Model save
Browse files- README.md +67 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1176 -0
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-7b
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: gemma-7b-borpo-noisy-5e-5
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-borpo-noisy-5e-5
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="silviasapora/gemma-7b-borpo-noisy-5e-5", 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 |
+
|
| 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": 2.986666666666667,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 254.74021465967573,
|
| 5 |
+
"train_runtime": 6927.757,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.923,
|
| 8 |
+
"train_steps_per_second": 0.045
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.986666666666667,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 254.74021465967573,
|
| 5 |
+
"train_runtime": 6927.757,
|
| 6 |
+
"train_samples": 6750,
|
| 7 |
+
"train_samples_per_second": 2.923,
|
| 8 |
+
"train_steps_per_second": 0.045
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.986666666666667,
|
| 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": 350208.0,
|
| 14 |
+
"learning_rate": 7.8125e-06,
|
| 15 |
+
"log_odds_chosen": 5.121338844299316,
|
| 16 |
+
"log_odds_ratio": -9.85741901397705,
|
| 17 |
+
"logits/chosen": 138.62344360351562,
|
| 18 |
+
"logits/rejected": 152.14088439941406,
|
| 19 |
+
"logps/chosen": -20.805774688720703,
|
| 20 |
+
"logps/rejected": -25.92710304260254,
|
| 21 |
+
"loss": 2767.5898,
|
| 22 |
+
"nll_loss": 8.236187934875488,
|
| 23 |
+
"rewards/accuracies": 0.543749988079071,
|
| 24 |
+
"rewards/chosen": -10.402887344360352,
|
| 25 |
+
"rewards/margins": 2.5606625080108643,
|
| 26 |
+
"rewards/rejected": -12.96355152130127,
|
| 27 |
+
"step": 5
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.09481481481481481,
|
| 31 |
+
"grad_norm": 18560.0,
|
| 32 |
+
"learning_rate": 1.5625e-05,
|
| 33 |
+
"log_odds_chosen": 2.4278221130371094,
|
| 34 |
+
"log_odds_ratio": -7.406900882720947,
|
| 35 |
+
"logits/chosen": 129.0441436767578,
|
| 36 |
+
"logits/rejected": 156.9169158935547,
|
| 37 |
+
"logps/chosen": -15.9468355178833,
|
| 38 |
+
"logps/rejected": -18.37396812438965,
|
| 39 |
+
"loss": 222.5123,
|
| 40 |
+
"nll_loss": 7.290040016174316,
|
| 41 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 42 |
+
"rewards/chosen": -7.97341775894165,
|
| 43 |
+
"rewards/margins": 1.2135658264160156,
|
| 44 |
+
"rewards/rejected": -9.186984062194824,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14222222222222222,
|
| 49 |
+
"grad_norm": 22144.0,
|
| 50 |
+
"learning_rate": 2.34375e-05,
|
| 51 |
+
"log_odds_chosen": 4.718414306640625,
|
| 52 |
+
"log_odds_ratio": -8.311258316040039,
|
| 53 |
+
"logits/chosen": 109.10520935058594,
|
| 54 |
+
"logits/rejected": 134.63148498535156,
|
| 55 |
+
"logps/chosen": -21.552303314208984,
|
| 56 |
+
"logps/rejected": -26.269397735595703,
|
| 57 |
+
"loss": 4043.0027,
|
| 58 |
+
"nll_loss": 8.928205490112305,
|
| 59 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 60 |
+
"rewards/chosen": -10.776151657104492,
|
| 61 |
+
"rewards/margins": 2.358546733856201,
|
| 62 |
+
"rewards/rejected": -13.134698867797852,
|
| 63 |
+
"step": 15
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.18962962962962962,
|
| 67 |
+
"grad_norm": 7392.0,
|
| 68 |
+
"learning_rate": 3.125e-05,
|
| 69 |
+
"log_odds_chosen": -0.21470332145690918,
|
| 70 |
+
"log_odds_ratio": -7.985512733459473,
|
| 71 |
+
"logits/chosen": 124.5482177734375,
|
| 72 |
+
"logits/rejected": 130.01638793945312,
|
| 73 |
+
"logps/chosen": -18.390491485595703,
|
| 74 |
+
"logps/rejected": -18.17677879333496,
|
| 75 |
+
"loss": 2356.9844,
|
| 76 |
+
"nll_loss": 8.458616256713867,
|
| 77 |
+
"rewards/accuracies": 0.543749988079071,
|
| 78 |
+
"rewards/chosen": -9.195245742797852,
|
| 79 |
+
"rewards/margins": -0.10685563087463379,
|
| 80 |
+
"rewards/rejected": -9.08838939666748,
|
| 81 |
+
"step": 20
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.23703703703703705,
|
| 85 |
+
"grad_norm": 15040.0,
|
| 86 |
+
"learning_rate": 3.90625e-05,
|
| 87 |
+
"log_odds_chosen": 4.443707466125488,
|
| 88 |
+
"log_odds_ratio": -7.725132942199707,
|
| 89 |
+
"logits/chosen": 117.22242736816406,
|
| 90 |
+
"logits/rejected": 125.91157531738281,
|
| 91 |
+
"logps/chosen": -18.476272583007812,
|
| 92 |
+
"logps/rejected": -22.920040130615234,
|
| 93 |
+
"loss": 2797.959,
|
| 94 |
+
"nll_loss": 7.461993217468262,
|
| 95 |
+
"rewards/accuracies": 0.53125,
|
| 96 |
+
"rewards/chosen": -9.238136291503906,
|
| 97 |
+
"rewards/margins": 2.2218832969665527,
|
| 98 |
+
"rewards/rejected": -11.460020065307617,
|
| 99 |
+
"step": 25
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28444444444444444,
|
| 103 |
+
"grad_norm": 44800.0,
|
| 104 |
+
"learning_rate": 4.6875e-05,
|
| 105 |
+
"log_odds_chosen": 1.0098240375518799,
|
| 106 |
+
"log_odds_ratio": -6.719524383544922,
|
| 107 |
+
"logits/chosen": 74.41526794433594,
|
| 108 |
+
"logits/rejected": 114.5694808959961,
|
| 109 |
+
"logps/chosen": -17.375003814697266,
|
| 110 |
+
"logps/rejected": -18.381793975830078,
|
| 111 |
+
"loss": 806.7089,
|
| 112 |
+
"nll_loss": 7.751124382019043,
|
| 113 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 114 |
+
"rewards/chosen": -8.687501907348633,
|
| 115 |
+
"rewards/margins": 0.5033950209617615,
|
| 116 |
+
"rewards/rejected": -9.190896987915039,
|
| 117 |
+
"step": 30
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.33185185185185184,
|
| 121 |
+
"grad_norm": 1488.0,
|
| 122 |
+
"learning_rate": 4.998613757348784e-05,
|
| 123 |
+
"log_odds_chosen": -0.7847718000411987,
|
| 124 |
+
"log_odds_ratio": -3.851313352584839,
|
| 125 |
+
"logits/chosen": 95.89997100830078,
|
| 126 |
+
"logits/rejected": 78.72817993164062,
|
| 127 |
+
"logps/chosen": -9.437823295593262,
|
| 128 |
+
"logps/rejected": -8.653680801391602,
|
| 129 |
+
"loss": 540.7871,
|
| 130 |
+
"nll_loss": 5.164363861083984,
|
| 131 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 132 |
+
"rewards/chosen": -4.718911647796631,
|
| 133 |
+
"rewards/margins": -0.3920711278915405,
|
| 134 |
+
"rewards/rejected": -4.326840400695801,
|
| 135 |
+
"step": 35
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37925925925925924,
|
| 139 |
+
"grad_norm": 572.0,
|
| 140 |
+
"learning_rate": 4.990147841143462e-05,
|
| 141 |
+
"log_odds_chosen": 0.18732526898384094,
|
| 142 |
+
"log_odds_ratio": -0.7723637819290161,
|
| 143 |
+
"logits/chosen": 186.49609375,
|
| 144 |
+
"logits/rejected": 200.811767578125,
|
| 145 |
+
"logps/chosen": -1.9941200017929077,
|
| 146 |
+
"logps/rejected": -2.152578830718994,
|
| 147 |
+
"loss": 89.2208,
|
| 148 |
+
"nll_loss": 2.3764045238494873,
|
| 149 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 150 |
+
"rewards/chosen": -0.9970600008964539,
|
| 151 |
+
"rewards/margins": 0.07922936975955963,
|
| 152 |
+
"rewards/rejected": -1.076289415359497,
|
| 153 |
+
"step": 40
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.4266666666666667,
|
| 157 |
+
"grad_norm": 194.0,
|
| 158 |
+
"learning_rate": 4.97401218720448e-05,
|
| 159 |
+
"log_odds_chosen": 0.16486576199531555,
|
| 160 |
+
"log_odds_ratio": -0.8942667245864868,
|
| 161 |
+
"logits/chosen": 235.48745727539062,
|
| 162 |
+
"logits/rejected": 228.4767608642578,
|
| 163 |
+
"logps/chosen": -1.7937572002410889,
|
| 164 |
+
"logps/rejected": -1.9410396814346313,
|
| 165 |
+
"loss": 73.9402,
|
| 166 |
+
"nll_loss": 2.0359139442443848,
|
| 167 |
+
"rewards/accuracies": 0.512499988079071,
|
| 168 |
+
"rewards/chosen": -0.8968786001205444,
|
| 169 |
+
"rewards/margins": 0.07364129275083542,
|
| 170 |
+
"rewards/rejected": -0.9705198407173157,
|
| 171 |
+
"step": 45
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4740740740740741,
|
| 175 |
+
"grad_norm": 108.5,
|
| 176 |
+
"learning_rate": 4.9502564938797946e-05,
|
| 177 |
+
"log_odds_chosen": 0.19230754673480988,
|
| 178 |
+
"log_odds_ratio": -0.7629112601280212,
|
| 179 |
+
"logits/chosen": 250.28610229492188,
|
| 180 |
+
"logits/rejected": 250.9467315673828,
|
| 181 |
+
"logps/chosen": -1.563698172569275,
|
| 182 |
+
"logps/rejected": -1.7324550151824951,
|
| 183 |
+
"loss": 71.6374,
|
| 184 |
+
"nll_loss": 1.9435456991195679,
|
| 185 |
+
"rewards/accuracies": 0.543749988079071,
|
| 186 |
+
"rewards/chosen": -0.7818490862846375,
|
| 187 |
+
"rewards/margins": 0.08437846601009369,
|
| 188 |
+
"rewards/rejected": -0.8662275075912476,
|
| 189 |
+
"step": 50
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5214814814814814,
|
| 193 |
+
"grad_norm": 158.0,
|
| 194 |
+
"learning_rate": 4.918953929490768e-05,
|
| 195 |
+
"log_odds_chosen": 0.12492300570011139,
|
| 196 |
+
"log_odds_ratio": -0.7422401309013367,
|
| 197 |
+
"logits/chosen": 259.849853515625,
|
| 198 |
+
"logits/rejected": 262.5646667480469,
|
| 199 |
+
"logps/chosen": -1.3962795734405518,
|
| 200 |
+
"logps/rejected": -1.4994373321533203,
|
| 201 |
+
"loss": 62.4039,
|
| 202 |
+
"nll_loss": 1.724207878112793,
|
| 203 |
+
"rewards/accuracies": 0.574999988079071,
|
| 204 |
+
"rewards/chosen": -0.6981397867202759,
|
| 205 |
+
"rewards/margins": 0.05157895013689995,
|
| 206 |
+
"rewards/rejected": -0.7497186660766602,
|
| 207 |
+
"step": 55
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.5688888888888889,
|
| 211 |
+
"grad_norm": 96.5,
|
| 212 |
+
"learning_rate": 4.88020090697132e-05,
|
| 213 |
+
"log_odds_chosen": 0.20564976334571838,
|
| 214 |
+
"log_odds_ratio": -0.6903149485588074,
|
| 215 |
+
"logits/chosen": 244.59274291992188,
|
| 216 |
+
"logits/rejected": 249.990478515625,
|
| 217 |
+
"logps/chosen": -1.2423059940338135,
|
| 218 |
+
"logps/rejected": -1.395542025566101,
|
| 219 |
+
"loss": 58.7184,
|
| 220 |
+
"nll_loss": 1.5578796863555908,
|
| 221 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 222 |
+
"rewards/chosen": -0.6211529970169067,
|
| 223 |
+
"rewards/margins": 0.0766180157661438,
|
| 224 |
+
"rewards/rejected": -0.6977710127830505,
|
| 225 |
+
"step": 60
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6162962962962963,
|
| 229 |
+
"grad_norm": 146.0,
|
| 230 |
+
"learning_rate": 4.834116786912897e-05,
|
| 231 |
+
"log_odds_chosen": 0.16068454086780548,
|
| 232 |
+
"log_odds_ratio": -0.7302707433700562,
|
| 233 |
+
"logits/chosen": 241.5931854248047,
|
| 234 |
+
"logits/rejected": 244.501220703125,
|
| 235 |
+
"logps/chosen": -1.264792799949646,
|
| 236 |
+
"logps/rejected": -1.3899643421173096,
|
| 237 |
+
"loss": 56.932,
|
| 238 |
+
"nll_loss": 1.5472662448883057,
|
| 239 |
+
"rewards/accuracies": 0.543749988079071,
|
| 240 |
+
"rewards/chosen": -0.632396399974823,
|
| 241 |
+
"rewards/margins": 0.06258574873209,
|
| 242 |
+
"rewards/rejected": -0.6949821710586548,
|
| 243 |
+
"step": 65
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.6637037037037037,
|
| 247 |
+
"grad_norm": 296.0,
|
| 248 |
+
"learning_rate": 4.7808435099299045e-05,
|
| 249 |
+
"log_odds_chosen": 0.2955527603626251,
|
| 250 |
+
"log_odds_ratio": -0.6667484641075134,
|
| 251 |
+
"logits/chosen": 230.53378295898438,
|
| 252 |
+
"logits/rejected": 237.8424530029297,
|
| 253 |
+
"logps/chosen": -1.1751458644866943,
|
| 254 |
+
"logps/rejected": -1.4215205907821655,
|
| 255 |
+
"loss": 56.7455,
|
| 256 |
+
"nll_loss": 1.5035736560821533,
|
| 257 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 258 |
+
"rewards/chosen": -0.5875729322433472,
|
| 259 |
+
"rewards/margins": 0.12318732589483261,
|
| 260 |
+
"rewards/rejected": -0.7107602953910828,
|
| 261 |
+
"step": 70
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7111111111111111,
|
| 265 |
+
"grad_norm": 96.0,
|
| 266 |
+
"learning_rate": 4.720545159477922e-05,
|
| 267 |
+
"log_odds_chosen": 0.2671827971935272,
|
| 268 |
+
"log_odds_ratio": -0.6653175950050354,
|
| 269 |
+
"logits/chosen": 244.9853057861328,
|
| 270 |
+
"logits/rejected": 243.1300811767578,
|
| 271 |
+
"logps/chosen": -1.1448094844818115,
|
| 272 |
+
"logps/rejected": -1.3471866846084595,
|
| 273 |
+
"loss": 54.6994,
|
| 274 |
+
"nll_loss": 1.492123007774353,
|
| 275 |
+
"rewards/accuracies": 0.637499988079071,
|
| 276 |
+
"rewards/chosen": -0.5724047422409058,
|
| 277 |
+
"rewards/margins": 0.10118861496448517,
|
| 278 |
+
"rewards/rejected": -0.6735933423042297,
|
| 279 |
+
"step": 75
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7585185185185185,
|
| 283 |
+
"grad_norm": 107.0,
|
| 284 |
+
"learning_rate": 4.653407456471222e-05,
|
| 285 |
+
"log_odds_chosen": 0.24170565605163574,
|
| 286 |
+
"log_odds_ratio": -0.7027624845504761,
|
| 287 |
+
"logits/chosen": 238.3944091796875,
|
| 288 |
+
"logits/rejected": 243.826171875,
|
| 289 |
+
"logps/chosen": -1.1621917486190796,
|
| 290 |
+
"logps/rejected": -1.3395326137542725,
|
| 291 |
+
"loss": 52.7796,
|
| 292 |
+
"nll_loss": 1.4673112630844116,
|
| 293 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 294 |
+
"rewards/chosen": -0.5810958743095398,
|
| 295 |
+
"rewards/margins": 0.08867045491933823,
|
| 296 |
+
"rewards/rejected": -0.6697663068771362,
|
| 297 |
+
"step": 80
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.8059259259259259,
|
| 301 |
+
"grad_norm": 67.5,
|
| 302 |
+
"learning_rate": 4.579637187256222e-05,
|
| 303 |
+
"log_odds_chosen": 0.2149166613817215,
|
| 304 |
+
"log_odds_ratio": -0.6902152895927429,
|
| 305 |
+
"logits/chosen": 237.4712677001953,
|
| 306 |
+
"logits/rejected": 251.7366485595703,
|
| 307 |
+
"logps/chosen": -1.0897717475891113,
|
| 308 |
+
"logps/rejected": -1.2465016841888428,
|
| 309 |
+
"loss": 51.8551,
|
| 310 |
+
"nll_loss": 1.3890597820281982,
|
| 311 |
+
"rewards/accuracies": 0.59375,
|
| 312 |
+
"rewards/chosen": -0.5448858737945557,
|
| 313 |
+
"rewards/margins": 0.07836499065160751,
|
| 314 |
+
"rewards/rejected": -0.6232508420944214,
|
| 315 |
+
"step": 85
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8533333333333334,
|
| 319 |
+
"grad_norm": 83.0,
|
| 320 |
+
"learning_rate": 4.499461566702685e-05,
|
| 321 |
+
"log_odds_chosen": 0.2485649585723877,
|
| 322 |
+
"log_odds_ratio": -0.6673235297203064,
|
| 323 |
+
"logits/chosen": 234.73666381835938,
|
| 324 |
+
"logits/rejected": 244.18630981445312,
|
| 325 |
+
"logps/chosen": -1.0668323040008545,
|
| 326 |
+
"logps/rejected": -1.230261206626892,
|
| 327 |
+
"loss": 52.656,
|
| 328 |
+
"nll_loss": 1.4122871160507202,
|
| 329 |
+
"rewards/accuracies": 0.612500011920929,
|
| 330 |
+
"rewards/chosen": -0.5334161520004272,
|
| 331 |
+
"rewards/margins": 0.08171448856592178,
|
| 332 |
+
"rewards/rejected": -0.615130603313446,
|
| 333 |
+
"step": 90
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.9007407407407407,
|
| 337 |
+
"grad_norm": 198.0,
|
| 338 |
+
"learning_rate": 4.413127538374411e-05,
|
| 339 |
+
"log_odds_chosen": 0.36127719283103943,
|
| 340 |
+
"log_odds_ratio": -0.5996707677841187,
|
| 341 |
+
"logits/chosen": 227.44729614257812,
|
| 342 |
+
"logits/rejected": 236.6618194580078,
|
| 343 |
+
"logps/chosen": -0.9856353998184204,
|
| 344 |
+
"logps/rejected": -1.2236969470977783,
|
| 345 |
+
"loss": 50.7134,
|
| 346 |
+
"nll_loss": 1.3108433485031128,
|
| 347 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 348 |
+
"rewards/chosen": -0.4928176999092102,
|
| 349 |
+
"rewards/margins": 0.11903069913387299,
|
| 350 |
+
"rewards/rejected": -0.6118484735488892,
|
| 351 |
+
"step": 95
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9481481481481482,
|
| 355 |
+
"grad_norm": 71.0,
|
| 356 |
+
"learning_rate": 4.320901013934887e-05,
|
| 357 |
+
"log_odds_chosen": 0.09772248566150665,
|
| 358 |
+
"log_odds_ratio": -0.7446527481079102,
|
| 359 |
+
"logits/chosen": 239.33163452148438,
|
| 360 |
+
"logits/rejected": 244.9424591064453,
|
| 361 |
+
"logps/chosen": -1.1100224256515503,
|
| 362 |
+
"logps/rejected": -1.1970586776733398,
|
| 363 |
+
"loss": 51.5177,
|
| 364 |
+
"nll_loss": 1.4378069639205933,
|
| 365 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 366 |
+
"rewards/chosen": -0.5550112128257751,
|
| 367 |
+
"rewards/margins": 0.04351811483502388,
|
| 368 |
+
"rewards/rejected": -0.5985293388366699,
|
| 369 |
+
"step": 100
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9955555555555555,
|
| 373 |
+
"grad_norm": 64.0,
|
| 374 |
+
"learning_rate": 4.223066054130568e-05,
|
| 375 |
+
"log_odds_chosen": 0.27767232060432434,
|
| 376 |
+
"log_odds_ratio": -0.6391838788986206,
|
| 377 |
+
"logits/chosen": 239.01522827148438,
|
| 378 |
+
"logits/rejected": 248.607421875,
|
| 379 |
+
"logps/chosen": -1.0211066007614136,
|
| 380 |
+
"logps/rejected": -1.21805739402771,
|
| 381 |
+
"loss": 49.8911,
|
| 382 |
+
"nll_loss": 1.339125633239746,
|
| 383 |
+
"rewards/accuracies": 0.625,
|
| 384 |
+
"rewards/chosen": -0.5105533003807068,
|
| 385 |
+
"rewards/margins": 0.09847544133663177,
|
| 386 |
+
"rewards/rejected": -0.609028697013855,
|
| 387 |
+
"step": 105
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.0429629629629629,
|
| 391 |
+
"grad_norm": 52.0,
|
| 392 |
+
"learning_rate": 4.1199239938743797e-05,
|
| 393 |
+
"log_odds_chosen": 0.2824048697948456,
|
| 394 |
+
"log_odds_ratio": -0.6530550718307495,
|
| 395 |
+
"logits/chosen": 236.0783233642578,
|
| 396 |
+
"logits/rejected": 240.7754364013672,
|
| 397 |
+
"logps/chosen": -0.9296471476554871,
|
| 398 |
+
"logps/rejected": -1.1144460439682007,
|
| 399 |
+
"loss": 46.0534,
|
| 400 |
+
"nll_loss": 1.2121281623840332,
|
| 401 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 402 |
+
"rewards/chosen": -0.46482357382774353,
|
| 403 |
+
"rewards/margins": 0.09239944070577621,
|
| 404 |
+
"rewards/rejected": -0.5572230219841003,
|
| 405 |
+
"step": 110
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.0903703703703704,
|
| 409 |
+
"grad_norm": 90.0,
|
| 410 |
+
"learning_rate": 4.0117925141242174e-05,
|
| 411 |
+
"log_odds_chosen": 0.4364786744117737,
|
| 412 |
+
"log_odds_ratio": -0.5888470411300659,
|
| 413 |
+
"logits/chosen": 231.71511840820312,
|
| 414 |
+
"logits/rejected": 239.5157928466797,
|
| 415 |
+
"logps/chosen": -0.8990312814712524,
|
| 416 |
+
"logps/rejected": -1.1864837408065796,
|
| 417 |
+
"loss": 45.2371,
|
| 418 |
+
"nll_loss": 1.1880322694778442,
|
| 419 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 420 |
+
"rewards/chosen": -0.4495156407356262,
|
| 421 |
+
"rewards/margins": 0.14372624456882477,
|
| 422 |
+
"rewards/rejected": -0.5932418704032898,
|
| 423 |
+
"step": 115
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.1377777777777778,
|
| 427 |
+
"grad_norm": 52.75,
|
| 428 |
+
"learning_rate": 3.899004663415084e-05,
|
| 429 |
+
"log_odds_chosen": 0.433866024017334,
|
| 430 |
+
"log_odds_ratio": -0.5812664031982422,
|
| 431 |
+
"logits/chosen": 228.9931640625,
|
| 432 |
+
"logits/rejected": 234.70193481445312,
|
| 433 |
+
"logps/chosen": -0.8852219581604004,
|
| 434 |
+
"logps/rejected": -1.1457195281982422,
|
| 435 |
+
"loss": 45.4878,
|
| 436 |
+
"nll_loss": 1.1959518194198608,
|
| 437 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 438 |
+
"rewards/chosen": -0.4426109790802002,
|
| 439 |
+
"rewards/margins": 0.13024874031543732,
|
| 440 |
+
"rewards/rejected": -0.5728597640991211,
|
| 441 |
+
"step": 120
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.1851851851851851,
|
| 445 |
+
"grad_norm": 59.75,
|
| 446 |
+
"learning_rate": 3.781907832058587e-05,
|
| 447 |
+
"log_odds_chosen": 0.4279605448246002,
|
| 448 |
+
"log_odds_ratio": -0.6149727702140808,
|
| 449 |
+
"logits/chosen": 228.2986602783203,
|
| 450 |
+
"logits/rejected": 231.1459503173828,
|
| 451 |
+
"logps/chosen": -0.9138356447219849,
|
| 452 |
+
"logps/rejected": -1.196605920791626,
|
| 453 |
+
"loss": 42.8771,
|
| 454 |
+
"nll_loss": 1.1416022777557373,
|
| 455 |
+
"rewards/accuracies": 0.65625,
|
| 456 |
+
"rewards/chosen": -0.45691782236099243,
|
| 457 |
+
"rewards/margins": 0.14138510823249817,
|
| 458 |
+
"rewards/rejected": -0.598302960395813,
|
| 459 |
+
"step": 125
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2325925925925927,
|
| 463 |
+
"grad_norm": 116.0,
|
| 464 |
+
"learning_rate": 3.660862682169282e-05,
|
| 465 |
+
"log_odds_chosen": 0.29699453711509705,
|
| 466 |
+
"log_odds_ratio": -0.6551839709281921,
|
| 467 |
+
"logits/chosen": 229.7410430908203,
|
| 468 |
+
"logits/rejected": 233.87667846679688,
|
| 469 |
+
"logps/chosen": -0.9139402508735657,
|
| 470 |
+
"logps/rejected": -1.0807549953460693,
|
| 471 |
+
"loss": 43.7064,
|
| 472 |
+
"nll_loss": 1.2109730243682861,
|
| 473 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 474 |
+
"rewards/chosen": -0.45697012543678284,
|
| 475 |
+
"rewards/margins": 0.08340738713741302,
|
| 476 |
+
"rewards/rejected": -0.5403774976730347,
|
| 477 |
+
"step": 130
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.28,
|
| 481 |
+
"grad_norm": 107.5,
|
| 482 |
+
"learning_rate": 3.5362420368134356e-05,
|
| 483 |
+
"log_odds_chosen": 0.39430850744247437,
|
| 484 |
+
"log_odds_ratio": -0.6013480424880981,
|
| 485 |
+
"logits/chosen": 232.01220703125,
|
| 486 |
+
"logits/rejected": 235.47061157226562,
|
| 487 |
+
"logps/chosen": -0.8469200134277344,
|
| 488 |
+
"logps/rejected": -1.0819423198699951,
|
| 489 |
+
"loss": 44.477,
|
| 490 |
+
"nll_loss": 1.1646451950073242,
|
| 491 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 492 |
+
"rewards/chosen": -0.4234600067138672,
|
| 493 |
+
"rewards/margins": 0.11751119047403336,
|
| 494 |
+
"rewards/rejected": -0.5409711599349976,
|
| 495 |
+
"step": 135
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.3274074074074074,
|
| 499 |
+
"grad_norm": 153.0,
|
| 500 |
+
"learning_rate": 3.408429731701635e-05,
|
| 501 |
+
"log_odds_chosen": 0.3346176743507385,
|
| 502 |
+
"log_odds_ratio": -0.6571810245513916,
|
| 503 |
+
"logits/chosen": 231.61083984375,
|
| 504 |
+
"logits/rejected": 233.050537109375,
|
| 505 |
+
"logps/chosen": -0.9530075192451477,
|
| 506 |
+
"logps/rejected": -1.1791385412216187,
|
| 507 |
+
"loss": 45.6471,
|
| 508 |
+
"nll_loss": 1.2547054290771484,
|
| 509 |
+
"rewards/accuracies": 0.625,
|
| 510 |
+
"rewards/chosen": -0.47650375962257385,
|
| 511 |
+
"rewards/margins": 0.11306555569171906,
|
| 512 |
+
"rewards/rejected": -0.5895692706108093,
|
| 513 |
+
"step": 140
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.374814814814815,
|
| 517 |
+
"grad_norm": 77.0,
|
| 518 |
+
"learning_rate": 3.2778194329621104e-05,
|
| 519 |
+
"log_odds_chosen": 0.3642648160457611,
|
| 520 |
+
"log_odds_ratio": -0.61054927110672,
|
| 521 |
+
"logits/chosen": 226.9085235595703,
|
| 522 |
+
"logits/rejected": 232.42965698242188,
|
| 523 |
+
"logps/chosen": -0.924142062664032,
|
| 524 |
+
"logps/rejected": -1.1682136058807373,
|
| 525 |
+
"loss": 44.7382,
|
| 526 |
+
"nll_loss": 1.1921000480651855,
|
| 527 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 528 |
+
"rewards/chosen": -0.462071031332016,
|
| 529 |
+
"rewards/margins": 0.12203577905893326,
|
| 530 |
+
"rewards/rejected": -0.5841068029403687,
|
| 531 |
+
"step": 145
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.4222222222222223,
|
| 535 |
+
"grad_norm": 55.25,
|
| 536 |
+
"learning_rate": 3.144813424636031e-05,
|
| 537 |
+
"log_odds_chosen": 0.282584011554718,
|
| 538 |
+
"log_odds_ratio": -0.6607434153556824,
|
| 539 |
+
"logits/chosen": 225.26840209960938,
|
| 540 |
+
"logits/rejected": 226.21768188476562,
|
| 541 |
+
"logps/chosen": -0.8751763105392456,
|
| 542 |
+
"logps/rejected": -1.040479302406311,
|
| 543 |
+
"loss": 42.9438,
|
| 544 |
+
"nll_loss": 1.1665369272232056,
|
| 545 |
+
"rewards/accuracies": 0.59375,
|
| 546 |
+
"rewards/chosen": -0.4375881552696228,
|
| 547 |
+
"rewards/margins": 0.08265147358179092,
|
| 548 |
+
"rewards/rejected": -0.5202396512031555,
|
| 549 |
+
"step": 150
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.4696296296296296,
|
| 553 |
+
"grad_norm": 86.5,
|
| 554 |
+
"learning_rate": 3.0098213696293542e-05,
|
| 555 |
+
"log_odds_chosen": 0.28907060623168945,
|
| 556 |
+
"log_odds_ratio": -0.6529395580291748,
|
| 557 |
+
"logits/chosen": 224.1173553466797,
|
| 558 |
+
"logits/rejected": 232.13858032226562,
|
| 559 |
+
"logps/chosen": -0.8760883212089539,
|
| 560 |
+
"logps/rejected": -1.0371713638305664,
|
| 561 |
+
"loss": 44.3679,
|
| 562 |
+
"nll_loss": 1.1804320812225342,
|
| 563 |
+
"rewards/accuracies": 0.643750011920929,
|
| 564 |
+
"rewards/chosen": -0.43804416060447693,
|
| 565 |
+
"rewards/margins": 0.08054153621196747,
|
| 566 |
+
"rewards/rejected": -0.5185856819152832,
|
| 567 |
+
"step": 155
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.5170370370370372,
|
| 571 |
+
"grad_norm": 68.0,
|
| 572 |
+
"learning_rate": 2.8732590479375165e-05,
|
| 573 |
+
"log_odds_chosen": 0.32775264978408813,
|
| 574 |
+
"log_odds_ratio": -0.612945020198822,
|
| 575 |
+
"logits/chosen": 230.9765625,
|
| 576 |
+
"logits/rejected": 230.38125610351562,
|
| 577 |
+
"logps/chosen": -0.951357364654541,
|
| 578 |
+
"logps/rejected": -1.1643283367156982,
|
| 579 |
+
"loss": 45.2404,
|
| 580 |
+
"nll_loss": 1.2451446056365967,
|
| 581 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 582 |
+
"rewards/chosen": -0.4756786823272705,
|
| 583 |
+
"rewards/margins": 0.10648546367883682,
|
| 584 |
+
"rewards/rejected": -0.5821641683578491,
|
| 585 |
+
"step": 160
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.5644444444444443,
|
| 589 |
+
"grad_norm": 38.75,
|
| 590 |
+
"learning_rate": 2.7355470760292956e-05,
|
| 591 |
+
"log_odds_chosen": 0.4192623198032379,
|
| 592 |
+
"log_odds_ratio": -0.5958751440048218,
|
| 593 |
+
"logits/chosen": 225.8223876953125,
|
| 594 |
+
"logits/rejected": 225.8138427734375,
|
| 595 |
+
"logps/chosen": -0.9021142721176147,
|
| 596 |
+
"logps/rejected": -1.1618568897247314,
|
| 597 |
+
"loss": 43.8924,
|
| 598 |
+
"nll_loss": 1.1659859418869019,
|
| 599 |
+
"rewards/accuracies": 0.675000011920929,
|
| 600 |
+
"rewards/chosen": -0.4510571360588074,
|
| 601 |
+
"rewards/margins": 0.12987132370471954,
|
| 602 |
+
"rewards/rejected": -0.5809284448623657,
|
| 603 |
+
"step": 165
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.6118518518518519,
|
| 607 |
+
"grad_norm": 82.5,
|
| 608 |
+
"learning_rate": 2.597109611334169e-05,
|
| 609 |
+
"log_odds_chosen": 0.43345552682876587,
|
| 610 |
+
"log_odds_ratio": -0.5985883474349976,
|
| 611 |
+
"logits/chosen": 222.00634765625,
|
| 612 |
+
"logits/rejected": 224.83993530273438,
|
| 613 |
+
"logps/chosen": -0.878454864025116,
|
| 614 |
+
"logps/rejected": -1.1517555713653564,
|
| 615 |
+
"loss": 44.3274,
|
| 616 |
+
"nll_loss": 1.1605513095855713,
|
| 617 |
+
"rewards/accuracies": 0.6875,
|
| 618 |
+
"rewards/chosen": -0.439227432012558,
|
| 619 |
+
"rewards/margins": 0.13665039837360382,
|
| 620 |
+
"rewards/rejected": -0.5758777856826782,
|
| 621 |
+
"step": 170
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.6592592592592592,
|
| 625 |
+
"grad_norm": 84.5,
|
| 626 |
+
"learning_rate": 2.458373045823404e-05,
|
| 627 |
+
"log_odds_chosen": 0.30794280767440796,
|
| 628 |
+
"log_odds_ratio": -0.6498476266860962,
|
| 629 |
+
"logits/chosen": 230.5714874267578,
|
| 630 |
+
"logits/rejected": 233.70053100585938,
|
| 631 |
+
"logps/chosen": -0.8978121876716614,
|
| 632 |
+
"logps/rejected": -1.0880495309829712,
|
| 633 |
+
"loss": 43.0845,
|
| 634 |
+
"nll_loss": 1.1957765817642212,
|
| 635 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 636 |
+
"rewards/chosen": -0.4489060938358307,
|
| 637 |
+
"rewards/margins": 0.0951187014579773,
|
| 638 |
+
"rewards/rejected": -0.5440247654914856,
|
| 639 |
+
"step": 175
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7066666666666666,
|
| 643 |
+
"grad_norm": 90.0,
|
| 644 |
+
"learning_rate": 2.3197646927086697e-05,
|
| 645 |
+
"log_odds_chosen": 0.13764269649982452,
|
| 646 |
+
"log_odds_ratio": -0.7279932498931885,
|
| 647 |
+
"logits/chosen": 232.72415161132812,
|
| 648 |
+
"logits/rejected": 232.6797637939453,
|
| 649 |
+
"logps/chosen": -0.9545130729675293,
|
| 650 |
+
"logps/rejected": -1.035469651222229,
|
| 651 |
+
"loss": 44.9809,
|
| 652 |
+
"nll_loss": 1.2627184391021729,
|
| 653 |
+
"rewards/accuracies": 0.5625,
|
| 654 |
+
"rewards/chosen": -0.47725653648376465,
|
| 655 |
+
"rewards/margins": 0.04047833010554314,
|
| 656 |
+
"rewards/rejected": -0.5177348256111145,
|
| 657 |
+
"step": 180
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.7540740740740741,
|
| 661 |
+
"grad_norm": 34.75,
|
| 662 |
+
"learning_rate": 2.1817114703032176e-05,
|
| 663 |
+
"log_odds_chosen": 0.345479816198349,
|
| 664 |
+
"log_odds_ratio": -0.6043921709060669,
|
| 665 |
+
"logits/chosen": 228.2840118408203,
|
| 666 |
+
"logits/rejected": 230.993408203125,
|
| 667 |
+
"logps/chosen": -0.8777548670768738,
|
| 668 |
+
"logps/rejected": -1.0991740226745605,
|
| 669 |
+
"loss": 43.1118,
|
| 670 |
+
"nll_loss": 1.1345906257629395,
|
| 671 |
+
"rewards/accuracies": 0.65625,
|
| 672 |
+
"rewards/chosen": -0.4388774335384369,
|
| 673 |
+
"rewards/margins": 0.11070958524942398,
|
| 674 |
+
"rewards/rejected": -0.5495870113372803,
|
| 675 |
+
"step": 185
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.8014814814814815,
|
| 679 |
+
"grad_norm": 56.0,
|
| 680 |
+
"learning_rate": 2.0446385870993467e-05,
|
| 681 |
+
"log_odds_chosen": 0.4094289243221283,
|
| 682 |
+
"log_odds_ratio": -0.6055964231491089,
|
| 683 |
+
"logits/chosen": 224.13839721679688,
|
| 684 |
+
"logits/rejected": 230.32461547851562,
|
| 685 |
+
"logps/chosen": -0.9269938468933105,
|
| 686 |
+
"logps/rejected": -1.1887009143829346,
|
| 687 |
+
"loss": 44.9884,
|
| 688 |
+
"nll_loss": 1.2038931846618652,
|
| 689 |
+
"rewards/accuracies": 0.699999988079071,
|
| 690 |
+
"rewards/chosen": -0.4634969234466553,
|
| 691 |
+
"rewards/margins": 0.1308535933494568,
|
| 692 |
+
"rewards/rejected": -0.5943504571914673,
|
| 693 |
+
"step": 190
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.8488888888888888,
|
| 697 |
+
"grad_norm": 54.75,
|
| 698 |
+
"learning_rate": 1.9089682321121834e-05,
|
| 699 |
+
"log_odds_chosen": 0.3977203965187073,
|
| 700 |
+
"log_odds_ratio": -0.6004412174224854,
|
| 701 |
+
"logits/chosen": 223.8681640625,
|
| 702 |
+
"logits/rejected": 231.8001251220703,
|
| 703 |
+
"logps/chosen": -0.8826181292533875,
|
| 704 |
+
"logps/rejected": -1.1249583959579468,
|
| 705 |
+
"loss": 43.2424,
|
| 706 |
+
"nll_loss": 1.154944658279419,
|
| 707 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 708 |
+
"rewards/chosen": -0.4413090646266937,
|
| 709 |
+
"rewards/margins": 0.12117014825344086,
|
| 710 |
+
"rewards/rejected": -0.5624791979789734,
|
| 711 |
+
"step": 195
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.8962962962962964,
|
| 715 |
+
"grad_norm": 39.25,
|
| 716 |
+
"learning_rate": 1.775118274523545e-05,
|
| 717 |
+
"log_odds_chosen": 0.29216670989990234,
|
| 718 |
+
"log_odds_ratio": -0.6725496053695679,
|
| 719 |
+
"logits/chosen": 224.38119506835938,
|
| 720 |
+
"logits/rejected": 227.98068237304688,
|
| 721 |
+
"logps/chosen": -0.9147941470146179,
|
| 722 |
+
"logps/rejected": -1.0938390493392944,
|
| 723 |
+
"loss": 45.3255,
|
| 724 |
+
"nll_loss": 1.1992260217666626,
|
| 725 |
+
"rewards/accuracies": 0.612500011920929,
|
| 726 |
+
"rewards/chosen": -0.45739707350730896,
|
| 727 |
+
"rewards/margins": 0.08952243626117706,
|
| 728 |
+
"rewards/rejected": -0.5469195246696472,
|
| 729 |
+
"step": 200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9437037037037037,
|
| 733 |
+
"grad_norm": 40.0,
|
| 734 |
+
"learning_rate": 1.643500976631037e-05,
|
| 735 |
+
"log_odds_chosen": 0.339089572429657,
|
| 736 |
+
"log_odds_ratio": -0.6290854215621948,
|
| 737 |
+
"logits/chosen": 223.801513671875,
|
| 738 |
+
"logits/rejected": 232.53506469726562,
|
| 739 |
+
"logps/chosen": -0.8788326382637024,
|
| 740 |
+
"logps/rejected": -1.1053264141082764,
|
| 741 |
+
"loss": 41.223,
|
| 742 |
+
"nll_loss": 1.1145111322402954,
|
| 743 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 744 |
+
"rewards/chosen": -0.4394163191318512,
|
| 745 |
+
"rewards/margins": 0.11324682086706161,
|
| 746 |
+
"rewards/rejected": -0.5526632070541382,
|
| 747 |
+
"step": 205
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.991111111111111,
|
| 751 |
+
"grad_norm": 57.25,
|
| 752 |
+
"learning_rate": 1.514521724066537e-05,
|
| 753 |
+
"log_odds_chosen": 0.2665654718875885,
|
| 754 |
+
"log_odds_ratio": -0.656215250492096,
|
| 755 |
+
"logits/chosen": 226.04345703125,
|
| 756 |
+
"logits/rejected": 222.84518432617188,
|
| 757 |
+
"logps/chosen": -0.8619743585586548,
|
| 758 |
+
"logps/rejected": -1.0236845016479492,
|
| 759 |
+
"loss": 41.679,
|
| 760 |
+
"nll_loss": 1.1403412818908691,
|
| 761 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 762 |
+
"rewards/chosen": -0.4309871792793274,
|
| 763 |
+
"rewards/margins": 0.08085508644580841,
|
| 764 |
+
"rewards/rejected": -0.5118422508239746,
|
| 765 |
+
"step": 210
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.0385185185185186,
|
| 769 |
+
"grad_norm": 41.0,
|
| 770 |
+
"learning_rate": 1.3885777771950348e-05,
|
| 771 |
+
"log_odds_chosen": 0.4965452253818512,
|
| 772 |
+
"log_odds_ratio": -0.5693584084510803,
|
| 773 |
+
"logits/chosen": 222.06100463867188,
|
| 774 |
+
"logits/rejected": 227.81369018554688,
|
| 775 |
+
"logps/chosen": -0.7932162284851074,
|
| 776 |
+
"logps/rejected": -1.053929328918457,
|
| 777 |
+
"loss": 38.5686,
|
| 778 |
+
"nll_loss": 1.0264673233032227,
|
| 779 |
+
"rewards/accuracies": 0.71875,
|
| 780 |
+
"rewards/chosen": -0.3966081142425537,
|
| 781 |
+
"rewards/margins": 0.13035652041435242,
|
| 782 |
+
"rewards/rejected": -0.5269646644592285,
|
| 783 |
+
"step": 215
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.0859259259259257,
|
| 787 |
+
"grad_norm": 61.0,
|
| 788 |
+
"learning_rate": 1.2660570475395683e-05,
|
| 789 |
+
"log_odds_chosen": 0.6846104860305786,
|
| 790 |
+
"log_odds_ratio": -0.5175246596336365,
|
| 791 |
+
"logits/chosen": 214.78866577148438,
|
| 792 |
+
"logits/rejected": 215.9294891357422,
|
| 793 |
+
"logps/chosen": -0.695962131023407,
|
| 794 |
+
"logps/rejected": -1.07391357421875,
|
| 795 |
+
"loss": 36.2322,
|
| 796 |
+
"nll_loss": 0.9726701974868774,
|
| 797 |
+
"rewards/accuracies": 0.75,
|
| 798 |
+
"rewards/chosen": -0.3479810655117035,
|
| 799 |
+
"rewards/margins": 0.1889757215976715,
|
| 800 |
+
"rewards/rejected": -0.536956787109375,
|
| 801 |
+
"step": 220
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.1333333333333333,
|
| 805 |
+
"grad_norm": 41.5,
|
| 806 |
+
"learning_rate": 1.1473369030008974e-05,
|
| 807 |
+
"log_odds_chosen": 0.4971061646938324,
|
| 808 |
+
"log_odds_ratio": -0.5805861353874207,
|
| 809 |
+
"logits/chosen": 220.0359649658203,
|
| 810 |
+
"logits/rejected": 221.62265014648438,
|
| 811 |
+
"logps/chosen": -0.6966139674186707,
|
| 812 |
+
"logps/rejected": -0.9382969737052917,
|
| 813 |
+
"loss": 36.4938,
|
| 814 |
+
"nll_loss": 0.9666304588317871,
|
| 815 |
+
"rewards/accuracies": 0.737500011920929,
|
| 816 |
+
"rewards/chosen": -0.3483069837093353,
|
| 817 |
+
"rewards/margins": 0.12084150314331055,
|
| 818 |
+
"rewards/rejected": -0.4691484868526459,
|
| 819 |
+
"step": 225
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.180740740740741,
|
| 823 |
+
"grad_norm": 46.5,
|
| 824 |
+
"learning_rate": 1.0327830055518842e-05,
|
| 825 |
+
"log_odds_chosen": 0.578362762928009,
|
| 826 |
+
"log_odds_ratio": -0.5245636105537415,
|
| 827 |
+
"logits/chosen": 215.7959747314453,
|
| 828 |
+
"logits/rejected": 220.932861328125,
|
| 829 |
+
"logps/chosen": -0.6989654302597046,
|
| 830 |
+
"logps/rejected": -1.0136926174163818,
|
| 831 |
+
"loss": 35.8101,
|
| 832 |
+
"nll_loss": 0.9475434422492981,
|
| 833 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 834 |
+
"rewards/chosen": -0.3494827151298523,
|
| 835 |
+
"rewards/margins": 0.1573636829853058,
|
| 836 |
+
"rewards/rejected": -0.5068463087081909,
|
| 837 |
+
"step": 230
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.228148148148148,
|
| 841 |
+
"grad_norm": 48.0,
|
| 842 |
+
"learning_rate": 9.227481849865235e-06,
|
| 843 |
+
"log_odds_chosen": 0.46400395035743713,
|
| 844 |
+
"log_odds_ratio": -0.5828996896743774,
|
| 845 |
+
"logits/chosen": 215.46646118164062,
|
| 846 |
+
"logits/rejected": 216.53781127929688,
|
| 847 |
+
"logps/chosen": -0.7348538041114807,
|
| 848 |
+
"logps/rejected": -0.9607747197151184,
|
| 849 |
+
"loss": 36.3354,
|
| 850 |
+
"nll_loss": 0.9704949259757996,
|
| 851 |
+
"rewards/accuracies": 0.675000011920929,
|
| 852 |
+
"rewards/chosen": -0.36742690205574036,
|
| 853 |
+
"rewards/margins": 0.11296045780181885,
|
| 854 |
+
"rewards/rejected": -0.4803873598575592,
|
| 855 |
+
"step": 235
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.2755555555555556,
|
| 859 |
+
"grad_norm": 52.0,
|
| 860 |
+
"learning_rate": 8.175713521924978e-06,
|
| 861 |
+
"log_odds_chosen": 0.608235776424408,
|
| 862 |
+
"log_odds_ratio": -0.5281478762626648,
|
| 863 |
+
"logits/chosen": 214.62643432617188,
|
| 864 |
+
"logits/rejected": 220.86477661132812,
|
| 865 |
+
"logps/chosen": -0.6918534636497498,
|
| 866 |
+
"logps/rejected": -0.998008131980896,
|
| 867 |
+
"loss": 36.3286,
|
| 868 |
+
"nll_loss": 0.984667956829071,
|
| 869 |
+
"rewards/accuracies": 0.706250011920929,
|
| 870 |
+
"rewards/chosen": -0.3459267318248749,
|
| 871 |
+
"rewards/margins": 0.15307730436325073,
|
| 872 |
+
"rewards/rejected": -0.499004065990448,
|
| 873 |
+
"step": 240
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.322962962962963,
|
| 877 |
+
"grad_norm": 56.0,
|
| 878 |
+
"learning_rate": 7.1757645529443665e-06,
|
| 879 |
+
"log_odds_chosen": 0.579681396484375,
|
| 880 |
+
"log_odds_ratio": -0.5280523896217346,
|
| 881 |
+
"logits/chosen": 211.4954071044922,
|
| 882 |
+
"logits/rejected": 222.38168334960938,
|
| 883 |
+
"logps/chosen": -0.7150880098342896,
|
| 884 |
+
"logps/rejected": -1.0089267492294312,
|
| 885 |
+
"loss": 35.879,
|
| 886 |
+
"nll_loss": 0.9763582944869995,
|
| 887 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 888 |
+
"rewards/chosen": -0.3575440049171448,
|
| 889 |
+
"rewards/margins": 0.1469193994998932,
|
| 890 |
+
"rewards/rejected": -0.5044633746147156,
|
| 891 |
+
"step": 245
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.3703703703703702,
|
| 895 |
+
"grad_norm": 43.25,
|
| 896 |
+
"learning_rate": 6.230714818829733e-06,
|
| 897 |
+
"log_odds_chosen": 0.6855599880218506,
|
| 898 |
+
"log_odds_ratio": -0.5006572604179382,
|
| 899 |
+
"logits/chosen": 212.50741577148438,
|
| 900 |
+
"logits/rejected": 216.546630859375,
|
| 901 |
+
"logps/chosen": -0.6900970339775085,
|
| 902 |
+
"logps/rejected": -1.0552127361297607,
|
| 903 |
+
"loss": 35.5148,
|
| 904 |
+
"nll_loss": 0.922285258769989,
|
| 905 |
+
"rewards/accuracies": 0.762499988079071,
|
| 906 |
+
"rewards/chosen": -0.3450485169887543,
|
| 907 |
+
"rewards/margins": 0.1825578510761261,
|
| 908 |
+
"rewards/rejected": -0.5276063680648804,
|
| 909 |
+
"step": 250
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.417777777777778,
|
| 913 |
+
"grad_norm": 39.25,
|
| 914 |
+
"learning_rate": 5.343475104027743e-06,
|
| 915 |
+
"log_odds_chosen": 0.6516860723495483,
|
| 916 |
+
"log_odds_ratio": -0.5264917612075806,
|
| 917 |
+
"logits/chosen": 208.18154907226562,
|
| 918 |
+
"logits/rejected": 220.366943359375,
|
| 919 |
+
"logps/chosen": -0.6484566926956177,
|
| 920 |
+
"logps/rejected": -0.9873029589653015,
|
| 921 |
+
"loss": 34.7539,
|
| 922 |
+
"nll_loss": 0.9242755770683289,
|
| 923 |
+
"rewards/accuracies": 0.75,
|
| 924 |
+
"rewards/chosen": -0.32422834634780884,
|
| 925 |
+
"rewards/margins": 0.16942311823368073,
|
| 926 |
+
"rewards/rejected": -0.49365147948265076,
|
| 927 |
+
"step": 255
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.4651851851851854,
|
| 931 |
+
"grad_norm": 43.75,
|
| 932 |
+
"learning_rate": 4.516778136213037e-06,
|
| 933 |
+
"log_odds_chosen": 0.649712085723877,
|
| 934 |
+
"log_odds_ratio": -0.532970130443573,
|
| 935 |
+
"logits/chosen": 206.3728485107422,
|
| 936 |
+
"logits/rejected": 213.7884063720703,
|
| 937 |
+
"logps/chosen": -0.6736012697219849,
|
| 938 |
+
"logps/rejected": -1.0033750534057617,
|
| 939 |
+
"loss": 35.165,
|
| 940 |
+
"nll_loss": 0.9300982356071472,
|
| 941 |
+
"rewards/accuracies": 0.706250011920929,
|
| 942 |
+
"rewards/chosen": -0.33680063486099243,
|
| 943 |
+
"rewards/margins": 0.16488692164421082,
|
| 944 |
+
"rewards/rejected": -0.5016875267028809,
|
| 945 |
+
"step": 260
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.5125925925925925,
|
| 949 |
+
"grad_norm": 47.25,
|
| 950 |
+
"learning_rate": 3.7531701693965554e-06,
|
| 951 |
+
"log_odds_chosen": 0.6154508590698242,
|
| 952 |
+
"log_odds_ratio": -0.5541368722915649,
|
| 953 |
+
"logits/chosen": 212.042724609375,
|
| 954 |
+
"logits/rejected": 211.31857299804688,
|
| 955 |
+
"logps/chosen": -0.6950085163116455,
|
| 956 |
+
"logps/rejected": -1.034769892692566,
|
| 957 |
+
"loss": 35.0564,
|
| 958 |
+
"nll_loss": 0.9271542429924011,
|
| 959 |
+
"rewards/accuracies": 0.762499988079071,
|
| 960 |
+
"rewards/chosen": -0.34750425815582275,
|
| 961 |
+
"rewards/margins": 0.1698807030916214,
|
| 962 |
+
"rewards/rejected": -0.517384946346283,
|
| 963 |
+
"step": 265
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.56,
|
| 967 |
+
"grad_norm": 46.75,
|
| 968 |
+
"learning_rate": 3.055003141378948e-06,
|
| 969 |
+
"log_odds_chosen": 0.6779940128326416,
|
| 970 |
+
"log_odds_ratio": -0.49747544527053833,
|
| 971 |
+
"logits/chosen": 207.754638671875,
|
| 972 |
+
"logits/rejected": 214.58056640625,
|
| 973 |
+
"logps/chosen": -0.7059646844863892,
|
| 974 |
+
"logps/rejected": -1.07295560836792,
|
| 975 |
+
"loss": 36.0368,
|
| 976 |
+
"nll_loss": 0.9621596336364746,
|
| 977 |
+
"rewards/accuracies": 0.75,
|
| 978 |
+
"rewards/chosen": -0.3529823422431946,
|
| 979 |
+
"rewards/margins": 0.1834954470396042,
|
| 980 |
+
"rewards/rejected": -0.53647780418396,
|
| 981 |
+
"step": 270
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.6074074074074076,
|
| 985 |
+
"grad_norm": 44.0,
|
| 986 |
+
"learning_rate": 2.424427429704365e-06,
|
| 987 |
+
"log_odds_chosen": 0.5993890166282654,
|
| 988 |
+
"log_odds_ratio": -0.5276089310646057,
|
| 989 |
+
"logits/chosen": 209.30874633789062,
|
| 990 |
+
"logits/rejected": 223.237548828125,
|
| 991 |
+
"logps/chosen": -0.713978111743927,
|
| 992 |
+
"logps/rejected": -1.013167142868042,
|
| 993 |
+
"loss": 36.1104,
|
| 994 |
+
"nll_loss": 0.9636555910110474,
|
| 995 |
+
"rewards/accuracies": 0.731249988079071,
|
| 996 |
+
"rewards/chosen": -0.3569890558719635,
|
| 997 |
+
"rewards/margins": 0.14959454536437988,
|
| 998 |
+
"rewards/rejected": -0.506583571434021,
|
| 999 |
+
"step": 275
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.6548148148148147,
|
| 1003 |
+
"grad_norm": 46.5,
|
| 1004 |
+
"learning_rate": 1.8633852284264508e-06,
|
| 1005 |
+
"log_odds_chosen": 0.641398549079895,
|
| 1006 |
+
"log_odds_ratio": -0.5212709307670593,
|
| 1007 |
+
"logits/chosen": 213.2962188720703,
|
| 1008 |
+
"logits/rejected": 211.81527709960938,
|
| 1009 |
+
"logps/chosen": -0.6817741394042969,
|
| 1010 |
+
"logps/rejected": -1.0385987758636475,
|
| 1011 |
+
"loss": 34.4299,
|
| 1012 |
+
"nll_loss": 0.8972374796867371,
|
| 1013 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1014 |
+
"rewards/chosen": -0.34088706970214844,
|
| 1015 |
+
"rewards/margins": 0.1784123331308365,
|
| 1016 |
+
"rewards/rejected": -0.5192993879318237,
|
| 1017 |
+
"step": 280
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.7022222222222223,
|
| 1021 |
+
"grad_norm": 44.0,
|
| 1022 |
+
"learning_rate": 1.3736045660864034e-06,
|
| 1023 |
+
"log_odds_chosen": 0.5530254244804382,
|
| 1024 |
+
"log_odds_ratio": -0.5443213582038879,
|
| 1025 |
+
"logits/chosen": 210.8925323486328,
|
| 1026 |
+
"logits/rejected": 220.40585327148438,
|
| 1027 |
+
"logps/chosen": -0.6845903396606445,
|
| 1028 |
+
"logps/rejected": -0.9716375470161438,
|
| 1029 |
+
"loss": 35.8726,
|
| 1030 |
+
"nll_loss": 0.9356240034103394,
|
| 1031 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1032 |
+
"rewards/chosen": -0.34229516983032227,
|
| 1033 |
+
"rewards/margins": 0.14352357387542725,
|
| 1034 |
+
"rewards/rejected": -0.4858187735080719,
|
| 1035 |
+
"step": 285
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.74962962962963,
|
| 1039 |
+
"grad_norm": 41.5,
|
| 1040 |
+
"learning_rate": 9.565939833279192e-07,
|
| 1041 |
+
"log_odds_chosen": 0.6134527325630188,
|
| 1042 |
+
"log_odds_ratio": -0.5205898284912109,
|
| 1043 |
+
"logits/chosen": 213.05337524414062,
|
| 1044 |
+
"logits/rejected": 214.83544921875,
|
| 1045 |
+
"logps/chosen": -0.7018261551856995,
|
| 1046 |
+
"logps/rejected": -1.010362982749939,
|
| 1047 |
+
"loss": 36.1954,
|
| 1048 |
+
"nll_loss": 0.976243793964386,
|
| 1049 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1050 |
+
"rewards/chosen": -0.35091307759284973,
|
| 1051 |
+
"rewards/margins": 0.15426845848560333,
|
| 1052 |
+
"rewards/rejected": -0.5051814913749695,
|
| 1053 |
+
"step": 290
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.797037037037037,
|
| 1057 |
+
"grad_norm": 107.5,
|
| 1058 |
+
"learning_rate": 6.136378865420872e-07,
|
| 1059 |
+
"log_odds_chosen": 0.6149747967720032,
|
| 1060 |
+
"log_odds_ratio": -0.5573899745941162,
|
| 1061 |
+
"logits/chosen": 213.50210571289062,
|
| 1062 |
+
"logits/rejected": 216.36660766601562,
|
| 1063 |
+
"logps/chosen": -0.7356334924697876,
|
| 1064 |
+
"logps/rejected": -1.0573605298995972,
|
| 1065 |
+
"loss": 35.9121,
|
| 1066 |
+
"nll_loss": 0.977961540222168,
|
| 1067 |
+
"rewards/accuracies": 0.75,
|
| 1068 |
+
"rewards/chosen": -0.3678167462348938,
|
| 1069 |
+
"rewards/margins": 0.16086354851722717,
|
| 1070 |
+
"rewards/rejected": -0.5286802649497986,
|
| 1071 |
+
"step": 295
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.8444444444444446,
|
| 1075 |
+
"grad_norm": 42.0,
|
| 1076 |
+
"learning_rate": 3.45792591853214e-07,
|
| 1077 |
+
"log_odds_chosen": 0.5418564081192017,
|
| 1078 |
+
"log_odds_ratio": -0.5683308839797974,
|
| 1079 |
+
"logits/chosen": 211.27420043945312,
|
| 1080 |
+
"logits/rejected": 218.8232879638672,
|
| 1081 |
+
"logps/chosen": -0.7445929646492004,
|
| 1082 |
+
"logps/rejected": -1.0526930093765259,
|
| 1083 |
+
"loss": 36.5699,
|
| 1084 |
+
"nll_loss": 1.0055818557739258,
|
| 1085 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1086 |
+
"rewards/chosen": -0.3722964823246002,
|
| 1087 |
+
"rewards/margins": 0.15405002236366272,
|
| 1088 |
+
"rewards/rejected": -0.5263465046882629,
|
| 1089 |
+
"step": 300
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.891851851851852,
|
| 1093 |
+
"grad_norm": 40.0,
|
| 1094 |
+
"learning_rate": 1.538830716302092e-07,
|
| 1095 |
+
"log_odds_chosen": 0.5894609093666077,
|
| 1096 |
+
"log_odds_ratio": -0.5367568731307983,
|
| 1097 |
+
"logits/chosen": 209.3128662109375,
|
| 1098 |
+
"logits/rejected": 217.3281707763672,
|
| 1099 |
+
"logps/chosen": -0.7301949262619019,
|
| 1100 |
+
"logps/rejected": -1.0283689498901367,
|
| 1101 |
+
"loss": 34.3942,
|
| 1102 |
+
"nll_loss": 0.911605954170227,
|
| 1103 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1104 |
+
"rewards/chosen": -0.3650974631309509,
|
| 1105 |
+
"rewards/margins": 0.14908698201179504,
|
| 1106 |
+
"rewards/rejected": -0.5141844749450684,
|
| 1107 |
+
"step": 305
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.9392592592592592,
|
| 1111 |
+
"grad_norm": 43.75,
|
| 1112 |
+
"learning_rate": 3.8500413544415025e-08,
|
| 1113 |
+
"log_odds_chosen": 0.626187801361084,
|
| 1114 |
+
"log_odds_ratio": -0.5338219404220581,
|
| 1115 |
+
"logits/chosen": 211.9689178466797,
|
| 1116 |
+
"logits/rejected": 221.7113800048828,
|
| 1117 |
+
"logps/chosen": -0.7173446416854858,
|
| 1118 |
+
"logps/rejected": -1.0214028358459473,
|
| 1119 |
+
"loss": 36.2169,
|
| 1120 |
+
"nll_loss": 0.935465931892395,
|
| 1121 |
+
"rewards/accuracies": 0.75,
|
| 1122 |
+
"rewards/chosen": -0.3586723208427429,
|
| 1123 |
+
"rewards/margins": 0.15202903747558594,
|
| 1124 |
+
"rewards/rejected": -0.5107014179229736,
|
| 1125 |
+
"step": 310
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.986666666666667,
|
| 1129 |
+
"grad_norm": 46.75,
|
| 1130 |
+
"learning_rate": 0.0,
|
| 1131 |
+
"log_odds_chosen": 0.6916165947914124,
|
| 1132 |
+
"log_odds_ratio": -0.48836082220077515,
|
| 1133 |
+
"logits/chosen": 211.38467407226562,
|
| 1134 |
+
"logits/rejected": 218.0712890625,
|
| 1135 |
+
"logps/chosen": -0.6407631039619446,
|
| 1136 |
+
"logps/rejected": -0.9831441044807434,
|
| 1137 |
+
"loss": 34.8711,
|
| 1138 |
+
"nll_loss": 0.9277550578117371,
|
| 1139 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1140 |
+
"rewards/chosen": -0.3203815519809723,
|
| 1141 |
+
"rewards/margins": 0.1711905300617218,
|
| 1142 |
+
"rewards/rejected": -0.4915720522403717,
|
| 1143 |
+
"step": 315
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.986666666666667,
|
| 1147 |
+
"step": 315,
|
| 1148 |
+
"total_flos": 0.0,
|
| 1149 |
+
"train_loss": 254.74021465967573,
|
| 1150 |
+
"train_runtime": 6927.757,
|
| 1151 |
+
"train_samples_per_second": 2.923,
|
| 1152 |
+
"train_steps_per_second": 0.045
|
| 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 |
+
}
|