 |
+
"loss": 21.7598,
|
| 706 |
+
"nll_loss": 1.3171429634094238,
|
| 707 |
+
"rewards/accuracies": 0.606249988079071,
|
| 708 |
+
"rewards/chosen": -0.5633249878883362,
|
| 709 |
+
"rewards/margins": 0.15832264721393585,
|
| 710 |
+
"rewards/rejected": -0.7216476202011108,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8957345971563981,
|
| 715 |
+
"grad_norm": 33.5,
|
| 716 |
+
"learning_rate": 3.55023654904709e-06,
|
| 717 |
+
"log_odds_chosen": 0.1588168442249298,
|
| 718 |
+
"log_odds_ratio": -0.7843244671821594,
|
| 719 |
+
"logits/chosen": 346.522216796875,
|
| 720 |
+
"logits/rejected": 360.4036865234375,
|
| 721 |
+
"logps/chosen": -1.248828649520874,
|
| 722 |
+
"logps/rejected": -1.3542238473892212,
|
| 723 |
+
"loss": 22.3502,
|
| 724 |
+
"nll_loss": 1.417950987815857,
|
| 725 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 726 |
+
"rewards/chosen": -0.624414324760437,
|
| 727 |
+
"rewards/margins": 0.05269758030772209,
|
| 728 |
+
"rewards/rejected": -0.6771119236946106,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.943127962085308,
|
| 733 |
+
"grad_norm": 38.25,
|
| 734 |
+
"learning_rate": 3.2870019532620744e-06,
|
| 735 |
+
"log_odds_chosen": 0.19328053295612335,
|
| 736 |
+
"log_odds_ratio": -0.6914501786231995,
|
| 737 |
+
"logits/chosen": 339.40057373046875,
|
| 738 |
+
"logits/rejected": 364.0176696777344,
|
| 739 |
+
"logps/chosen": -1.1170837879180908,
|
| 740 |
+
"logps/rejected": -1.2566817998886108,
|
| 741 |
+
"loss": 20.5383,
|
| 742 |
+
"nll_loss": 1.269911527633667,
|
| 743 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 744 |
+
"rewards/chosen": -0.5585418939590454,
|
| 745 |
+
"rewards/margins": 0.06979899108409882,
|
| 746 |
+
"rewards/rejected": -0.6283408999443054,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.9905213270142181,
|
| 751 |
+
"grad_norm": 28.25,
|
| 752 |
+
"learning_rate": 3.0290434481330746e-06,
|
| 753 |
+
"log_odds_chosen": 0.2612162232398987,
|
| 754 |
+
"log_odds_ratio": -0.672803521156311,
|
| 755 |
+
"logits/chosen": 355.0847473144531,
|
| 756 |
+
"logits/rejected": 357.27752685546875,
|
| 757 |
+
"logps/chosen": -1.1057822704315186,
|
| 758 |
+
"logps/rejected": -1.2985315322875977,
|
| 759 |
+
"loss": 20.8159,
|
| 760 |
+
"nll_loss": 1.2727569341659546,
|
| 761 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 762 |
+
"rewards/chosen": -0.5528911352157593,
|
| 763 |
+
"rewards/margins": 0.09637459367513657,
|
| 764 |
+
"rewards/rejected": -0.6492657661437988,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.037914691943128,
|
| 769 |
+
"grad_norm": 28.875,
|
| 770 |
+
"learning_rate": 2.77715555439007e-06,
|
| 771 |
+
"log_odds_chosen": 0.3204149305820465,
|
| 772 |
+
"log_odds_ratio": -0.6769343018531799,
|
| 773 |
+
"logits/chosen": 358.9181823730469,
|
| 774 |
+
"logits/rejected": 361.39923095703125,
|
| 775 |
+
"logps/chosen": -1.1455085277557373,
|
| 776 |
+
"logps/rejected": -1.3854711055755615,
|
| 777 |
+
"loss": 20.8727,
|
| 778 |
+
"nll_loss": 1.2813211679458618,
|
| 779 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 780 |
+
"rewards/chosen": -0.5727542638778687,
|
| 781 |
+
"rewards/margins": 0.1199813038110733,
|
| 782 |
+
"rewards/rejected": -0.6927355527877808,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.085308056872038,
|
| 787 |
+
"grad_norm": 33.25,
|
| 788 |
+
"learning_rate": 2.532114095079137e-06,
|
| 789 |
+
"log_odds_chosen": 0.41654521226882935,
|
| 790 |
+
"log_odds_ratio": -0.6568773984909058,
|
| 791 |
+
"logits/chosen": 366.5960388183594,
|
| 792 |
+
"logits/rejected": 352.90496826171875,
|
| 793 |
+
"logps/chosen": -1.1304197311401367,
|
| 794 |
+
"logps/rejected": -1.4528872966766357,
|
| 795 |
+
"loss": 20.8523,
|
| 796 |
+
"nll_loss": 1.349096655845642,
|
| 797 |
+
"rewards/accuracies": 0.637499988079071,
|
| 798 |
+
"rewards/chosen": -0.5652098655700684,
|
| 799 |
+
"rewards/margins": 0.1612338125705719,
|
| 800 |
+
"rewards/rejected": -0.7264436483383179,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.132701421800948,
|
| 805 |
+
"grad_norm": 26.75,
|
| 806 |
+
"learning_rate": 2.2946738060017947e-06,
|
| 807 |
+
"log_odds_chosen": 0.2243175506591797,
|
| 808 |
+
"log_odds_ratio": -0.7143479585647583,
|
| 809 |
+
"logits/chosen": 332.3900451660156,
|
| 810 |
+
"logits/rejected": 345.49566650390625,
|
| 811 |
+
"logps/chosen": -1.181205153465271,
|
| 812 |
+
"logps/rejected": -1.3491504192352295,
|
| 813 |
+
"loss": 20.5804,
|
| 814 |
+
"nll_loss": 1.364635705947876,
|
| 815 |
+
"rewards/accuracies": 0.59375,
|
| 816 |
+
"rewards/chosen": -0.5906025767326355,
|
| 817 |
+
"rewards/margins": 0.08397253602743149,
|
| 818 |
+
"rewards/rejected": -0.6745752096176147,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.1800947867298577,
|
| 823 |
+
"grad_norm": 31.75,
|
| 824 |
+
"learning_rate": 2.0655660111037685e-06,
|
| 825 |
+
"log_odds_chosen": 0.3350989818572998,
|
| 826 |
+
"log_odds_ratio": -0.6724013090133667,
|
| 827 |
+
"logits/chosen": 369.9901123046875,
|
| 828 |
+
"logits/rejected": 377.08782958984375,
|
| 829 |
+
"logps/chosen": -1.1567586660385132,
|
| 830 |
+
"logps/rejected": -1.401942253112793,
|
| 831 |
+
"loss": 20.7933,
|
| 832 |
+
"nll_loss": 1.311959981918335,
|
| 833 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 834 |
+
"rewards/chosen": -0.5783793330192566,
|
| 835 |
+
"rewards/margins": 0.12259165942668915,
|
| 836 |
+
"rewards/rejected": -0.7009711265563965,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.227488151658768,
|
| 841 |
+
"grad_norm": 33.25,
|
| 842 |
+
"learning_rate": 1.8454963699730471e-06,
|
| 843 |
+
"log_odds_chosen": 0.2978705167770386,
|
| 844 |
+
"log_odds_ratio": -0.6846665143966675,
|
| 845 |
+
"logits/chosen": 344.5165100097656,
|
| 846 |
+
"logits/rejected": 354.6488342285156,
|
| 847 |
+
"logps/chosen": -1.1545495986938477,
|
| 848 |
+
"logps/rejected": -1.3718467950820923,
|
| 849 |
+
"loss": 20.9976,
|
| 850 |
+
"nll_loss": 1.339095950126648,
|
| 851 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 852 |
+
"rewards/chosen": -0.5772747993469238,
|
| 853 |
+
"rewards/margins": 0.10864856094121933,
|
| 854 |
+
"rewards/rejected": -0.6859233975410461,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2748815165876777,
|
| 859 |
+
"grad_norm": 30.75,
|
| 860 |
+
"learning_rate": 1.6351427043849955e-06,
|
| 861 |
+
"log_odds_chosen": 0.3817955255508423,
|
| 862 |
+
"log_odds_ratio": -0.6393663287162781,
|
| 863 |
+
"logits/chosen": 349.63629150390625,
|
| 864 |
+
"logits/rejected": 357.9341735839844,
|
| 865 |
+
"logps/chosen": -1.0905182361602783,
|
| 866 |
+
"logps/rejected": -1.3862955570220947,
|
| 867 |
+
"loss": 20.6181,
|
| 868 |
+
"nll_loss": 1.3354963064193726,
|
| 869 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 870 |
+
"rewards/chosen": -0.5452591180801392,
|
| 871 |
+
"rewards/margins": 0.14788874983787537,
|
| 872 |
+
"rewards/rejected": -0.6931477785110474,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.322274881516588,
|
| 877 |
+
"grad_norm": 26.875,
|
| 878 |
+
"learning_rate": 1.4351529105888735e-06,
|
| 879 |
+
"log_odds_chosen": 0.31271064281463623,
|
| 880 |
+
"log_odds_ratio": -0.6530163884162903,
|
| 881 |
+
"logits/chosen": 331.0179138183594,
|
| 882 |
+
"logits/rejected": 351.15350341796875,
|
| 883 |
+
"logps/chosen": -1.0642945766448975,
|
| 884 |
+
"logps/rejected": -1.2827186584472656,
|
| 885 |
+
"loss": 20.275,
|
| 886 |
+
"nll_loss": 1.2534034252166748,
|
| 887 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 888 |
+
"rewards/chosen": -0.5321472883224487,
|
| 889 |
+
"rewards/margins": 0.1092119961977005,
|
| 890 |
+
"rewards/rejected": -0.6413593292236328,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3696682464454977,
|
| 895 |
+
"grad_norm": 28.625,
|
| 896 |
+
"learning_rate": 1.2461429637659466e-06,
|
| 897 |
+
"log_odds_chosen": 0.5228301882743835,
|
| 898 |
+
"log_odds_ratio": -0.573235273361206,
|
| 899 |
+
"logits/chosen": 358.1142883300781,
|
| 900 |
+
"logits/rejected": 357.08905029296875,
|
| 901 |
+
"logps/chosen": -1.084540843963623,
|
| 902 |
+
"logps/rejected": -1.4777616262435913,
|
| 903 |
+
"loss": 19.7075,
|
| 904 |
+
"nll_loss": 1.2512117624282837,
|
| 905 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 906 |
+
"rewards/chosen": -0.5422704219818115,
|
| 907 |
+
"rewards/margins": 0.19661036133766174,
|
| 908 |
+
"rewards/rejected": -0.7388808131217957,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.4170616113744074,
|
| 913 |
+
"grad_norm": 26.75,
|
| 914 |
+
"learning_rate": 1.0686950208055486e-06,
|
| 915 |
+
"log_odds_chosen": 0.421553373336792,
|
| 916 |
+
"log_odds_ratio": -0.6212750673294067,
|
| 917 |
+
"logits/chosen": 343.3612976074219,
|
| 918 |
+
"logits/rejected": 357.06671142578125,
|
| 919 |
+
"logps/chosen": -1.0726759433746338,
|
| 920 |
+
"logps/rejected": -1.3855555057525635,
|
| 921 |
+
"loss": 20.2572,
|
| 922 |
+
"nll_loss": 1.2650914192199707,
|
| 923 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 924 |
+
"rewards/chosen": -0.5363379716873169,
|
| 925 |
+
"rewards/margins": 0.15643975138664246,
|
| 926 |
+
"rewards/rejected": -0.6927777528762817,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.4644549763033177,
|
| 931 |
+
"grad_norm": 27.75,
|
| 932 |
+
"learning_rate": 9.033556272426075e-07,
|
| 933 |
+
"log_odds_chosen": 0.4636826515197754,
|
| 934 |
+
"log_odds_ratio": -0.6046215891838074,
|
| 935 |
+
"logits/chosen": 346.56195068359375,
|
| 936 |
+
"logits/rejected": 366.7275390625,
|
| 937 |
+
"logps/chosen": -1.0248494148254395,
|
| 938 |
+
"logps/rejected": -1.3430243730545044,
|
| 939 |
+
"loss": 20.2488,
|
| 940 |
+
"nll_loss": 1.2966502904891968,
|
| 941 |
+
"rewards/accuracies": 0.637499988079071,
|
| 942 |
+
"rewards/chosen": -0.5124247074127197,
|
| 943 |
+
"rewards/margins": 0.1590874344110489,
|
| 944 |
+
"rewards/rejected": -0.6715121865272522,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5118483412322274,
|
| 949 |
+
"grad_norm": 27.875,
|
| 950 |
+
"learning_rate": 7.506340338793111e-07,
|
| 951 |
+
"log_odds_chosen": 0.3095242977142334,
|
| 952 |
+
"log_odds_ratio": -0.6864131689071655,
|
| 953 |
+
"logits/chosen": 350.9102783203125,
|
| 954 |
+
"logits/rejected": 345.47369384765625,
|
| 955 |
+
"logps/chosen": -1.0581636428833008,
|
| 956 |
+
"logps/rejected": -1.2898659706115723,
|
| 957 |
+
"loss": 19.9452,
|
| 958 |
+
"nll_loss": 1.1947921514511108,
|
| 959 |
+
"rewards/accuracies": 0.668749988079071,
|
| 960 |
+
"rewards/chosen": -0.5290818214416504,
|
| 961 |
+
"rewards/margins": 0.11585123836994171,
|
| 962 |
+
"rewards/rejected": -0.6449329853057861,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.5592417061611377,
|
| 967 |
+
"grad_norm": 26.125,
|
| 968 |
+
"learning_rate": 6.110006282757897e-07,
|
| 969 |
+
"log_odds_chosen": 0.3585304915904999,
|
| 970 |
+
"log_odds_ratio": -0.6409192681312561,
|
| 971 |
+
"logits/chosen": 346.3677673339844,
|
| 972 |
+
"logits/rejected": 351.1196594238281,
|
| 973 |
+
"logps/chosen": -1.0988967418670654,
|
| 974 |
+
"logps/rejected": -1.3483343124389648,
|
| 975 |
+
"loss": 20.3609,
|
| 976 |
+
"nll_loss": 1.2682546377182007,
|
| 977 |
+
"rewards/accuracies": 0.643750011920929,
|
| 978 |
+
"rewards/chosen": -0.5494483709335327,
|
| 979 |
+
"rewards/margins": 0.1247188076376915,
|
| 980 |
+
"rewards/rejected": -0.6741671562194824,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.6066350710900474,
|
| 985 |
+
"grad_norm": 32.25,
|
| 986 |
+
"learning_rate": 4.848854859408731e-07,
|
| 987 |
+
"log_odds_chosen": 0.3773255944252014,
|
| 988 |
+
"log_odds_ratio": -0.6349105834960938,
|
| 989 |
+
"logits/chosen": 331.9405212402344,
|
| 990 |
+
"logits/rejected": 360.78240966796875,
|
| 991 |
+
"logps/chosen": -1.0606693029403687,
|
| 992 |
+
"logps/rejected": -1.3333818912506104,
|
| 993 |
+
"loss": 20.7172,
|
| 994 |
+
"nll_loss": 1.2634260654449463,
|
| 995 |
+
"rewards/accuracies": 0.643750011920929,
|
| 996 |
+
"rewards/chosen": -0.5303346514701843,
|
| 997 |
+
"rewards/margins": 0.13635624945163727,
|
| 998 |
+
"rewards/rejected": -0.6666909456253052,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.654028436018957,
|
| 1003 |
+
"grad_norm": 35.5,
|
| 1004 |
+
"learning_rate": 3.7267704568529015e-07,
|
| 1005 |
+
"log_odds_chosen": 0.3883630633354187,
|
| 1006 |
+
"log_odds_ratio": -0.6591242551803589,
|
| 1007 |
+
"logits/chosen": 358.720458984375,
|
| 1008 |
+
"logits/rejected": 362.07733154296875,
|
| 1009 |
+
"logps/chosen": -1.1015071868896484,
|
| 1010 |
+
"logps/rejected": -1.3767952919006348,
|
| 1011 |
+
"loss": 20.1116,
|
| 1012 |
+
"nll_loss": 1.2957426309585571,
|
| 1013 |
+
"rewards/accuracies": 0.581250011920929,
|
| 1014 |
+
"rewards/chosen": -0.5507535934448242,
|
| 1015 |
+
"rewards/margins": 0.13764409720897675,
|
| 1016 |
+
"rewards/rejected": -0.6883976459503174,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.7014218009478674,
|
| 1021 |
+
"grad_norm": 27.125,
|
| 1022 |
+
"learning_rate": 2.7472091321728067e-07,
|
| 1023 |
+
"log_odds_chosen": 0.4095282554626465,
|
| 1024 |
+
"log_odds_ratio": -0.6598828434944153,
|
| 1025 |
+
"logits/chosen": 338.1583557128906,
|
| 1026 |
+
"logits/rejected": 358.77423095703125,
|
| 1027 |
+
"logps/chosen": -1.065172553062439,
|
| 1028 |
+
"logps/rejected": -1.3705942630767822,
|
| 1029 |
+
"loss": 21.1264,
|
| 1030 |
+
"nll_loss": 1.314082384109497,
|
| 1031 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1032 |
+
"rewards/chosen": -0.5325862765312195,
|
| 1033 |
+
"rewards/margins": 0.15271088480949402,
|
| 1034 |
+
"rewards/rejected": -0.6852971315383911,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.748815165876777,
|
| 1039 |
+
"grad_norm": 28.125,
|
| 1040 |
+
"learning_rate": 1.9131879666558385e-07,
|
| 1041 |
+
"log_odds_chosen": 0.3623240888118744,
|
| 1042 |
+
"log_odds_ratio": -0.6543713212013245,
|
| 1043 |
+
"logits/chosen": 346.6483459472656,
|
| 1044 |
+
"logits/rejected": 349.0867614746094,
|
| 1045 |
+
"logps/chosen": -1.1220111846923828,
|
| 1046 |
+
"logps/rejected": -1.3769835233688354,
|
| 1047 |
+
"loss": 20.8844,
|
| 1048 |
+
"nll_loss": 1.2857048511505127,
|
| 1049 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1050 |
+
"rewards/chosen": -0.5610055923461914,
|
| 1051 |
+
"rewards/margins": 0.1274861991405487,
|
| 1052 |
+
"rewards/rejected": -0.6884917616844177,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.7962085308056874,
|
| 1057 |
+
"grad_norm": 29.875,
|
| 1058 |
+
"learning_rate": 1.2272757730841744e-07,
|
| 1059 |
+
"log_odds_chosen": 0.4991677701473236,
|
| 1060 |
+
"log_odds_ratio": -0.5991695523262024,
|
| 1061 |
+
"logits/chosen": 338.85333251953125,
|
| 1062 |
+
"logits/rejected": 345.8002624511719,
|
| 1063 |
+
"logps/chosen": -1.084472417831421,
|
| 1064 |
+
"logps/rejected": -1.4532811641693115,
|
| 1065 |
+
"loss": 20.8087,
|
| 1066 |
+
"nll_loss": 1.2487828731536865,
|
| 1067 |
+
"rewards/accuracies": 0.6875,
|
| 1068 |
+
"rewards/chosen": -0.5422362089157104,
|
| 1069 |
+
"rewards/margins": 0.1844043731689453,
|
| 1070 |
+
"rewards/rejected": -0.7266405820846558,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.843601895734597,
|
| 1075 |
+
"grad_norm": 30.5,
|
| 1076 |
+
"learning_rate": 6.91585183706428e-08,
|
| 1077 |
+
"log_odds_chosen": 0.4421129822731018,
|
| 1078 |
+
"log_odds_ratio": -0.6238647699356079,
|
| 1079 |
+
"logits/chosen": 352.36297607421875,
|
| 1080 |
+
"logits/rejected": 363.74932861328125,
|
| 1081 |
+
"logps/chosen": -1.0848851203918457,
|
| 1082 |
+
"logps/rejected": -1.4058189392089844,
|
| 1083 |
+
"loss": 20.2124,
|
| 1084 |
+
"nll_loss": 1.2451623678207397,
|
| 1085 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1086 |
+
"rewards/chosen": -0.5424425601959229,
|
| 1087 |
+
"rewards/margins": 0.16046686470508575,
|
| 1088 |
+
"rewards/rejected": -0.7029094696044922,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.890995260663507,
|
| 1093 |
+
"grad_norm": 29.0,
|
| 1094 |
+
"learning_rate": 3.077661432604184e-08,
|
| 1095 |
+
"log_odds_chosen": 0.5017856359481812,
|
| 1096 |
+
"log_odds_ratio": -0.5793642401695251,
|
| 1097 |
+
"logits/chosen": 331.6553649902344,
|
| 1098 |
+
"logits/rejected": 360.7071838378906,
|
| 1099 |
+
"logps/chosen": -1.0543787479400635,
|
| 1100 |
+
"logps/rejected": -1.3993524312973022,
|
| 1101 |
+
"loss": 19.7715,
|
| 1102 |
+
"nll_loss": 1.265842318534851,
|
| 1103 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1104 |
+
"rewards/chosen": -0.5271893739700317,
|
| 1105 |
+
"rewards/margins": 0.1724867820739746,
|
| 1106 |
+
"rewards/rejected": -0.6996762156486511,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.938388625592417,
|
| 1111 |
+
"grad_norm": 94.0,
|
| 1112 |
+
"learning_rate": 7.700082708883006e-09,
|
| 1113 |
+
"log_odds_chosen": 0.4347440302371979,
|
| 1114 |
+
"log_odds_ratio": -0.60276859998703,
|
| 1115 |
+
"logits/chosen": 349.830810546875,
|
| 1116 |
+
"logits/rejected": 379.3188781738281,
|
| 1117 |
+
"logps/chosen": -1.1171703338623047,
|
| 1118 |
+
"logps/rejected": -1.4191040992736816,
|
| 1119 |
+
"loss": 20.9093,
|
| 1120 |
+
"nll_loss": 1.3094590902328491,
|
| 1121 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1122 |
+
"rewards/chosen": -0.5585851669311523,
|
| 1123 |
+
"rewards/margins": 0.15096691250801086,
|
| 1124 |
+
"rewards/rejected": -0.7095520496368408,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.985781990521327,
|
| 1129 |
+
"grad_norm": 30.125,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 0.4408469796180725,
|
| 1132 |
+
"log_odds_ratio": -0.6020691990852356,
|
| 1133 |
+
"logits/chosen": 353.28302001953125,
|
| 1134 |
+
"logits/rejected": 368.32501220703125,
|
| 1135 |
+
"logps/chosen": -1.073335886001587,
|
| 1136 |
+
"logps/rejected": -1.3923397064208984,
|
| 1137 |
+
"loss": 20.0639,
|
| 1138 |
+
"nll_loss": 1.2285845279693604,
|
| 1139 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1140 |
+
"rewards/chosen": -0.5366679430007935,
|
| 1141 |
+
"rewards/margins": 0.15950192511081696,
|
| 1142 |
+
"rewards/rejected": -0.6961698532104492,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.985781990521327,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 31.780521017407615,
|
| 1150 |
+
"train_runtime": 7128.5338,
|
| 1151 |
+
"train_samples_per_second": 2.841,
|
| 1152 |
+
"train_steps_per_second": 0.044
|
| 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": 2,
|
| 1174 |
+
"trial_name": null,
|
| 1175 |
+
"trial_params": null
|
| 1176 |
+
}
|