Model save
Browse files- README.md +66 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +2472 -0
README.md
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: zephyr-orpo-7b-hehe
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for zephyr-orpo-7b-hehe
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1).
|
| 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="awsuineg/zephyr-orpo-7b-hehe", 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 |
+
This model was trained with ORPO, a method introduced in [ORPO: Monolithic Preference Optimization without Reference Model](https://huggingface.co/papers/2403.07691).
|
| 33 |
+
|
| 34 |
+
### Framework versions
|
| 35 |
+
|
| 36 |
+
- TRL: 0.12.1
|
| 37 |
+
- Transformers: 4.46.2
|
| 38 |
+
- Pytorch: 2.5.1
|
| 39 |
+
- Datasets: 3.1.0
|
| 40 |
+
- Tokenizers: 0.20.3
|
| 41 |
+
|
| 42 |
+
## Citations
|
| 43 |
+
|
| 44 |
+
Cite ORPO as:
|
| 45 |
+
|
| 46 |
+
```bibtex
|
| 47 |
+
@article{hong2024orpo,
|
| 48 |
+
title = {{ORPO: Monolithic Preference Optimization without Reference Model}},
|
| 49 |
+
author = {Jiwoo Hong and Noah Lee and James Thorne},
|
| 50 |
+
year = 2024,
|
| 51 |
+
eprint = {arXiv:2403.07691}
|
| 52 |
+
}
|
| 53 |
+
```
|
| 54 |
+
|
| 55 |
+
Cite TRL as:
|
| 56 |
+
|
| 57 |
+
```bibtex
|
| 58 |
+
@misc{vonwerra2022trl,
|
| 59 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 60 |
+
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},
|
| 61 |
+
year = 2020,
|
| 62 |
+
journal = {GitHub repository},
|
| 63 |
+
publisher = {GitHub},
|
| 64 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 65 |
+
}
|
| 66 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.4294498003771459,
|
| 5 |
+
"train_runtime": 7808.9602,
|
| 6 |
+
"train_samples": 7210,
|
| 7 |
+
"train_samples_per_second": 2.77,
|
| 8 |
+
"train_steps_per_second": 0.173
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.4294498003771459,
|
| 5 |
+
"train_runtime": 7808.9602,
|
| 6 |
+
"train_samples": 7210,
|
| 7 |
+
"train_samples_per_second": 2.77,
|
| 8 |
+
"train_steps_per_second": 0.173
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2472 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1353,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.022172949002217297,
|
| 13 |
+
"grad_norm": NaN,
|
| 14 |
+
"learning_rate": 1.9512195121951218e-06,
|
| 15 |
+
"log_odds_chosen": NaN,
|
| 16 |
+
"log_odds_ratio": NaN,
|
| 17 |
+
"logits/chosen": -2.5637989044189453,
|
| 18 |
+
"logits/rejected": -2.5734333992004395,
|
| 19 |
+
"logps/chosen": NaN,
|
| 20 |
+
"logps/rejected": NaN,
|
| 21 |
+
"loss": 1.8527,
|
| 22 |
+
"nll_loss": 1.088592767715454,
|
| 23 |
+
"rewards/accuracies": 0.512499988079071,
|
| 24 |
+
"rewards/chosen": NaN,
|
| 25 |
+
"rewards/margins": NaN,
|
| 26 |
+
"rewards/rejected": NaN,
|
| 27 |
+
"step": 10
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.04434589800443459,
|
| 31 |
+
"grad_norm": NaN,
|
| 32 |
+
"learning_rate": 3.9024390243902435e-06,
|
| 33 |
+
"log_odds_chosen": NaN,
|
| 34 |
+
"log_odds_ratio": NaN,
|
| 35 |
+
"logits/chosen": -2.5235965251922607,
|
| 36 |
+
"logits/rejected": -2.5694117546081543,
|
| 37 |
+
"logps/chosen": NaN,
|
| 38 |
+
"logps/rejected": NaN,
|
| 39 |
+
"loss": 1.6216,
|
| 40 |
+
"nll_loss": 1.042040467262268,
|
| 41 |
+
"rewards/accuracies": 0.612500011920929,
|
| 42 |
+
"rewards/chosen": NaN,
|
| 43 |
+
"rewards/margins": NaN,
|
| 44 |
+
"rewards/rejected": NaN,
|
| 45 |
+
"step": 20
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.06651884700665188,
|
| 49 |
+
"grad_norm": 0.7492002248764038,
|
| 50 |
+
"learning_rate": 5.853658536585366e-06,
|
| 51 |
+
"log_odds_chosen": NaN,
|
| 52 |
+
"log_odds_ratio": NaN,
|
| 53 |
+
"logits/chosen": -2.439809560775757,
|
| 54 |
+
"logits/rejected": -2.4339804649353027,
|
| 55 |
+
"logps/chosen": NaN,
|
| 56 |
+
"logps/rejected": NaN,
|
| 57 |
+
"loss": 1.3856,
|
| 58 |
+
"nll_loss": 1.040069818496704,
|
| 59 |
+
"rewards/accuracies": 0.5625,
|
| 60 |
+
"rewards/chosen": NaN,
|
| 61 |
+
"rewards/margins": NaN,
|
| 62 |
+
"rewards/rejected": NaN,
|
| 63 |
+
"step": 30
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.08869179600886919,
|
| 67 |
+
"grad_norm": NaN,
|
| 68 |
+
"learning_rate": 7.804878048780487e-06,
|
| 69 |
+
"log_odds_chosen": NaN,
|
| 70 |
+
"log_odds_ratio": NaN,
|
| 71 |
+
"logits/chosen": -2.5076301097869873,
|
| 72 |
+
"logits/rejected": -2.538536310195923,
|
| 73 |
+
"logps/chosen": NaN,
|
| 74 |
+
"logps/rejected": NaN,
|
| 75 |
+
"loss": 1.5238,
|
| 76 |
+
"nll_loss": 1.0452662706375122,
|
| 77 |
+
"rewards/accuracies": 0.637499988079071,
|
| 78 |
+
"rewards/chosen": NaN,
|
| 79 |
+
"rewards/margins": NaN,
|
| 80 |
+
"rewards/rejected": NaN,
|
| 81 |
+
"step": 40
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.11086474501108648,
|
| 85 |
+
"grad_norm": NaN,
|
| 86 |
+
"learning_rate": 7.999071182730533e-06,
|
| 87 |
+
"log_odds_chosen": NaN,
|
| 88 |
+
"log_odds_ratio": NaN,
|
| 89 |
+
"logits/chosen": -2.5563275814056396,
|
| 90 |
+
"logits/rejected": -2.5710463523864746,
|
| 91 |
+
"logps/chosen": NaN,
|
| 92 |
+
"logps/rejected": NaN,
|
| 93 |
+
"loss": 1.7136,
|
| 94 |
+
"nll_loss": 0.9837453961372375,
|
| 95 |
+
"rewards/accuracies": 0.574999988079071,
|
| 96 |
+
"rewards/chosen": NaN,
|
| 97 |
+
"rewards/margins": NaN,
|
| 98 |
+
"rewards/rejected": NaN,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.13303769401330376,
|
| 103 |
+
"grad_norm": NaN,
|
| 104 |
+
"learning_rate": 7.995861010152277e-06,
|
| 105 |
+
"log_odds_chosen": NaN,
|
| 106 |
+
"log_odds_ratio": NaN,
|
| 107 |
+
"logits/chosen": -2.422978162765503,
|
| 108 |
+
"logits/rejected": -2.497622013092041,
|
| 109 |
+
"logps/chosen": NaN,
|
| 110 |
+
"logps/rejected": NaN,
|
| 111 |
+
"loss": 1.4346,
|
| 112 |
+
"nll_loss": 1.1125876903533936,
|
| 113 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 114 |
+
"rewards/chosen": NaN,
|
| 115 |
+
"rewards/margins": NaN,
|
| 116 |
+
"rewards/rejected": NaN,
|
| 117 |
+
"step": 60
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.15521064301552107,
|
| 121 |
+
"grad_norm": NaN,
|
| 122 |
+
"learning_rate": 7.990359855463492e-06,
|
| 123 |
+
"log_odds_chosen": NaN,
|
| 124 |
+
"log_odds_ratio": NaN,
|
| 125 |
+
"logits/chosen": -2.4739160537719727,
|
| 126 |
+
"logits/rejected": -2.5015368461608887,
|
| 127 |
+
"logps/chosen": NaN,
|
| 128 |
+
"logps/rejected": NaN,
|
| 129 |
+
"loss": 1.7054,
|
| 130 |
+
"nll_loss": 1.0281422138214111,
|
| 131 |
+
"rewards/accuracies": 0.5,
|
| 132 |
+
"rewards/chosen": NaN,
|
| 133 |
+
"rewards/margins": NaN,
|
| 134 |
+
"rewards/rejected": NaN,
|
| 135 |
+
"step": 70
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.17738359201773837,
|
| 139 |
+
"grad_norm": NaN,
|
| 140 |
+
"learning_rate": 7.982570872689543e-06,
|
| 141 |
+
"log_odds_chosen": NaN,
|
| 142 |
+
"log_odds_ratio": NaN,
|
| 143 |
+
"logits/chosen": -2.5495190620422363,
|
| 144 |
+
"logits/rejected": -2.5661299228668213,
|
| 145 |
+
"logps/chosen": NaN,
|
| 146 |
+
"logps/rejected": NaN,
|
| 147 |
+
"loss": 1.9447,
|
| 148 |
+
"nll_loss": 1.0821787118911743,
|
| 149 |
+
"rewards/accuracies": 0.550000011920929,
|
| 150 |
+
"rewards/chosen": NaN,
|
| 151 |
+
"rewards/margins": NaN,
|
| 152 |
+
"rewards/rejected": NaN,
|
| 153 |
+
"step": 80
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.19955654101995565,
|
| 157 |
+
"grad_norm": NaN,
|
| 158 |
+
"learning_rate": 7.972498527556375e-06,
|
| 159 |
+
"log_odds_chosen": NaN,
|
| 160 |
+
"log_odds_ratio": NaN,
|
| 161 |
+
"logits/chosen": -2.539638042449951,
|
| 162 |
+
"logits/rejected": -2.565432071685791,
|
| 163 |
+
"logps/chosen": NaN,
|
| 164 |
+
"logps/rejected": NaN,
|
| 165 |
+
"loss": 1.4908,
|
| 166 |
+
"nll_loss": 1.0412414073944092,
|
| 167 |
+
"rewards/accuracies": 0.6875,
|
| 168 |
+
"rewards/chosen": NaN,
|
| 169 |
+
"rewards/margins": NaN,
|
| 170 |
+
"rewards/rejected": NaN,
|
| 171 |
+
"step": 90
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.22172949002217296,
|
| 175 |
+
"grad_norm": NaN,
|
| 176 |
+
"learning_rate": 7.960148594930148e-06,
|
| 177 |
+
"log_odds_chosen": NaN,
|
| 178 |
+
"log_odds_ratio": NaN,
|
| 179 |
+
"logits/chosen": -2.5148587226867676,
|
| 180 |
+
"logits/rejected": -2.531787395477295,
|
| 181 |
+
"logps/chosen": NaN,
|
| 182 |
+
"logps/rejected": NaN,
|
| 183 |
+
"loss": 1.3802,
|
| 184 |
+
"nll_loss": 1.0808457136154175,
|
| 185 |
+
"rewards/accuracies": 0.550000011920929,
|
| 186 |
+
"rewards/chosen": NaN,
|
| 187 |
+
"rewards/margins": NaN,
|
| 188 |
+
"rewards/rejected": NaN,
|
| 189 |
+
"step": 100
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.24390243902439024,
|
| 193 |
+
"grad_norm": NaN,
|
| 194 |
+
"learning_rate": 7.945528155506268e-06,
|
| 195 |
+
"log_odds_chosen": NaN,
|
| 196 |
+
"log_odds_ratio": NaN,
|
| 197 |
+
"logits/chosen": -2.4936861991882324,
|
| 198 |
+
"logits/rejected": -2.5372276306152344,
|
| 199 |
+
"logps/chosen": NaN,
|
| 200 |
+
"logps/rejected": NaN,
|
| 201 |
+
"loss": 1.4052,
|
| 202 |
+
"nll_loss": 1.0503720045089722,
|
| 203 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 204 |
+
"rewards/chosen": NaN,
|
| 205 |
+
"rewards/margins": NaN,
|
| 206 |
+
"rewards/rejected": NaN,
|
| 207 |
+
"step": 110
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.2660753880266075,
|
| 211 |
+
"grad_norm": NaN,
|
| 212 |
+
"learning_rate": 7.928645591749765e-06,
|
| 213 |
+
"log_odds_chosen": NaN,
|
| 214 |
+
"log_odds_ratio": NaN,
|
| 215 |
+
"logits/chosen": -2.5514473915100098,
|
| 216 |
+
"logits/rejected": -2.5984888076782227,
|
| 217 |
+
"logps/chosen": NaN,
|
| 218 |
+
"logps/rejected": NaN,
|
| 219 |
+
"loss": 1.6079,
|
| 220 |
+
"nll_loss": 1.0542423725128174,
|
| 221 |
+
"rewards/accuracies": 0.574999988079071,
|
| 222 |
+
"rewards/chosen": NaN,
|
| 223 |
+
"rewards/margins": NaN,
|
| 224 |
+
"rewards/rejected": NaN,
|
| 225 |
+
"step": 120
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.28824833702882485,
|
| 229 |
+
"grad_norm": NaN,
|
| 230 |
+
"learning_rate": 7.909510583089285e-06,
|
| 231 |
+
"log_odds_chosen": NaN,
|
| 232 |
+
"log_odds_ratio": NaN,
|
| 233 |
+
"logits/chosen": -2.513597011566162,
|
| 234 |
+
"logits/rejected": -2.509930372238159,
|
| 235 |
+
"logps/chosen": NaN,
|
| 236 |
+
"logps/rejected": NaN,
|
| 237 |
+
"loss": 1.4307,
|
| 238 |
+
"nll_loss": 0.9876394271850586,
|
| 239 |
+
"rewards/accuracies": 0.625,
|
| 240 |
+
"rewards/chosen": NaN,
|
| 241 |
+
"rewards/margins": NaN,
|
| 242 |
+
"rewards/rejected": NaN,
|
| 243 |
+
"step": 130
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.31042128603104213,
|
| 247 |
+
"grad_norm": 0.5560552477836609,
|
| 248 |
+
"learning_rate": 7.888134100367517e-06,
|
| 249 |
+
"log_odds_chosen": NaN,
|
| 250 |
+
"log_odds_ratio": NaN,
|
| 251 |
+
"logits/chosen": -2.5325217247009277,
|
| 252 |
+
"logits/rejected": -2.5615830421447754,
|
| 253 |
+
"logps/chosen": NaN,
|
| 254 |
+
"logps/rejected": NaN,
|
| 255 |
+
"loss": 1.5637,
|
| 256 |
+
"nll_loss": 1.090905785560608,
|
| 257 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 258 |
+
"rewards/chosen": NaN,
|
| 259 |
+
"rewards/margins": NaN,
|
| 260 |
+
"rewards/rejected": NaN,
|
| 261 |
+
"step": 140
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.3325942350332594,
|
| 265 |
+
"grad_norm": NaN,
|
| 266 |
+
"learning_rate": 7.864528399551163e-06,
|
| 267 |
+
"log_odds_chosen": NaN,
|
| 268 |
+
"log_odds_ratio": NaN,
|
| 269 |
+
"logits/chosen": -2.5349364280700684,
|
| 270 |
+
"logits/rejected": -2.555640459060669,
|
| 271 |
+
"logps/chosen": NaN,
|
| 272 |
+
"logps/rejected": NaN,
|
| 273 |
+
"loss": 1.2739,
|
| 274 |
+
"nll_loss": 0.984653115272522,
|
| 275 |
+
"rewards/accuracies": 0.612500011920929,
|
| 276 |
+
"rewards/chosen": NaN,
|
| 277 |
+
"rewards/margins": NaN,
|
| 278 |
+
"rewards/rejected": NaN,
|
| 279 |
+
"step": 150
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.35476718403547675,
|
| 283 |
+
"grad_norm": 0.49410519003868103,
|
| 284 |
+
"learning_rate": 7.83870701470413e-06,
|
| 285 |
+
"log_odds_chosen": NaN,
|
| 286 |
+
"log_odds_ratio": NaN,
|
| 287 |
+
"logits/chosen": -2.579796075820923,
|
| 288 |
+
"logits/rejected": -2.6130154132843018,
|
| 289 |
+
"logps/chosen": NaN,
|
| 290 |
+
"logps/rejected": NaN,
|
| 291 |
+
"loss": 1.6816,
|
| 292 |
+
"nll_loss": 0.9771108627319336,
|
| 293 |
+
"rewards/accuracies": 0.637499988079071,
|
| 294 |
+
"rewards/chosen": NaN,
|
| 295 |
+
"rewards/margins": NaN,
|
| 296 |
+
"rewards/rejected": NaN,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.376940133037694,
|
| 301 |
+
"grad_norm": NaN,
|
| 302 |
+
"learning_rate": 7.810684750227926e-06,
|
| 303 |
+
"log_odds_chosen": NaN,
|
| 304 |
+
"log_odds_ratio": NaN,
|
| 305 |
+
"logits/chosen": -2.5355019569396973,
|
| 306 |
+
"logits/rejected": -2.5664353370666504,
|
| 307 |
+
"logps/chosen": NaN,
|
| 308 |
+
"logps/rejected": NaN,
|
| 309 |
+
"loss": 1.4045,
|
| 310 |
+
"nll_loss": 0.9972349405288696,
|
| 311 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 312 |
+
"rewards/chosen": NaN,
|
| 313 |
+
"rewards/margins": NaN,
|
| 314 |
+
"rewards/rejected": NaN,
|
| 315 |
+
"step": 170
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.3991130820399113,
|
| 319 |
+
"grad_norm": NaN,
|
| 320 |
+
"learning_rate": 7.780477672373715e-06,
|
| 321 |
+
"log_odds_chosen": NaN,
|
| 322 |
+
"log_odds_ratio": NaN,
|
| 323 |
+
"logits/chosen": -2.610077381134033,
|
| 324 |
+
"logits/rejected": -2.6662240028381348,
|
| 325 |
+
"logps/chosen": NaN,
|
| 326 |
+
"logps/rejected": NaN,
|
| 327 |
+
"loss": 1.3857,
|
| 328 |
+
"nll_loss": 0.9627612829208374,
|
| 329 |
+
"rewards/accuracies": 0.637499988079071,
|
| 330 |
+
"rewards/chosen": NaN,
|
| 331 |
+
"rewards/margins": NaN,
|
| 332 |
+
"rewards/rejected": NaN,
|
| 333 |
+
"step": 180
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.4212860310421286,
|
| 337 |
+
"grad_norm": NaN,
|
| 338 |
+
"learning_rate": 7.748103100030918e-06,
|
| 339 |
+
"log_odds_chosen": NaN,
|
| 340 |
+
"log_odds_ratio": NaN,
|
| 341 |
+
"logits/chosen": -2.5606350898742676,
|
| 342 |
+
"logits/rejected": -2.599088430404663,
|
| 343 |
+
"logps/chosen": NaN,
|
| 344 |
+
"logps/rejected": NaN,
|
| 345 |
+
"loss": 1.7501,
|
| 346 |
+
"nll_loss": 0.9361652135848999,
|
| 347 |
+
"rewards/accuracies": 0.574999988079071,
|
| 348 |
+
"rewards/chosen": NaN,
|
| 349 |
+
"rewards/margins": NaN,
|
| 350 |
+
"rewards/rejected": NaN,
|
| 351 |
+
"step": 190
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.4434589800443459,
|
| 355 |
+
"grad_norm": NaN,
|
| 356 |
+
"learning_rate": 7.713579594797617e-06,
|
| 357 |
+
"log_odds_chosen": NaN,
|
| 358 |
+
"log_odds_ratio": NaN,
|
| 359 |
+
"logits/chosen": -2.5864617824554443,
|
| 360 |
+
"logits/rejected": -2.6177988052368164,
|
| 361 |
+
"logps/chosen": NaN,
|
| 362 |
+
"logps/rejected": NaN,
|
| 363 |
+
"loss": 1.5321,
|
| 364 |
+
"nll_loss": 0.9839666485786438,
|
| 365 |
+
"rewards/accuracies": 0.550000011920929,
|
| 366 |
+
"rewards/chosen": NaN,
|
| 367 |
+
"rewards/margins": NaN,
|
| 368 |
+
"rewards/rejected": NaN,
|
| 369 |
+
"step": 200
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.4656319290465632,
|
| 373 |
+
"grad_norm": NaN,
|
| 374 |
+
"learning_rate": 7.676926950338484e-06,
|
| 375 |
+
"log_odds_chosen": NaN,
|
| 376 |
+
"log_odds_ratio": NaN,
|
| 377 |
+
"logits/chosen": -2.5560379028320312,
|
| 378 |
+
"logits/rejected": -2.6088876724243164,
|
| 379 |
+
"logps/chosen": NaN,
|
| 380 |
+
"logps/rejected": NaN,
|
| 381 |
+
"loss": 1.4582,
|
| 382 |
+
"nll_loss": 0.9833539724349976,
|
| 383 |
+
"rewards/accuracies": 0.625,
|
| 384 |
+
"rewards/chosen": NaN,
|
| 385 |
+
"rewards/margins": NaN,
|
| 386 |
+
"rewards/rejected": NaN,
|
| 387 |
+
"step": 210
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.4878048780487805,
|
| 391 |
+
"grad_norm": NaN,
|
| 392 |
+
"learning_rate": 7.638166181036278e-06,
|
| 393 |
+
"log_odds_chosen": NaN,
|
| 394 |
+
"log_odds_ratio": NaN,
|
| 395 |
+
"logits/chosen": -2.6120612621307373,
|
| 396 |
+
"logits/rejected": -2.648507595062256,
|
| 397 |
+
"logps/chosen": NaN,
|
| 398 |
+
"logps/rejected": NaN,
|
| 399 |
+
"loss": 1.7413,
|
| 400 |
+
"nll_loss": 0.9990829229354858,
|
| 401 |
+
"rewards/accuracies": 0.550000011920929,
|
| 402 |
+
"rewards/chosen": NaN,
|
| 403 |
+
"rewards/margins": NaN,
|
| 404 |
+
"rewards/rejected": NaN,
|
| 405 |
+
"step": 220
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.5099778270509978,
|
| 409 |
+
"grad_norm": NaN,
|
| 410 |
+
"learning_rate": 7.597319509943522e-06,
|
| 411 |
+
"log_odds_chosen": NaN,
|
| 412 |
+
"log_odds_ratio": NaN,
|
| 413 |
+
"logits/chosen": -2.5437393188476562,
|
| 414 |
+
"logits/rejected": -2.5923430919647217,
|
| 415 |
+
"logps/chosen": NaN,
|
| 416 |
+
"logps/rejected": NaN,
|
| 417 |
+
"loss": 1.3477,
|
| 418 |
+
"nll_loss": 0.9254695177078247,
|
| 419 |
+
"rewards/accuracies": 0.550000011920929,
|
| 420 |
+
"rewards/chosen": NaN,
|
| 421 |
+
"rewards/margins": NaN,
|
| 422 |
+
"rewards/rejected": NaN,
|
| 423 |
+
"step": 230
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.532150776053215,
|
| 427 |
+
"grad_norm": NaN,
|
| 428 |
+
"learning_rate": 7.554410356041128e-06,
|
| 429 |
+
"log_odds_chosen": NaN,
|
| 430 |
+
"log_odds_ratio": NaN,
|
| 431 |
+
"logits/chosen": -2.5858166217803955,
|
| 432 |
+
"logits/rejected": -2.645839214324951,
|
| 433 |
+
"logps/chosen": NaN,
|
| 434 |
+
"logps/rejected": NaN,
|
| 435 |
+
"loss": 1.8364,
|
| 436 |
+
"nll_loss": 0.9360870122909546,
|
| 437 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 438 |
+
"rewards/chosen": NaN,
|
| 439 |
+
"rewards/margins": NaN,
|
| 440 |
+
"rewards/rejected": NaN,
|
| 441 |
+
"step": 240
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.5543237250554324,
|
| 445 |
+
"grad_norm": NaN,
|
| 446 |
+
"learning_rate": 7.509463320811409e-06,
|
| 447 |
+
"log_odds_chosen": NaN,
|
| 448 |
+
"log_odds_ratio": NaN,
|
| 449 |
+
"logits/chosen": -2.603842258453369,
|
| 450 |
+
"logits/rejected": -2.65669584274292,
|
| 451 |
+
"logps/chosen": NaN,
|
| 452 |
+
"logps/rejected": NaN,
|
| 453 |
+
"loss": 1.9066,
|
| 454 |
+
"nll_loss": 0.8704744577407837,
|
| 455 |
+
"rewards/accuracies": 0.612500011920929,
|
| 456 |
+
"rewards/chosen": NaN,
|
| 457 |
+
"rewards/margins": NaN,
|
| 458 |
+
"rewards/rejected": NaN,
|
| 459 |
+
"step": 250
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.5764966740576497,
|
| 463 |
+
"grad_norm": NaN,
|
| 464 |
+
"learning_rate": 7.462504174133093e-06,
|
| 465 |
+
"log_odds_chosen": NaN,
|
| 466 |
+
"log_odds_ratio": NaN,
|
| 467 |
+
"logits/chosen": -2.5409696102142334,
|
| 468 |
+
"logits/rejected": -2.58134388923645,
|
| 469 |
+
"logps/chosen": NaN,
|
| 470 |
+
"logps/rejected": NaN,
|
| 471 |
+
"loss": 1.4614,
|
| 472 |
+
"nll_loss": 0.8922187685966492,
|
| 473 |
+
"rewards/accuracies": 0.625,
|
| 474 |
+
"rewards/chosen": NaN,
|
| 475 |
+
"rewards/margins": NaN,
|
| 476 |
+
"rewards/rejected": NaN,
|
| 477 |
+
"step": 260
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.5986696230598669,
|
| 481 |
+
"grad_norm": NaN,
|
| 482 |
+
"learning_rate": 7.413559839506442e-06,
|
| 483 |
+
"log_odds_chosen": NaN,
|
| 484 |
+
"log_odds_ratio": NaN,
|
| 485 |
+
"logits/chosen": -2.6056294441223145,
|
| 486 |
+
"logits/rejected": -2.59102725982666,
|
| 487 |
+
"logps/chosen": NaN,
|
| 488 |
+
"logps/rejected": NaN,
|
| 489 |
+
"loss": 1.4101,
|
| 490 |
+
"nll_loss": 0.9519271850585938,
|
| 491 |
+
"rewards/accuracies": 0.512499988079071,
|
| 492 |
+
"rewards/chosen": NaN,
|
| 493 |
+
"rewards/margins": NaN,
|
| 494 |
+
"rewards/rejected": NaN,
|
| 495 |
+
"step": 270
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.6208425720620843,
|
| 499 |
+
"grad_norm": NaN,
|
| 500 |
+
"learning_rate": 7.362658378616977e-06,
|
| 501 |
+
"log_odds_chosen": NaN,
|
| 502 |
+
"log_odds_ratio": NaN,
|
| 503 |
+
"logits/chosen": -2.518075942993164,
|
| 504 |
+
"logits/rejected": -2.5656614303588867,
|
| 505 |
+
"logps/chosen": NaN,
|
| 506 |
+
"logps/rejected": NaN,
|
| 507 |
+
"loss": 1.7219,
|
| 508 |
+
"nll_loss": 0.9626830816268921,
|
| 509 |
+
"rewards/accuracies": 0.675000011920929,
|
| 510 |
+
"rewards/chosen": NaN,
|
| 511 |
+
"rewards/margins": NaN,
|
| 512 |
+
"rewards/rejected": NaN,
|
| 513 |
+
"step": 280
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.6430155210643016,
|
| 517 |
+
"grad_norm": NaN,
|
| 518 |
+
"learning_rate": 7.309828975246615e-06,
|
| 519 |
+
"log_odds_chosen": NaN,
|
| 520 |
+
"log_odds_ratio": NaN,
|
| 521 |
+
"logits/chosen": -2.478224277496338,
|
| 522 |
+
"logits/rejected": -2.4818527698516846,
|
| 523 |
+
"logps/chosen": NaN,
|
| 524 |
+
"logps/rejected": NaN,
|
| 525 |
+
"loss": 1.4698,
|
| 526 |
+
"nll_loss": 0.9482153654098511,
|
| 527 |
+
"rewards/accuracies": 0.637499988079071,
|
| 528 |
+
"rewards/chosen": NaN,
|
| 529 |
+
"rewards/margins": NaN,
|
| 530 |
+
"rewards/rejected": NaN,
|
| 531 |
+
"step": 290
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.6651884700665188,
|
| 535 |
+
"grad_norm": NaN,
|
| 536 |
+
"learning_rate": 7.255101918541482e-06,
|
| 537 |
+
"log_odds_chosen": NaN,
|
| 538 |
+
"log_odds_ratio": NaN,
|
| 539 |
+
"logits/chosen": -2.5367703437805176,
|
| 540 |
+
"logits/rejected": -2.618227005004883,
|
| 541 |
+
"logps/chosen": NaN,
|
| 542 |
+
"logps/rejected": NaN,
|
| 543 |
+
"loss": 1.3779,
|
| 544 |
+
"nll_loss": 0.9249935150146484,
|
| 545 |
+
"rewards/accuracies": 0.612500011920929,
|
| 546 |
+
"rewards/chosen": NaN,
|
| 547 |
+
"rewards/margins": NaN,
|
| 548 |
+
"rewards/rejected": NaN,
|
| 549 |
+
"step": 300
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.6873614190687362,
|
| 553 |
+
"grad_norm": NaN,
|
| 554 |
+
"learning_rate": 7.198508585645966e-06,
|
| 555 |
+
"log_odds_chosen": NaN,
|
| 556 |
+
"log_odds_ratio": NaN,
|
| 557 |
+
"logits/chosen": -2.498709201812744,
|
| 558 |
+
"logits/rejected": -2.5306169986724854,
|
| 559 |
+
"logps/chosen": NaN,
|
| 560 |
+
"logps/rejected": NaN,
|
| 561 |
+
"loss": 1.4247,
|
| 562 |
+
"nll_loss": 0.9907130002975464,
|
| 563 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 564 |
+
"rewards/chosen": NaN,
|
| 565 |
+
"rewards/margins": NaN,
|
| 566 |
+
"rewards/rejected": NaN,
|
| 567 |
+
"step": 310
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.7095343680709535,
|
| 571 |
+
"grad_norm": 0.5340009331703186,
|
| 572 |
+
"learning_rate": 7.140081423712985e-06,
|
| 573 |
+
"log_odds_chosen": NaN,
|
| 574 |
+
"log_odds_ratio": NaN,
|
| 575 |
+
"logits/chosen": -2.576976776123047,
|
| 576 |
+
"logits/rejected": -2.5837340354919434,
|
| 577 |
+
"logps/chosen": NaN,
|
| 578 |
+
"logps/rejected": NaN,
|
| 579 |
+
"loss": 1.3265,
|
| 580 |
+
"nll_loss": 0.9744553565979004,
|
| 581 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 582 |
+
"rewards/chosen": NaN,
|
| 583 |
+
"rewards/margins": NaN,
|
| 584 |
+
"rewards/rejected": NaN,
|
| 585 |
+
"step": 320
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.7317073170731707,
|
| 589 |
+
"grad_norm": NaN,
|
| 590 |
+
"learning_rate": 7.079853931300778e-06,
|
| 591 |
+
"log_odds_chosen": NaN,
|
| 592 |
+
"log_odds_ratio": NaN,
|
| 593 |
+
"logits/chosen": -2.5235087871551514,
|
| 594 |
+
"logits/rejected": -2.537680149078369,
|
| 595 |
+
"logps/chosen": NaN,
|
| 596 |
+
"logps/rejected": NaN,
|
| 597 |
+
"loss": 1.3516,
|
| 598 |
+
"nll_loss": 0.9415532350540161,
|
| 599 |
+
"rewards/accuracies": 0.612500011920929,
|
| 600 |
+
"rewards/chosen": NaN,
|
| 601 |
+
"rewards/margins": NaN,
|
| 602 |
+
"rewards/rejected": NaN,
|
| 603 |
+
"step": 330
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.753880266075388,
|
| 607 |
+
"grad_norm": NaN,
|
| 608 |
+
"learning_rate": 7.017860639166877e-06,
|
| 609 |
+
"log_odds_chosen": NaN,
|
| 610 |
+
"log_odds_ratio": NaN,
|
| 611 |
+
"logits/chosen": -2.561636447906494,
|
| 612 |
+
"logits/rejected": -2.6029891967773438,
|
| 613 |
+
"logps/chosen": NaN,
|
| 614 |
+
"logps/rejected": NaN,
|
| 615 |
+
"loss": 1.0444,
|
| 616 |
+
"nll_loss": 0.87201327085495,
|
| 617 |
+
"rewards/accuracies": 0.625,
|
| 618 |
+
"rewards/chosen": NaN,
|
| 619 |
+
"rewards/margins": NaN,
|
| 620 |
+
"rewards/rejected": NaN,
|
| 621 |
+
"step": 340
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.7760532150776053,
|
| 625 |
+
"grad_norm": NaN,
|
| 626 |
+
"learning_rate": 6.95413709047029e-06,
|
| 627 |
+
"log_odds_chosen": NaN,
|
| 628 |
+
"log_odds_ratio": NaN,
|
| 629 |
+
"logits/chosen": -2.6087021827697754,
|
| 630 |
+
"logits/rejected": -2.631678342819214,
|
| 631 |
+
"logps/chosen": NaN,
|
| 632 |
+
"logps/rejected": NaN,
|
| 633 |
+
"loss": 1.241,
|
| 634 |
+
"nll_loss": 0.9579289555549622,
|
| 635 |
+
"rewards/accuracies": 0.574999988079071,
|
| 636 |
+
"rewards/chosen": NaN,
|
| 637 |
+
"rewards/margins": NaN,
|
| 638 |
+
"rewards/rejected": NaN,
|
| 639 |
+
"step": 350
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.7982261640798226,
|
| 643 |
+
"grad_norm": 0.6355442404747009,
|
| 644 |
+
"learning_rate": 6.888719820393224e-06,
|
| 645 |
+
"log_odds_chosen": NaN,
|
| 646 |
+
"log_odds_ratio": NaN,
|
| 647 |
+
"logits/chosen": -2.5300655364990234,
|
| 648 |
+
"logits/rejected": -2.538252592086792,
|
| 649 |
+
"logps/chosen": NaN,
|
| 650 |
+
"logps/rejected": NaN,
|
| 651 |
+
"loss": 1.1996,
|
| 652 |
+
"nll_loss": 0.9421980977058411,
|
| 653 |
+
"rewards/accuracies": 0.550000011920929,
|
| 654 |
+
"rewards/chosen": NaN,
|
| 655 |
+
"rewards/margins": NaN,
|
| 656 |
+
"rewards/rejected": NaN,
|
| 657 |
+
"step": 360
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.8203991130820399,
|
| 661 |
+
"grad_norm": NaN,
|
| 662 |
+
"learning_rate": 6.821646335194051e-06,
|
| 663 |
+
"log_odds_chosen": NaN,
|
| 664 |
+
"log_odds_ratio": NaN,
|
| 665 |
+
"logits/chosen": -2.4321510791778564,
|
| 666 |
+
"logits/rejected": -2.488974094390869,
|
| 667 |
+
"logps/chosen": NaN,
|
| 668 |
+
"logps/rejected": NaN,
|
| 669 |
+
"loss": 1.5151,
|
| 670 |
+
"nll_loss": 0.9701131582260132,
|
| 671 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 672 |
+
"rewards/chosen": NaN,
|
| 673 |
+
"rewards/margins": NaN,
|
| 674 |
+
"rewards/rejected": NaN,
|
| 675 |
+
"step": 370
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.8425720620842572,
|
| 679 |
+
"grad_norm": NaN,
|
| 680 |
+
"learning_rate": 6.752955090703516e-06,
|
| 681 |
+
"log_odds_chosen": NaN,
|
| 682 |
+
"log_odds_ratio": NaN,
|
| 683 |
+
"logits/chosen": -2.5075130462646484,
|
| 684 |
+
"logits/rejected": -2.576809883117676,
|
| 685 |
+
"logps/chosen": NaN,
|
| 686 |
+
"logps/rejected": NaN,
|
| 687 |
+
"loss": 1.6486,
|
| 688 |
+
"nll_loss": 0.8745073080062866,
|
| 689 |
+
"rewards/accuracies": 0.699999988079071,
|
| 690 |
+
"rewards/chosen": NaN,
|
| 691 |
+
"rewards/margins": NaN,
|
| 692 |
+
"rewards/rejected": NaN,
|
| 693 |
+
"step": 380
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.8647450110864745,
|
| 697 |
+
"grad_norm": NaN,
|
| 698 |
+
"learning_rate": 6.682685470276513e-06,
|
| 699 |
+
"log_odds_chosen": NaN,
|
| 700 |
+
"log_odds_ratio": NaN,
|
| 701 |
+
"logits/chosen": -2.609452247619629,
|
| 702 |
+
"logits/rejected": -2.6642494201660156,
|
| 703 |
+
"logps/chosen": NaN,
|
| 704 |
+
"logps/rejected": NaN,
|
| 705 |
+
"loss": 1.6961,
|
| 706 |
+
"nll_loss": 0.8854317665100098,
|
| 707 |
+
"rewards/accuracies": 0.574999988079071,
|
| 708 |
+
"rewards/chosen": NaN,
|
| 709 |
+
"rewards/margins": NaN,
|
| 710 |
+
"rewards/rejected": NaN,
|
| 711 |
+
"step": 390
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.8869179600886918,
|
| 715 |
+
"grad_norm": 0.49902427196502686,
|
| 716 |
+
"learning_rate": 6.610877762212086e-06,
|
| 717 |
+
"log_odds_chosen": NaN,
|
| 718 |
+
"log_odds_ratio": NaN,
|
| 719 |
+
"logits/chosen": -2.5146374702453613,
|
| 720 |
+
"logits/rejected": -2.529111385345459,
|
| 721 |
+
"logps/chosen": NaN,
|
| 722 |
+
"logps/rejected": NaN,
|
| 723 |
+
"loss": 1.2123,
|
| 724 |
+
"nll_loss": 0.9021833539009094,
|
| 725 |
+
"rewards/accuracies": 0.612500011920929,
|
| 726 |
+
"rewards/chosen": NaN,
|
| 727 |
+
"rewards/margins": NaN,
|
| 728 |
+
"rewards/rejected": NaN,
|
| 729 |
+
"step": 400
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.9090909090909091,
|
| 733 |
+
"grad_norm": NaN,
|
| 734 |
+
"learning_rate": 6.537573136654582e-06,
|
| 735 |
+
"log_odds_chosen": NaN,
|
| 736 |
+
"log_odds_ratio": NaN,
|
| 737 |
+
"logits/chosen": -2.5666518211364746,
|
| 738 |
+
"logits/rejected": -2.594228744506836,
|
| 739 |
+
"logps/chosen": NaN,
|
| 740 |
+
"logps/rejected": NaN,
|
| 741 |
+
"loss": 1.4251,
|
| 742 |
+
"nll_loss": 0.9093640446662903,
|
| 743 |
+
"rewards/accuracies": 0.574999988079071,
|
| 744 |
+
"rewards/chosen": NaN,
|
| 745 |
+
"rewards/margins": NaN,
|
| 746 |
+
"rewards/rejected": NaN,
|
| 747 |
+
"step": 410
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.9312638580931264,
|
| 751 |
+
"grad_norm": NaN,
|
| 752 |
+
"learning_rate": 6.462813621989207e-06,
|
| 753 |
+
"log_odds_chosen": NaN,
|
| 754 |
+
"log_odds_ratio": NaN,
|
| 755 |
+
"logits/chosen": -2.551781177520752,
|
| 756 |
+
"logits/rejected": -2.5560076236724854,
|
| 757 |
+
"logps/chosen": NaN,
|
| 758 |
+
"logps/rejected": NaN,
|
| 759 |
+
"loss": 1.3258,
|
| 760 |
+
"nll_loss": 0.9460923075675964,
|
| 761 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 762 |
+
"rewards/chosen": NaN,
|
| 763 |
+
"rewards/margins": NaN,
|
| 764 |
+
"rewards/rejected": NaN,
|
| 765 |
+
"step": 420
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.9534368070953437,
|
| 769 |
+
"grad_norm": NaN,
|
| 770 |
+
"learning_rate": 6.386642080745528e-06,
|
| 771 |
+
"log_odds_chosen": NaN,
|
| 772 |
+
"log_odds_ratio": NaN,
|
| 773 |
+
"logits/chosen": -2.5597198009490967,
|
| 774 |
+
"logits/rejected": -2.5898067951202393,
|
| 775 |
+
"logps/chosen": NaN,
|
| 776 |
+
"logps/rejected": NaN,
|
| 777 |
+
"loss": 1.2546,
|
| 778 |
+
"nll_loss": 0.9564792513847351,
|
| 779 |
+
"rewards/accuracies": 0.574999988079071,
|
| 780 |
+
"rewards/chosen": NaN,
|
| 781 |
+
"rewards/margins": NaN,
|
| 782 |
+
"rewards/rejected": NaN,
|
| 783 |
+
"step": 430
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.975609756097561,
|
| 787 |
+
"grad_norm": 0.4311073124408722,
|
| 788 |
+
"learning_rate": 6.30910218502272e-06,
|
| 789 |
+
"log_odds_chosen": NaN,
|
| 790 |
+
"log_odds_ratio": NaN,
|
| 791 |
+
"logits/chosen": -2.6047396659851074,
|
| 792 |
+
"logits/rejected": -2.6136045455932617,
|
| 793 |
+
"logps/chosen": NaN,
|
| 794 |
+
"logps/rejected": NaN,
|
| 795 |
+
"loss": 1.3434,
|
| 796 |
+
"nll_loss": 0.96406489610672,
|
| 797 |
+
"rewards/accuracies": 0.625,
|
| 798 |
+
"rewards/chosen": NaN,
|
| 799 |
+
"rewards/margins": NaN,
|
| 800 |
+
"rewards/rejected": NaN,
|
| 801 |
+
"step": 440
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.9977827050997783,
|
| 805 |
+
"grad_norm": NaN,
|
| 806 |
+
"learning_rate": 6.230238391450653e-06,
|
| 807 |
+
"log_odds_chosen": NaN,
|
| 808 |
+
"log_odds_ratio": NaN,
|
| 809 |
+
"logits/chosen": -2.598021984100342,
|
| 810 |
+
"logits/rejected": -2.635017156600952,
|
| 811 |
+
"logps/chosen": NaN,
|
| 812 |
+
"logps/rejected": NaN,
|
| 813 |
+
"loss": 1.5232,
|
| 814 |
+
"nll_loss": 0.9541953802108765,
|
| 815 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 816 |
+
"rewards/chosen": NaN,
|
| 817 |
+
"rewards/margins": NaN,
|
| 818 |
+
"rewards/rejected": NaN,
|
| 819 |
+
"step": 450
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 1.0199556541019956,
|
| 823 |
+
"grad_norm": NaN,
|
| 824 |
+
"learning_rate": 6.150095915701193e-06,
|
| 825 |
+
"log_odds_chosen": NaN,
|
| 826 |
+
"log_odds_ratio": NaN,
|
| 827 |
+
"logits/chosen": -2.572929859161377,
|
| 828 |
+
"logits/rejected": -2.6330859661102295,
|
| 829 |
+
"logps/chosen": NaN,
|
| 830 |
+
"logps/rejected": NaN,
|
| 831 |
+
"loss": 1.4393,
|
| 832 |
+
"nll_loss": 0.9072533845901489,
|
| 833 |
+
"rewards/accuracies": 0.612500011920929,
|
| 834 |
+
"rewards/chosen": NaN,
|
| 835 |
+
"rewards/margins": NaN,
|
| 836 |
+
"rewards/rejected": NaN,
|
| 837 |
+
"step": 460
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 1.042128603104213,
|
| 841 |
+
"grad_norm": NaN,
|
| 842 |
+
"learning_rate": 6.06872070656429e-06,
|
| 843 |
+
"log_odds_chosen": NaN,
|
| 844 |
+
"log_odds_ratio": NaN,
|
| 845 |
+
"logits/chosen": -2.659424066543579,
|
| 846 |
+
"logits/rejected": -2.665045976638794,
|
| 847 |
+
"logps/chosen": NaN,
|
| 848 |
+
"logps/rejected": NaN,
|
| 849 |
+
"loss": 1.2597,
|
| 850 |
+
"nll_loss": 0.9222477674484253,
|
| 851 |
+
"rewards/accuracies": 0.512499988079071,
|
| 852 |
+
"rewards/chosen": NaN,
|
| 853 |
+
"rewards/margins": NaN,
|
| 854 |
+
"rewards/rejected": NaN,
|
| 855 |
+
"step": 470
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1.06430155210643,
|
| 859 |
+
"grad_norm": NaN,
|
| 860 |
+
"learning_rate": 5.986159419603766e-06,
|
| 861 |
+
"log_odds_chosen": NaN,
|
| 862 |
+
"log_odds_ratio": NaN,
|
| 863 |
+
"logits/chosen": -2.5539708137512207,
|
| 864 |
+
"logits/rejected": -2.578723430633545,
|
| 865 |
+
"logps/chosen": NaN,
|
| 866 |
+
"logps/rejected": NaN,
|
| 867 |
+
"loss": 1.5468,
|
| 868 |
+
"nll_loss": 0.9003454446792603,
|
| 869 |
+
"rewards/accuracies": 0.612500011920929,
|
| 870 |
+
"rewards/chosen": NaN,
|
| 871 |
+
"rewards/margins": NaN,
|
| 872 |
+
"rewards/rejected": NaN,
|
| 873 |
+
"step": 480
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.0864745011086474,
|
| 877 |
+
"grad_norm": NaN,
|
| 878 |
+
"learning_rate": 5.902459390407861e-06,
|
| 879 |
+
"log_odds_chosen": NaN,
|
| 880 |
+
"log_odds_ratio": NaN,
|
| 881 |
+
"logits/chosen": -2.5614101886749268,
|
| 882 |
+
"logits/rejected": -2.61173677444458,
|
| 883 |
+
"logps/chosen": NaN,
|
| 884 |
+
"logps/rejected": NaN,
|
| 885 |
+
"loss": 1.4411,
|
| 886 |
+
"nll_loss": 0.9182124137878418,
|
| 887 |
+
"rewards/accuracies": 0.574999988079071,
|
| 888 |
+
"rewards/chosen": NaN,
|
| 889 |
+
"rewards/margins": NaN,
|
| 890 |
+
"rewards/rejected": NaN,
|
| 891 |
+
"step": 490
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.1086474501108647,
|
| 895 |
+
"grad_norm": 0.7572781443595886,
|
| 896 |
+
"learning_rate": 5.817668607449911e-06,
|
| 897 |
+
"log_odds_chosen": NaN,
|
| 898 |
+
"log_odds_ratio": NaN,
|
| 899 |
+
"logits/chosen": -2.621619939804077,
|
| 900 |
+
"logits/rejected": -2.688084363937378,
|
| 901 |
+
"logps/chosen": NaN,
|
| 902 |
+
"logps/rejected": NaN,
|
| 903 |
+
"loss": 1.2006,
|
| 904 |
+
"nll_loss": 0.9091912508010864,
|
| 905 |
+
"rewards/accuracies": 0.574999988079071,
|
| 906 |
+
"rewards/chosen": NaN,
|
| 907 |
+
"rewards/margins": NaN,
|
| 908 |
+
"rewards/rejected": NaN,
|
| 909 |
+
"step": 500
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1.130820399113082,
|
| 913 |
+
"grad_norm": NaN,
|
| 914 |
+
"learning_rate": 5.731835684574692e-06,
|
| 915 |
+
"log_odds_chosen": NaN,
|
| 916 |
+
"log_odds_ratio": NaN,
|
| 917 |
+
"logits/chosen": -2.59326171875,
|
| 918 |
+
"logits/rejected": -2.6192498207092285,
|
| 919 |
+
"logps/chosen": NaN,
|
| 920 |
+
"logps/rejected": NaN,
|
| 921 |
+
"loss": 1.392,
|
| 922 |
+
"nll_loss": 0.8921745419502258,
|
| 923 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 924 |
+
"rewards/chosen": NaN,
|
| 925 |
+
"rewards/margins": NaN,
|
| 926 |
+
"rewards/rejected": NaN,
|
| 927 |
+
"step": 510
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 1.1529933481152994,
|
| 931 |
+
"grad_norm": NaN,
|
| 932 |
+
"learning_rate": 5.645009833126218e-06,
|
| 933 |
+
"log_odds_chosen": NaN,
|
| 934 |
+
"log_odds_ratio": NaN,
|
| 935 |
+
"logits/chosen": -2.570634365081787,
|
| 936 |
+
"logits/rejected": -2.5929832458496094,
|
| 937 |
+
"logps/chosen": NaN,
|
| 938 |
+
"logps/rejected": NaN,
|
| 939 |
+
"loss": 1.5643,
|
| 940 |
+
"nll_loss": 0.9609503746032715,
|
| 941 |
+
"rewards/accuracies": 0.612500011920929,
|
| 942 |
+
"rewards/chosen": NaN,
|
| 943 |
+
"rewards/margins": NaN,
|
| 944 |
+
"rewards/rejected": NaN,
|
| 945 |
+
"step": 520
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 1.1751662971175167,
|
| 949 |
+
"grad_norm": NaN,
|
| 950 |
+
"learning_rate": 5.557240833732967e-06,
|
| 951 |
+
"log_odds_chosen": NaN,
|
| 952 |
+
"log_odds_ratio": NaN,
|
| 953 |
+
"logits/chosen": -2.5438590049743652,
|
| 954 |
+
"logits/rejected": -2.586803913116455,
|
| 955 |
+
"logps/chosen": NaN,
|
| 956 |
+
"logps/rejected": NaN,
|
| 957 |
+
"loss": 1.4816,
|
| 958 |
+
"nll_loss": 0.8677853345870972,
|
| 959 |
+
"rewards/accuracies": 0.5625,
|
| 960 |
+
"rewards/chosen": NaN,
|
| 961 |
+
"rewards/margins": NaN,
|
| 962 |
+
"rewards/rejected": NaN,
|
| 963 |
+
"step": 530
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 1.1973392461197339,
|
| 967 |
+
"grad_norm": NaN,
|
| 968 |
+
"learning_rate": 5.46857900776672e-06,
|
| 969 |
+
"log_odds_chosen": NaN,
|
| 970 |
+
"log_odds_ratio": NaN,
|
| 971 |
+
"logits/chosen": -2.5750033855438232,
|
| 972 |
+
"logits/rejected": -2.618818521499634,
|
| 973 |
+
"logps/chosen": NaN,
|
| 974 |
+
"logps/rejected": NaN,
|
| 975 |
+
"loss": 1.3517,
|
| 976 |
+
"nll_loss": 0.8420497179031372,
|
| 977 |
+
"rewards/accuracies": 0.5625,
|
| 978 |
+
"rewards/chosen": NaN,
|
| 979 |
+
"rewards/margins": NaN,
|
| 980 |
+
"rewards/rejected": NaN,
|
| 981 |
+
"step": 540
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1.2195121951219512,
|
| 985 |
+
"grad_norm": NaN,
|
| 986 |
+
"learning_rate": 5.3790751884913605e-06,
|
| 987 |
+
"log_odds_chosen": NaN,
|
| 988 |
+
"log_odds_ratio": NaN,
|
| 989 |
+
"logits/chosen": -2.576446771621704,
|
| 990 |
+
"logits/rejected": -2.5843629837036133,
|
| 991 |
+
"logps/chosen": NaN,
|
| 992 |
+
"logps/rejected": NaN,
|
| 993 |
+
"loss": 1.5026,
|
| 994 |
+
"nll_loss": 0.8621179461479187,
|
| 995 |
+
"rewards/accuracies": 0.5625,
|
| 996 |
+
"rewards/chosen": NaN,
|
| 997 |
+
"rewards/margins": NaN,
|
| 998 |
+
"rewards/rejected": NaN,
|
| 999 |
+
"step": 550
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 1.2416851441241685,
|
| 1003 |
+
"grad_norm": NaN,
|
| 1004 |
+
"learning_rate": 5.288780691918196e-06,
|
| 1005 |
+
"log_odds_chosen": NaN,
|
| 1006 |
+
"log_odds_ratio": NaN,
|
| 1007 |
+
"logits/chosen": -2.569793939590454,
|
| 1008 |
+
"logits/rejected": -2.656283140182495,
|
| 1009 |
+
"logps/chosen": NaN,
|
| 1010 |
+
"logps/rejected": NaN,
|
| 1011 |
+
"loss": 1.3906,
|
| 1012 |
+
"nll_loss": 1.0533077716827393,
|
| 1013 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1014 |
+
"rewards/chosen": NaN,
|
| 1015 |
+
"rewards/margins": NaN,
|
| 1016 |
+
"rewards/rejected": NaN,
|
| 1017 |
+
"step": 560
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 1.2638580931263859,
|
| 1021 |
+
"grad_norm": NaN,
|
| 1022 |
+
"learning_rate": 5.197747287384502e-06,
|
| 1023 |
+
"log_odds_chosen": NaN,
|
| 1024 |
+
"log_odds_ratio": NaN,
|
| 1025 |
+
"logits/chosen": -2.586745500564575,
|
| 1026 |
+
"logits/rejected": -2.6093835830688477,
|
| 1027 |
+
"logps/chosen": NaN,
|
| 1028 |
+
"logps/rejected": NaN,
|
| 1029 |
+
"loss": 1.5754,
|
| 1030 |
+
"nll_loss": 0.9478577375411987,
|
| 1031 |
+
"rewards/accuracies": 0.5,
|
| 1032 |
+
"rewards/chosen": NaN,
|
| 1033 |
+
"rewards/margins": NaN,
|
| 1034 |
+
"rewards/rejected": NaN,
|
| 1035 |
+
"step": 570
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 1.2860310421286032,
|
| 1039 |
+
"grad_norm": NaN,
|
| 1040 |
+
"learning_rate": 5.106027167872141e-06,
|
| 1041 |
+
"log_odds_chosen": NaN,
|
| 1042 |
+
"log_odds_ratio": NaN,
|
| 1043 |
+
"logits/chosen": -2.5553150177001953,
|
| 1044 |
+
"logits/rejected": -2.610725164413452,
|
| 1045 |
+
"logps/chosen": NaN,
|
| 1046 |
+
"logps/rejected": NaN,
|
| 1047 |
+
"loss": 1.4337,
|
| 1048 |
+
"nll_loss": 0.8642798662185669,
|
| 1049 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1050 |
+
"rewards/chosen": NaN,
|
| 1051 |
+
"rewards/margins": NaN,
|
| 1052 |
+
"rewards/rejected": NaN,
|
| 1053 |
+
"step": 580
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 1.3082039911308203,
|
| 1057 |
+
"grad_norm": NaN,
|
| 1058 |
+
"learning_rate": 5.013672920083319e-06,
|
| 1059 |
+
"log_odds_chosen": NaN,
|
| 1060 |
+
"log_odds_ratio": NaN,
|
| 1061 |
+
"logits/chosen": -2.579271078109741,
|
| 1062 |
+
"logits/rejected": -2.635814666748047,
|
| 1063 |
+
"logps/chosen": NaN,
|
| 1064 |
+
"logps/rejected": NaN,
|
| 1065 |
+
"loss": 1.2762,
|
| 1066 |
+
"nll_loss": 0.9140083193778992,
|
| 1067 |
+
"rewards/accuracies": 0.550000011920929,
|
| 1068 |
+
"rewards/chosen": NaN,
|
| 1069 |
+
"rewards/margins": NaN,
|
| 1070 |
+
"rewards/rejected": NaN,
|
| 1071 |
+
"step": 590
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.3303769401330376,
|
| 1075 |
+
"grad_norm": NaN,
|
| 1076 |
+
"learning_rate": 4.920737494290572e-06,
|
| 1077 |
+
"log_odds_chosen": NaN,
|
| 1078 |
+
"log_odds_ratio": NaN,
|
| 1079 |
+
"logits/chosen": -2.5220165252685547,
|
| 1080 |
+
"logits/rejected": -2.551558256149292,
|
| 1081 |
+
"logps/chosen": NaN,
|
| 1082 |
+
"logps/rejected": NaN,
|
| 1083 |
+
"loss": 1.6071,
|
| 1084 |
+
"nll_loss": 0.9066599011421204,
|
| 1085 |
+
"rewards/accuracies": 0.512499988079071,
|
| 1086 |
+
"rewards/chosen": NaN,
|
| 1087 |
+
"rewards/margins": NaN,
|
| 1088 |
+
"rewards/rejected": NaN,
|
| 1089 |
+
"step": 600
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 1.352549889135255,
|
| 1093 |
+
"grad_norm": NaN,
|
| 1094 |
+
"learning_rate": 4.827274173978333e-06,
|
| 1095 |
+
"log_odds_chosen": NaN,
|
| 1096 |
+
"log_odds_ratio": NaN,
|
| 1097 |
+
"logits/chosen": -2.560403347015381,
|
| 1098 |
+
"logits/rejected": -2.639423131942749,
|
| 1099 |
+
"logps/chosen": NaN,
|
| 1100 |
+
"logps/rejected": NaN,
|
| 1101 |
+
"loss": 1.2709,
|
| 1102 |
+
"nll_loss": 0.8913120031356812,
|
| 1103 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1104 |
+
"rewards/chosen": NaN,
|
| 1105 |
+
"rewards/margins": NaN,
|
| 1106 |
+
"rewards/rejected": NaN,
|
| 1107 |
+
"step": 610
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 1.3747228381374723,
|
| 1111 |
+
"grad_norm": NaN,
|
| 1112 |
+
"learning_rate": 4.733336545293438e-06,
|
| 1113 |
+
"log_odds_chosen": NaN,
|
| 1114 |
+
"log_odds_ratio": NaN,
|
| 1115 |
+
"logits/chosen": -2.616939067840576,
|
| 1116 |
+
"logits/rejected": -2.650655746459961,
|
| 1117 |
+
"logps/chosen": NaN,
|
| 1118 |
+
"logps/rejected": NaN,
|
| 1119 |
+
"loss": 1.4557,
|
| 1120 |
+
"nll_loss": 0.9315293431282043,
|
| 1121 |
+
"rewards/accuracies": 0.550000011920929,
|
| 1122 |
+
"rewards/chosen": NaN,
|
| 1123 |
+
"rewards/margins": NaN,
|
| 1124 |
+
"rewards/rejected": NaN,
|
| 1125 |
+
"step": 620
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 1.3968957871396896,
|
| 1129 |
+
"grad_norm": 0.6212561130523682,
|
| 1130 |
+
"learning_rate": 4.638978466322108e-06,
|
| 1131 |
+
"log_odds_chosen": NaN,
|
| 1132 |
+
"log_odds_ratio": NaN,
|
| 1133 |
+
"logits/chosen": -2.531559944152832,
|
| 1134 |
+
"logits/rejected": -2.5307424068450928,
|
| 1135 |
+
"logps/chosen": NaN,
|
| 1136 |
+
"logps/rejected": NaN,
|
| 1137 |
+
"loss": 0.9584,
|
| 1138 |
+
"nll_loss": 0.9440839886665344,
|
| 1139 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1140 |
+
"rewards/chosen": NaN,
|
| 1141 |
+
"rewards/margins": NaN,
|
| 1142 |
+
"rewards/rejected": NaN,
|
| 1143 |
+
"step": 630
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.4190687361419068,
|
| 1147 |
+
"grad_norm": NaN,
|
| 1148 |
+
"learning_rate": 4.5442540362110285e-06,
|
| 1149 |
+
"log_odds_chosen": NaN,
|
| 1150 |
+
"log_odds_ratio": NaN,
|
| 1151 |
+
"logits/chosen": -2.546278476715088,
|
| 1152 |
+
"logits/rejected": -2.5663809776306152,
|
| 1153 |
+
"logps/chosen": NaN,
|
| 1154 |
+
"logps/rejected": NaN,
|
| 1155 |
+
"loss": 1.4305,
|
| 1156 |
+
"nll_loss": 0.8402501940727234,
|
| 1157 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1158 |
+
"rewards/chosen": NaN,
|
| 1159 |
+
"rewards/margins": NaN,
|
| 1160 |
+
"rewards/rejected": NaN,
|
| 1161 |
+
"step": 640
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 1.441241685144124,
|
| 1165 |
+
"grad_norm": 0.784209132194519,
|
| 1166 |
+
"learning_rate": 4.449217564150197e-06,
|
| 1167 |
+
"log_odds_chosen": NaN,
|
| 1168 |
+
"log_odds_ratio": NaN,
|
| 1169 |
+
"logits/chosen": -2.547091484069824,
|
| 1170 |
+
"logits/rejected": -2.6290016174316406,
|
| 1171 |
+
"logps/chosen": NaN,
|
| 1172 |
+
"logps/rejected": NaN,
|
| 1173 |
+
"loss": 1.1191,
|
| 1174 |
+
"nll_loss": 0.9092384576797485,
|
| 1175 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1176 |
+
"rewards/chosen": NaN,
|
| 1177 |
+
"rewards/margins": NaN,
|
| 1178 |
+
"rewards/rejected": NaN,
|
| 1179 |
+
"step": 650
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.4634146341463414,
|
| 1183 |
+
"grad_norm": NaN,
|
| 1184 |
+
"learning_rate": 4.353923538235369e-06,
|
| 1185 |
+
"log_odds_chosen": NaN,
|
| 1186 |
+
"log_odds_ratio": NaN,
|
| 1187 |
+
"logits/chosen": -2.60884952545166,
|
| 1188 |
+
"logits/rejected": -2.640436887741089,
|
| 1189 |
+
"logps/chosen": NaN,
|
| 1190 |
+
"logps/rejected": NaN,
|
| 1191 |
+
"loss": 1.2808,
|
| 1192 |
+
"nll_loss": 0.872983455657959,
|
| 1193 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1194 |
+
"rewards/chosen": NaN,
|
| 1195 |
+
"rewards/margins": NaN,
|
| 1196 |
+
"rewards/rejected": NaN,
|
| 1197 |
+
"step": 660
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 1.4855875831485588,
|
| 1201 |
+
"grad_norm": NaN,
|
| 1202 |
+
"learning_rate": 4.2584265942279114e-06,
|
| 1203 |
+
"log_odds_chosen": NaN,
|
| 1204 |
+
"log_odds_ratio": NaN,
|
| 1205 |
+
"logits/chosen": -2.5404248237609863,
|
| 1206 |
+
"logits/rejected": -2.5652973651885986,
|
| 1207 |
+
"logps/chosen": NaN,
|
| 1208 |
+
"logps/rejected": NaN,
|
| 1209 |
+
"loss": 1.5307,
|
| 1210 |
+
"nll_loss": 0.8896579742431641,
|
| 1211 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1212 |
+
"rewards/chosen": NaN,
|
| 1213 |
+
"rewards/margins": NaN,
|
| 1214 |
+
"rewards/rejected": NaN,
|
| 1215 |
+
"step": 670
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 1.507760532150776,
|
| 1219 |
+
"grad_norm": NaN,
|
| 1220 |
+
"learning_rate": 4.162781484230005e-06,
|
| 1221 |
+
"log_odds_chosen": NaN,
|
| 1222 |
+
"log_odds_ratio": NaN,
|
| 1223 |
+
"logits/chosen": -2.554001569747925,
|
| 1224 |
+
"logits/rejected": -2.5924289226531982,
|
| 1225 |
+
"logps/chosen": NaN,
|
| 1226 |
+
"logps/rejected": NaN,
|
| 1227 |
+
"loss": 1.5124,
|
| 1228 |
+
"nll_loss": 0.9706124067306519,
|
| 1229 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1230 |
+
"rewards/chosen": NaN,
|
| 1231 |
+
"rewards/margins": NaN,
|
| 1232 |
+
"rewards/rejected": NaN,
|
| 1233 |
+
"step": 680
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 1.5299334811529932,
|
| 1237 |
+
"grad_norm": NaN,
|
| 1238 |
+
"learning_rate": 4.067043045293142e-06,
|
| 1239 |
+
"log_odds_chosen": NaN,
|
| 1240 |
+
"log_odds_ratio": NaN,
|
| 1241 |
+
"logits/chosen": -2.6452507972717285,
|
| 1242 |
+
"logits/rejected": -2.6547276973724365,
|
| 1243 |
+
"logps/chosen": NaN,
|
| 1244 |
+
"logps/rejected": NaN,
|
| 1245 |
+
"loss": 1.3232,
|
| 1246 |
+
"nll_loss": 0.9081094861030579,
|
| 1247 |
+
"rewards/accuracies": 0.5625,
|
| 1248 |
+
"rewards/chosen": NaN,
|
| 1249 |
+
"rewards/margins": NaN,
|
| 1250 |
+
"rewards/rejected": NaN,
|
| 1251 |
+
"step": 690
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 1.5521064301552108,
|
| 1255 |
+
"grad_norm": NaN,
|
| 1256 |
+
"learning_rate": 3.971266167977914e-06,
|
| 1257 |
+
"log_odds_chosen": NaN,
|
| 1258 |
+
"log_odds_ratio": NaN,
|
| 1259 |
+
"logits/chosen": -2.5739452838897705,
|
| 1260 |
+
"logits/rejected": -2.6191649436950684,
|
| 1261 |
+
"logps/chosen": NaN,
|
| 1262 |
+
"logps/rejected": NaN,
|
| 1263 |
+
"loss": 1.1834,
|
| 1264 |
+
"nll_loss": 0.8611727952957153,
|
| 1265 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1266 |
+
"rewards/chosen": NaN,
|
| 1267 |
+
"rewards/margins": NaN,
|
| 1268 |
+
"rewards/rejected": NaN,
|
| 1269 |
+
"step": 700
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.5742793791574279,
|
| 1273 |
+
"grad_norm": 0.7607054710388184,
|
| 1274 |
+
"learning_rate": 3.875505764883128e-06,
|
| 1275 |
+
"log_odds_chosen": NaN,
|
| 1276 |
+
"log_odds_ratio": NaN,
|
| 1277 |
+
"logits/chosen": -2.5171356201171875,
|
| 1278 |
+
"logits/rejected": -2.5558667182922363,
|
| 1279 |
+
"logps/chosen": NaN,
|
| 1280 |
+
"logps/rejected": NaN,
|
| 1281 |
+
"loss": 1.3944,
|
| 1282 |
+
"nll_loss": 0.9226005673408508,
|
| 1283 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1284 |
+
"rewards/chosen": NaN,
|
| 1285 |
+
"rewards/margins": NaN,
|
| 1286 |
+
"rewards/rejected": NaN,
|
| 1287 |
+
"step": 710
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 1.5964523281596452,
|
| 1291 |
+
"grad_norm": NaN,
|
| 1292 |
+
"learning_rate": 3.7798167391622746e-06,
|
| 1293 |
+
"log_odds_chosen": NaN,
|
| 1294 |
+
"log_odds_ratio": NaN,
|
| 1295 |
+
"logits/chosen": -2.4788334369659424,
|
| 1296 |
+
"logits/rejected": -2.556370735168457,
|
| 1297 |
+
"logps/chosen": NaN,
|
| 1298 |
+
"logps/rejected": NaN,
|
| 1299 |
+
"loss": 1.2499,
|
| 1300 |
+
"nll_loss": 0.8890296816825867,
|
| 1301 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1302 |
+
"rewards/chosen": NaN,
|
| 1303 |
+
"rewards/margins": NaN,
|
| 1304 |
+
"rewards/rejected": NaN,
|
| 1305 |
+
"step": 720
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.6186252771618626,
|
| 1309 |
+
"grad_norm": 0.5698145031929016,
|
| 1310 |
+
"learning_rate": 3.684253953045438e-06,
|
| 1311 |
+
"log_odds_chosen": NaN,
|
| 1312 |
+
"log_odds_ratio": NaN,
|
| 1313 |
+
"logits/chosen": -2.59308123588562,
|
| 1314 |
+
"logits/rejected": -2.5706467628479004,
|
| 1315 |
+
"logps/chosen": NaN,
|
| 1316 |
+
"logps/rejected": NaN,
|
| 1317 |
+
"loss": 1.3174,
|
| 1318 |
+
"nll_loss": 0.8769053220748901,
|
| 1319 |
+
"rewards/accuracies": 0.625,
|
| 1320 |
+
"rewards/chosen": NaN,
|
| 1321 |
+
"rewards/margins": NaN,
|
| 1322 |
+
"rewards/rejected": NaN,
|
| 1323 |
+
"step": 730
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 1.6407982261640797,
|
| 1327 |
+
"grad_norm": NaN,
|
| 1328 |
+
"learning_rate": 3.588872196384632e-06,
|
| 1329 |
+
"log_odds_chosen": NaN,
|
| 1330 |
+
"log_odds_ratio": NaN,
|
| 1331 |
+
"logits/chosen": -2.515533924102783,
|
| 1332 |
+
"logits/rejected": -2.539431571960449,
|
| 1333 |
+
"logps/chosen": NaN,
|
| 1334 |
+
"logps/rejected": NaN,
|
| 1335 |
+
"loss": 1.4826,
|
| 1336 |
+
"nll_loss": 0.9430069923400879,
|
| 1337 |
+
"rewards/accuracies": 0.5625,
|
| 1338 |
+
"rewards/chosen": NaN,
|
| 1339 |
+
"rewards/margins": NaN,
|
| 1340 |
+
"rewards/rejected": NaN,
|
| 1341 |
+
"step": 740
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.6629711751662972,
|
| 1345 |
+
"grad_norm": 1.2568813562393188,
|
| 1346 |
+
"learning_rate": 3.493726155240664e-06,
|
| 1347 |
+
"log_odds_chosen": NaN,
|
| 1348 |
+
"log_odds_ratio": NaN,
|
| 1349 |
+
"logits/chosen": -2.6210389137268066,
|
| 1350 |
+
"logits/rejected": -2.6431758403778076,
|
| 1351 |
+
"logps/chosen": NaN,
|
| 1352 |
+
"logps/rejected": NaN,
|
| 1353 |
+
"loss": 1.2068,
|
| 1354 |
+
"nll_loss": 0.9099509119987488,
|
| 1355 |
+
"rewards/accuracies": 0.5625,
|
| 1356 |
+
"rewards/chosen": NaN,
|
| 1357 |
+
"rewards/margins": NaN,
|
| 1358 |
+
"rewards/rejected": NaN,
|
| 1359 |
+
"step": 750
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.6851441241685143,
|
| 1363 |
+
"grad_norm": NaN,
|
| 1364 |
+
"learning_rate": 3.3988703805294946e-06,
|
| 1365 |
+
"log_odds_chosen": NaN,
|
| 1366 |
+
"log_odds_ratio": NaN,
|
| 1367 |
+
"logits/chosen": -2.572796106338501,
|
| 1368 |
+
"logits/rejected": -2.6425790786743164,
|
| 1369 |
+
"logps/chosen": NaN,
|
| 1370 |
+
"logps/rejected": NaN,
|
| 1371 |
+
"loss": 1.3102,
|
| 1372 |
+
"nll_loss": 0.9549806714057922,
|
| 1373 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1374 |
+
"rewards/chosen": NaN,
|
| 1375 |
+
"rewards/margins": NaN,
|
| 1376 |
+
"rewards/rejected": NaN,
|
| 1377 |
+
"step": 760
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.7073170731707317,
|
| 1381 |
+
"grad_norm": 0.7996054887771606,
|
| 1382 |
+
"learning_rate": 3.3043592567460748e-06,
|
| 1383 |
+
"log_odds_chosen": NaN,
|
| 1384 |
+
"log_odds_ratio": NaN,
|
| 1385 |
+
"logits/chosen": -2.5201961994171143,
|
| 1386 |
+
"logits/rejected": -2.552128553390503,
|
| 1387 |
+
"logps/chosen": NaN,
|
| 1388 |
+
"logps/rejected": NaN,
|
| 1389 |
+
"loss": 1.5766,
|
| 1390 |
+
"nll_loss": 0.9187144041061401,
|
| 1391 |
+
"rewards/accuracies": 0.625,
|
| 1392 |
+
"rewards/chosen": NaN,
|
| 1393 |
+
"rewards/margins": NaN,
|
| 1394 |
+
"rewards/rejected": NaN,
|
| 1395 |
+
"step": 770
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.729490022172949,
|
| 1399 |
+
"grad_norm": NaN,
|
| 1400 |
+
"learning_rate": 3.2102469707836174e-06,
|
| 1401 |
+
"log_odds_chosen": NaN,
|
| 1402 |
+
"log_odds_ratio": NaN,
|
| 1403 |
+
"logits/chosen": -2.5832736492156982,
|
| 1404 |
+
"logits/rejected": -2.6100733280181885,
|
| 1405 |
+
"logps/chosen": NaN,
|
| 1406 |
+
"logps/rejected": NaN,
|
| 1407 |
+
"loss": 1.1691,
|
| 1408 |
+
"nll_loss": 0.8904333114624023,
|
| 1409 |
+
"rewards/accuracies": 0.5625,
|
| 1410 |
+
"rewards/chosen": NaN,
|
| 1411 |
+
"rewards/margins": NaN,
|
| 1412 |
+
"rewards/rejected": NaN,
|
| 1413 |
+
"step": 780
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.7516629711751663,
|
| 1417 |
+
"grad_norm": NaN,
|
| 1418 |
+
"learning_rate": 3.1165874808661342e-06,
|
| 1419 |
+
"log_odds_chosen": NaN,
|
| 1420 |
+
"log_odds_ratio": NaN,
|
| 1421 |
+
"logits/chosen": -2.599666118621826,
|
| 1422 |
+
"logits/rejected": -2.62807297706604,
|
| 1423 |
+
"logps/chosen": NaN,
|
| 1424 |
+
"logps/rejected": NaN,
|
| 1425 |
+
"loss": 1.4813,
|
| 1426 |
+
"nll_loss": 0.8979522585868835,
|
| 1427 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1428 |
+
"rewards/chosen": NaN,
|
| 1429 |
+
"rewards/margins": NaN,
|
| 1430 |
+
"rewards/rejected": NaN,
|
| 1431 |
+
"step": 790
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 1.7738359201773837,
|
| 1435 |
+
"grad_norm": NaN,
|
| 1436 |
+
"learning_rate": 3.0234344856121086e-06,
|
| 1437 |
+
"log_odds_chosen": NaN,
|
| 1438 |
+
"log_odds_ratio": NaN,
|
| 1439 |
+
"logits/chosen": -2.6034371852874756,
|
| 1440 |
+
"logits/rejected": -2.6399123668670654,
|
| 1441 |
+
"logps/chosen": NaN,
|
| 1442 |
+
"logps/rejected": NaN,
|
| 1443 |
+
"loss": 1.2004,
|
| 1444 |
+
"nll_loss": 0.9016131162643433,
|
| 1445 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1446 |
+
"rewards/chosen": NaN,
|
| 1447 |
+
"rewards/margins": NaN,
|
| 1448 |
+
"rewards/rejected": NaN,
|
| 1449 |
+
"step": 800
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.7960088691796008,
|
| 1453 |
+
"grad_norm": NaN,
|
| 1454 |
+
"learning_rate": 2.9308413932469805e-06,
|
| 1455 |
+
"log_odds_chosen": NaN,
|
| 1456 |
+
"log_odds_ratio": NaN,
|
| 1457 |
+
"logits/chosen": -2.5875086784362793,
|
| 1458 |
+
"logits/rejected": -2.5995872020721436,
|
| 1459 |
+
"logps/chosen": NaN,
|
| 1460 |
+
"logps/rejected": NaN,
|
| 1461 |
+
"loss": 1.2187,
|
| 1462 |
+
"nll_loss": 0.895309567451477,
|
| 1463 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1464 |
+
"rewards/chosen": NaN,
|
| 1465 |
+
"rewards/margins": NaN,
|
| 1466 |
+
"rewards/rejected": NaN,
|
| 1467 |
+
"step": 810
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 1.8181818181818183,
|
| 1471 |
+
"grad_norm": NaN,
|
| 1472 |
+
"learning_rate": 2.8388612909821512e-06,
|
| 1473 |
+
"log_odds_chosen": NaN,
|
| 1474 |
+
"log_odds_ratio": NaN,
|
| 1475 |
+
"logits/chosen": -2.545532703399658,
|
| 1476 |
+
"logits/rejected": -2.5655813217163086,
|
| 1477 |
+
"logps/chosen": NaN,
|
| 1478 |
+
"logps/rejected": NaN,
|
| 1479 |
+
"loss": 1.4789,
|
| 1480 |
+
"nll_loss": 0.8789796829223633,
|
| 1481 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1482 |
+
"rewards/chosen": NaN,
|
| 1483 |
+
"rewards/margins": NaN,
|
| 1484 |
+
"rewards/rejected": NaN,
|
| 1485 |
+
"step": 820
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.8403547671840355,
|
| 1489 |
+
"grad_norm": NaN,
|
| 1490 |
+
"learning_rate": 2.7475469145780162e-06,
|
| 1491 |
+
"log_odds_chosen": NaN,
|
| 1492 |
+
"log_odds_ratio": NaN,
|
| 1493 |
+
"logits/chosen": -2.634237766265869,
|
| 1494 |
+
"logits/rejected": -2.6819849014282227,
|
| 1495 |
+
"logps/chosen": NaN,
|
| 1496 |
+
"logps/rejected": NaN,
|
| 1497 |
+
"loss": 2.0414,
|
| 1498 |
+
"nll_loss": 0.8710411190986633,
|
| 1499 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1500 |
+
"rewards/chosen": NaN,
|
| 1501 |
+
"rewards/margins": NaN,
|
| 1502 |
+
"rewards/rejected": NaN,
|
| 1503 |
+
"step": 830
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 1.8625277161862528,
|
| 1507 |
+
"grad_norm": NaN,
|
| 1508 |
+
"learning_rate": 2.6569506181085155e-06,
|
| 1509 |
+
"log_odds_chosen": NaN,
|
| 1510 |
+
"log_odds_ratio": NaN,
|
| 1511 |
+
"logits/chosen": -2.569899559020996,
|
| 1512 |
+
"logits/rejected": -2.627354145050049,
|
| 1513 |
+
"logps/chosen": NaN,
|
| 1514 |
+
"logps/rejected": NaN,
|
| 1515 |
+
"loss": 1.1485,
|
| 1516 |
+
"nll_loss": 0.8969793319702148,
|
| 1517 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1518 |
+
"rewards/chosen": NaN,
|
| 1519 |
+
"rewards/margins": NaN,
|
| 1520 |
+
"rewards/rejected": NaN,
|
| 1521 |
+
"step": 840
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.8847006651884701,
|
| 1525 |
+
"grad_norm": NaN,
|
| 1526 |
+
"learning_rate": 2.5671243439445098e-06,
|
| 1527 |
+
"log_odds_chosen": NaN,
|
| 1528 |
+
"log_odds_ratio": NaN,
|
| 1529 |
+
"logits/chosen": -2.5224175453186035,
|
| 1530 |
+
"logits/rejected": -2.521833896636963,
|
| 1531 |
+
"logps/chosen": NaN,
|
| 1532 |
+
"logps/rejected": NaN,
|
| 1533 |
+
"loss": 1.5147,
|
| 1534 |
+
"nll_loss": 1.0038148164749146,
|
| 1535 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1536 |
+
"rewards/chosen": NaN,
|
| 1537 |
+
"rewards/margins": NaN,
|
| 1538 |
+
"rewards/rejected": NaN,
|
| 1539 |
+
"step": 850
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 1.9068736141906872,
|
| 1543 |
+
"grad_norm": NaN,
|
| 1544 |
+
"learning_rate": 2.4781195929731997e-06,
|
| 1545 |
+
"log_odds_chosen": NaN,
|
| 1546 |
+
"log_odds_ratio": NaN,
|
| 1547 |
+
"logits/chosen": -2.5353684425354004,
|
| 1548 |
+
"logits/rejected": -2.554072141647339,
|
| 1549 |
+
"logps/chosen": NaN,
|
| 1550 |
+
"logps/rejected": NaN,
|
| 1551 |
+
"loss": 1.2603,
|
| 1552 |
+
"nll_loss": 0.8757489323616028,
|
| 1553 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1554 |
+
"rewards/chosen": NaN,
|
| 1555 |
+
"rewards/margins": NaN,
|
| 1556 |
+
"rewards/rejected": NaN,
|
| 1557 |
+
"step": 860
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 1.9290465631929048,
|
| 1561 |
+
"grad_norm": NaN,
|
| 1562 |
+
"learning_rate": 2.3899873950706803e-06,
|
| 1563 |
+
"log_odds_chosen": NaN,
|
| 1564 |
+
"log_odds_ratio": NaN,
|
| 1565 |
+
"logits/chosen": -2.6149215698242188,
|
| 1566 |
+
"logits/rejected": -2.663322687149048,
|
| 1567 |
+
"logps/chosen": NaN,
|
| 1568 |
+
"logps/rejected": NaN,
|
| 1569 |
+
"loss": 1.3348,
|
| 1570 |
+
"nll_loss": 0.8890805244445801,
|
| 1571 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1572 |
+
"rewards/chosen": NaN,
|
| 1573 |
+
"rewards/margins": NaN,
|
| 1574 |
+
"rewards/rejected": NaN,
|
| 1575 |
+
"step": 870
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 1.951219512195122,
|
| 1579 |
+
"grad_norm": NaN,
|
| 1580 |
+
"learning_rate": 2.3027782798445205e-06,
|
| 1581 |
+
"log_odds_chosen": NaN,
|
| 1582 |
+
"log_odds_ratio": NaN,
|
| 1583 |
+
"logits/chosen": -2.5764241218566895,
|
| 1584 |
+
"logits/rejected": -2.5978610515594482,
|
| 1585 |
+
"logps/chosen": NaN,
|
| 1586 |
+
"logps/rejected": NaN,
|
| 1587 |
+
"loss": 1.255,
|
| 1588 |
+
"nll_loss": 0.9303677678108215,
|
| 1589 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 1590 |
+
"rewards/chosen": NaN,
|
| 1591 |
+
"rewards/margins": NaN,
|
| 1592 |
+
"rewards/rejected": NaN,
|
| 1593 |
+
"step": 880
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 1.9733924611973392,
|
| 1597 |
+
"grad_norm": NaN,
|
| 1598 |
+
"learning_rate": 2.216542247663192e-06,
|
| 1599 |
+
"log_odds_chosen": NaN,
|
| 1600 |
+
"log_odds_ratio": NaN,
|
| 1601 |
+
"logits/chosen": -2.5386908054351807,
|
| 1602 |
+
"logits/rejected": -2.5890002250671387,
|
| 1603 |
+
"logps/chosen": NaN,
|
| 1604 |
+
"logps/rejected": NaN,
|
| 1605 |
+
"loss": 1.228,
|
| 1606 |
+
"nll_loss": 0.880374550819397,
|
| 1607 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1608 |
+
"rewards/chosen": NaN,
|
| 1609 |
+
"rewards/margins": NaN,
|
| 1610 |
+
"rewards/rejected": NaN,
|
| 1611 |
+
"step": 890
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.9955654101995566,
|
| 1615 |
+
"grad_norm": NaN,
|
| 1616 |
+
"learning_rate": 2.1313287409889075e-06,
|
| 1617 |
+
"log_odds_chosen": NaN,
|
| 1618 |
+
"log_odds_ratio": NaN,
|
| 1619 |
+
"logits/chosen": -2.6515865325927734,
|
| 1620 |
+
"logits/rejected": -2.698002338409424,
|
| 1621 |
+
"logps/chosen": NaN,
|
| 1622 |
+
"logps/rejected": NaN,
|
| 1623 |
+
"loss": 1.4501,
|
| 1624 |
+
"nll_loss": 0.8617309331893921,
|
| 1625 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1626 |
+
"rewards/chosen": NaN,
|
| 1627 |
+
"rewards/margins": NaN,
|
| 1628 |
+
"rewards/rejected": NaN,
|
| 1629 |
+
"step": 900
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 2.0177383592017737,
|
| 1633 |
+
"grad_norm": NaN,
|
| 1634 |
+
"learning_rate": 2.0471866160303494e-06,
|
| 1635 |
+
"log_odds_chosen": NaN,
|
| 1636 |
+
"log_odds_ratio": NaN,
|
| 1637 |
+
"logits/chosen": -2.576596975326538,
|
| 1638 |
+
"logits/rejected": -2.634167194366455,
|
| 1639 |
+
"logps/chosen": NaN,
|
| 1640 |
+
"logps/rejected": NaN,
|
| 1641 |
+
"loss": 1.658,
|
| 1642 |
+
"nll_loss": 0.9685592651367188,
|
| 1643 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1644 |
+
"rewards/chosen": NaN,
|
| 1645 |
+
"rewards/margins": NaN,
|
| 1646 |
+
"rewards/rejected": NaN,
|
| 1647 |
+
"step": 910
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 2.0399113082039912,
|
| 1651 |
+
"grad_norm": NaN,
|
| 1652 |
+
"learning_rate": 1.9641641147314996e-06,
|
| 1653 |
+
"log_odds_chosen": NaN,
|
| 1654 |
+
"log_odds_ratio": NaN,
|
| 1655 |
+
"logits/chosen": -2.556303024291992,
|
| 1656 |
+
"logits/rejected": -2.6210145950317383,
|
| 1657 |
+
"logps/chosen": NaN,
|
| 1658 |
+
"logps/rejected": NaN,
|
| 1659 |
+
"loss": 1.6006,
|
| 1660 |
+
"nll_loss": 0.9573662877082825,
|
| 1661 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1662 |
+
"rewards/chosen": NaN,
|
| 1663 |
+
"rewards/margins": NaN,
|
| 1664 |
+
"rewards/rejected": NaN,
|
| 1665 |
+
"step": 920
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 2.0620842572062084,
|
| 1669 |
+
"grad_norm": NaN,
|
| 1670 |
+
"learning_rate": 1.88230883711267e-06,
|
| 1671 |
+
"log_odds_chosen": NaN,
|
| 1672 |
+
"log_odds_ratio": NaN,
|
| 1673 |
+
"logits/chosen": -2.6028056144714355,
|
| 1674 |
+
"logits/rejected": -2.648607015609741,
|
| 1675 |
+
"logps/chosen": NaN,
|
| 1676 |
+
"logps/rejected": NaN,
|
| 1677 |
+
"loss": 1.3477,
|
| 1678 |
+
"nll_loss": 0.8656437993049622,
|
| 1679 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1680 |
+
"rewards/chosen": NaN,
|
| 1681 |
+
"rewards/margins": NaN,
|
| 1682 |
+
"rewards/rejected": NaN,
|
| 1683 |
+
"step": 930
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 2.084257206208426,
|
| 1687 |
+
"grad_norm": NaN,
|
| 1688 |
+
"learning_rate": 1.8016677139795635e-06,
|
| 1689 |
+
"log_odds_chosen": NaN,
|
| 1690 |
+
"log_odds_ratio": NaN,
|
| 1691 |
+
"logits/chosen": -2.5672965049743652,
|
| 1692 |
+
"logits/rejected": -2.5980045795440674,
|
| 1693 |
+
"logps/chosen": NaN,
|
| 1694 |
+
"logps/rejected": NaN,
|
| 1695 |
+
"loss": 1.404,
|
| 1696 |
+
"nll_loss": 0.8821004033088684,
|
| 1697 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1698 |
+
"rewards/chosen": NaN,
|
| 1699 |
+
"rewards/margins": NaN,
|
| 1700 |
+
"rewards/rejected": NaN,
|
| 1701 |
+
"step": 940
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 2.106430155210643,
|
| 1705 |
+
"grad_norm": NaN,
|
| 1706 |
+
"learning_rate": 1.7222869800160197e-06,
|
| 1707 |
+
"log_odds_chosen": NaN,
|
| 1708 |
+
"log_odds_ratio": NaN,
|
| 1709 |
+
"logits/chosen": -2.567884683609009,
|
| 1710 |
+
"logits/rejected": -2.6413073539733887,
|
| 1711 |
+
"logps/chosen": NaN,
|
| 1712 |
+
"logps/rejected": NaN,
|
| 1713 |
+
"loss": 1.5904,
|
| 1714 |
+
"nll_loss": 0.8698426485061646,
|
| 1715 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1716 |
+
"rewards/chosen": NaN,
|
| 1717 |
+
"rewards/margins": NaN,
|
| 1718 |
+
"rewards/rejected": NaN,
|
| 1719 |
+
"step": 950
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 2.12860310421286,
|
| 1723 |
+
"grad_norm": NaN,
|
| 1724 |
+
"learning_rate": 1.6442121472758776e-06,
|
| 1725 |
+
"log_odds_chosen": NaN,
|
| 1726 |
+
"log_odds_ratio": NaN,
|
| 1727 |
+
"logits/chosen": -2.5741963386535645,
|
| 1728 |
+
"logits/rejected": -2.6005654335021973,
|
| 1729 |
+
"logps/chosen": NaN,
|
| 1730 |
+
"logps/rejected": NaN,
|
| 1731 |
+
"loss": 1.7035,
|
| 1732 |
+
"nll_loss": 0.8915897607803345,
|
| 1733 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1734 |
+
"rewards/chosen": NaN,
|
| 1735 |
+
"rewards/margins": NaN,
|
| 1736 |
+
"rewards/rejected": NaN,
|
| 1737 |
+
"step": 960
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 2.1507760532150777,
|
| 1741 |
+
"grad_norm": NaN,
|
| 1742 |
+
"learning_rate": 1.5674879790891504e-06,
|
| 1743 |
+
"log_odds_chosen": NaN,
|
| 1744 |
+
"log_odds_ratio": NaN,
|
| 1745 |
+
"logits/chosen": -2.5121145248413086,
|
| 1746 |
+
"logits/rejected": -2.5986428260803223,
|
| 1747 |
+
"logps/chosen": NaN,
|
| 1748 |
+
"logps/rejected": NaN,
|
| 1749 |
+
"loss": 1.429,
|
| 1750 |
+
"nll_loss": 0.8993185758590698,
|
| 1751 |
+
"rewards/accuracies": 0.550000011920929,
|
| 1752 |
+
"rewards/chosen": NaN,
|
| 1753 |
+
"rewards/margins": NaN,
|
| 1754 |
+
"rewards/rejected": NaN,
|
| 1755 |
+
"step": 970
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 2.172949002217295,
|
| 1759 |
+
"grad_norm": NaN,
|
| 1760 |
+
"learning_rate": 1.4921584643974772e-06,
|
| 1761 |
+
"log_odds_chosen": NaN,
|
| 1762 |
+
"log_odds_ratio": NaN,
|
| 1763 |
+
"logits/chosen": -2.5978806018829346,
|
| 1764 |
+
"logits/rejected": -2.6225571632385254,
|
| 1765 |
+
"logps/chosen": NaN,
|
| 1766 |
+
"logps/rejected": NaN,
|
| 1767 |
+
"loss": 1.6976,
|
| 1768 |
+
"nll_loss": 0.8835185766220093,
|
| 1769 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1770 |
+
"rewards/chosen": NaN,
|
| 1771 |
+
"rewards/margins": NaN,
|
| 1772 |
+
"rewards/rejected": NaN,
|
| 1773 |
+
"step": 980
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 2.1951219512195124,
|
| 1777 |
+
"grad_norm": NaN,
|
| 1778 |
+
"learning_rate": 1.4182667925335472e-06,
|
| 1779 |
+
"log_odds_chosen": NaN,
|
| 1780 |
+
"log_odds_ratio": NaN,
|
| 1781 |
+
"logits/chosen": -2.5312979221343994,
|
| 1782 |
+
"logits/rejected": -2.619032382965088,
|
| 1783 |
+
"logps/chosen": NaN,
|
| 1784 |
+
"logps/rejected": NaN,
|
| 1785 |
+
"loss": 1.2105,
|
| 1786 |
+
"nll_loss": 0.8996240496635437,
|
| 1787 |
+
"rewards/accuracies": 0.6875,
|
| 1788 |
+
"rewards/chosen": NaN,
|
| 1789 |
+
"rewards/margins": NaN,
|
| 1790 |
+
"rewards/rejected": NaN,
|
| 1791 |
+
"step": 990
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 2.2172949002217295,
|
| 1795 |
+
"grad_norm": NaN,
|
| 1796 |
+
"learning_rate": 1.3458553284589852e-06,
|
| 1797 |
+
"log_odds_chosen": NaN,
|
| 1798 |
+
"log_odds_ratio": NaN,
|
| 1799 |
+
"logits/chosen": -2.586155414581299,
|
| 1800 |
+
"logits/rejected": -2.641432523727417,
|
| 1801 |
+
"logps/chosen": NaN,
|
| 1802 |
+
"logps/rejected": NaN,
|
| 1803 |
+
"loss": 1.4627,
|
| 1804 |
+
"nll_loss": 0.8832103610038757,
|
| 1805 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1806 |
+
"rewards/chosen": NaN,
|
| 1807 |
+
"rewards/margins": NaN,
|
| 1808 |
+
"rewards/rejected": NaN,
|
| 1809 |
+
"step": 1000
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 2.2394678492239466,
|
| 1813 |
+
"grad_norm": NaN,
|
| 1814 |
+
"learning_rate": 1.2749655884748788e-06,
|
| 1815 |
+
"log_odds_chosen": NaN,
|
| 1816 |
+
"log_odds_ratio": NaN,
|
| 1817 |
+
"logits/chosen": -2.649963617324829,
|
| 1818 |
+
"logits/rejected": -2.721240758895874,
|
| 1819 |
+
"logps/chosen": NaN,
|
| 1820 |
+
"logps/rejected": NaN,
|
| 1821 |
+
"loss": 0.9071,
|
| 1822 |
+
"nll_loss": 0.8600472211837769,
|
| 1823 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1824 |
+
"rewards/chosen": NaN,
|
| 1825 |
+
"rewards/margins": NaN,
|
| 1826 |
+
"rewards/rejected": NaN,
|
| 1827 |
+
"step": 1010
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 2.261640798226164,
|
| 1831 |
+
"grad_norm": NaN,
|
| 1832 |
+
"learning_rate": 1.205638216418864e-06,
|
| 1833 |
+
"log_odds_chosen": NaN,
|
| 1834 |
+
"log_odds_ratio": NaN,
|
| 1835 |
+
"logits/chosen": -2.6075966358184814,
|
| 1836 |
+
"logits/rejected": -2.6658735275268555,
|
| 1837 |
+
"logps/chosen": NaN,
|
| 1838 |
+
"logps/rejected": NaN,
|
| 1839 |
+
"loss": 1.3184,
|
| 1840 |
+
"nll_loss": 0.9156390428543091,
|
| 1841 |
+
"rewards/accuracies": 0.625,
|
| 1842 |
+
"rewards/chosen": NaN,
|
| 1843 |
+
"rewards/margins": NaN,
|
| 1844 |
+
"rewards/rejected": NaN,
|
| 1845 |
+
"step": 1020
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 2.2838137472283813,
|
| 1849 |
+
"grad_norm": NaN,
|
| 1850 |
+
"learning_rate": 1.1379129603624472e-06,
|
| 1851 |
+
"log_odds_chosen": NaN,
|
| 1852 |
+
"log_odds_ratio": NaN,
|
| 1853 |
+
"logits/chosen": -2.5418643951416016,
|
| 1854 |
+
"logits/rejected": -2.5810935497283936,
|
| 1855 |
+
"logps/chosen": NaN,
|
| 1856 |
+
"logps/rejected": NaN,
|
| 1857 |
+
"loss": 1.3225,
|
| 1858 |
+
"nll_loss": 0.9440028071403503,
|
| 1859 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1860 |
+
"rewards/chosen": NaN,
|
| 1861 |
+
"rewards/margins": NaN,
|
| 1862 |
+
"rewards/rejected": NaN,
|
| 1863 |
+
"step": 1030
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 2.305986696230599,
|
| 1867 |
+
"grad_norm": 0.6439170241355896,
|
| 1868 |
+
"learning_rate": 1.0718286498218834e-06,
|
| 1869 |
+
"log_odds_chosen": NaN,
|
| 1870 |
+
"log_odds_ratio": NaN,
|
| 1871 |
+
"logits/chosen": -2.5944840908050537,
|
| 1872 |
+
"logits/rejected": -2.6368420124053955,
|
| 1873 |
+
"logps/chosen": NaN,
|
| 1874 |
+
"logps/rejected": NaN,
|
| 1875 |
+
"loss": 1.3297,
|
| 1876 |
+
"nll_loss": 0.8718999028205872,
|
| 1877 |
+
"rewards/accuracies": 0.6875,
|
| 1878 |
+
"rewards/chosen": NaN,
|
| 1879 |
+
"rewards/margins": NaN,
|
| 1880 |
+
"rewards/rejected": NaN,
|
| 1881 |
+
"step": 1040
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 2.328159645232816,
|
| 1885 |
+
"grad_norm": NaN,
|
| 1886 |
+
"learning_rate": 1.0074231734957184e-06,
|
| 1887 |
+
"log_odds_chosen": NaN,
|
| 1888 |
+
"log_odds_ratio": NaN,
|
| 1889 |
+
"logits/chosen": -2.5355989933013916,
|
| 1890 |
+
"logits/rejected": -2.5640969276428223,
|
| 1891 |
+
"logps/chosen": NaN,
|
| 1892 |
+
"logps/rejected": NaN,
|
| 1893 |
+
"loss": 1.1733,
|
| 1894 |
+
"nll_loss": 0.8988415002822876,
|
| 1895 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 1896 |
+
"rewards/chosen": NaN,
|
| 1897 |
+
"rewards/margins": NaN,
|
| 1898 |
+
"rewards/rejected": NaN,
|
| 1899 |
+
"step": 1050
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 2.3503325942350335,
|
| 1903 |
+
"grad_norm": NaN,
|
| 1904 |
+
"learning_rate": 9.447334575417189e-07,
|
| 1905 |
+
"log_odds_chosen": NaN,
|
| 1906 |
+
"log_odds_ratio": NaN,
|
| 1907 |
+
"logits/chosen": -2.605184555053711,
|
| 1908 |
+
"logits/rejected": -2.6584599018096924,
|
| 1909 |
+
"logps/chosen": NaN,
|
| 1910 |
+
"logps/rejected": NaN,
|
| 1911 |
+
"loss": 1.4157,
|
| 1912 |
+
"nll_loss": 0.8994711637496948,
|
| 1913 |
+
"rewards/accuracies": 0.625,
|
| 1914 |
+
"rewards/chosen": NaN,
|
| 1915 |
+
"rewards/margins": NaN,
|
| 1916 |
+
"rewards/rejected": NaN,
|
| 1917 |
+
"step": 1060
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 2.3725055432372506,
|
| 1921 |
+
"grad_norm": NaN,
|
| 1922 |
+
"learning_rate": 8.837954444056825e-07,
|
| 1923 |
+
"log_odds_chosen": NaN,
|
| 1924 |
+
"log_odds_ratio": NaN,
|
| 1925 |
+
"logits/chosen": -2.521874189376831,
|
| 1926 |
+
"logits/rejected": -2.5487990379333496,
|
| 1927 |
+
"logps/chosen": NaN,
|
| 1928 |
+
"logps/rejected": NaN,
|
| 1929 |
+
"loss": 1.504,
|
| 1930 |
+
"nll_loss": 0.8935952186584473,
|
| 1931 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1932 |
+
"rewards/chosen": NaN,
|
| 1933 |
+
"rewards/margins": NaN,
|
| 1934 |
+
"rewards/rejected": NaN,
|
| 1935 |
+
"step": 1070
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 2.3946784922394677,
|
| 1939 |
+
"grad_norm": 0.6280531883239746,
|
| 1940 |
+
"learning_rate": 8.246440722142325e-07,
|
| 1941 |
+
"log_odds_chosen": NaN,
|
| 1942 |
+
"log_odds_ratio": NaN,
|
| 1943 |
+
"logits/chosen": -2.612034320831299,
|
| 1944 |
+
"logits/rejected": -2.640134811401367,
|
| 1945 |
+
"logps/chosen": NaN,
|
| 1946 |
+
"logps/rejected": NaN,
|
| 1947 |
+
"loss": 1.1654,
|
| 1948 |
+
"nll_loss": 0.9181571006774902,
|
| 1949 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1950 |
+
"rewards/chosen": NaN,
|
| 1951 |
+
"rewards/margins": NaN,
|
| 1952 |
+
"rewards/rejected": NaN,
|
| 1953 |
+
"step": 1080
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 2.4168514412416853,
|
| 1957 |
+
"grad_norm": NaN,
|
| 1958 |
+
"learning_rate": 7.67313254743438e-07,
|
| 1959 |
+
"log_odds_chosen": NaN,
|
| 1960 |
+
"log_odds_ratio": NaN,
|
| 1961 |
+
"logits/chosen": -2.6642110347747803,
|
| 1962 |
+
"logits/rejected": -2.6616063117980957,
|
| 1963 |
+
"logps/chosen": NaN,
|
| 1964 |
+
"logps/rejected": NaN,
|
| 1965 |
+
"loss": 1.3256,
|
| 1966 |
+
"nll_loss": 0.9149085283279419,
|
| 1967 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1968 |
+
"rewards/chosen": NaN,
|
| 1969 |
+
"rewards/margins": NaN,
|
| 1970 |
+
"rewards/rejected": NaN,
|
| 1971 |
+
"step": 1090
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 2.4390243902439024,
|
| 1975 |
+
"grad_norm": NaN,
|
| 1976 |
+
"learning_rate": 7.118358619747322e-07,
|
| 1977 |
+
"log_odds_chosen": NaN,
|
| 1978 |
+
"log_odds_ratio": NaN,
|
| 1979 |
+
"logits/chosen": -2.5517077445983887,
|
| 1980 |
+
"logits/rejected": -2.5954713821411133,
|
| 1981 |
+
"logps/chosen": NaN,
|
| 1982 |
+
"logps/rejected": NaN,
|
| 1983 |
+
"loss": 1.3177,
|
| 1984 |
+
"nll_loss": 0.9105826616287231,
|
| 1985 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1986 |
+
"rewards/chosen": NaN,
|
| 1987 |
+
"rewards/margins": NaN,
|
| 1988 |
+
"rewards/rejected": NaN,
|
| 1989 |
+
"step": 1100
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 2.4611973392461195,
|
| 1993 |
+
"grad_norm": NaN,
|
| 1994 |
+
"learning_rate": 6.582437012492725e-07,
|
| 1995 |
+
"log_odds_chosen": NaN,
|
| 1996 |
+
"log_odds_ratio": NaN,
|
| 1997 |
+
"logits/chosen": -2.6221938133239746,
|
| 1998 |
+
"logits/rejected": -2.6449408531188965,
|
| 1999 |
+
"logps/chosen": NaN,
|
| 2000 |
+
"logps/rejected": NaN,
|
| 2001 |
+
"loss": 1.387,
|
| 2002 |
+
"nll_loss": 0.891291618347168,
|
| 2003 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 2004 |
+
"rewards/chosen": NaN,
|
| 2005 |
+
"rewards/margins": NaN,
|
| 2006 |
+
"rewards/rejected": NaN,
|
| 2007 |
+
"step": 1110
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 2.483370288248337,
|
| 2011 |
+
"grad_norm": NaN,
|
| 2012 |
+
"learning_rate": 6.065674990315623e-07,
|
| 2013 |
+
"log_odds_chosen": NaN,
|
| 2014 |
+
"log_odds_ratio": NaN,
|
| 2015 |
+
"logits/chosen": -2.5601844787597656,
|
| 2016 |
+
"logits/rejected": -2.58024263381958,
|
| 2017 |
+
"logps/chosen": NaN,
|
| 2018 |
+
"logps/rejected": NaN,
|
| 2019 |
+
"loss": 1.2113,
|
| 2020 |
+
"nll_loss": 0.8098522424697876,
|
| 2021 |
+
"rewards/accuracies": 0.612500011920929,
|
| 2022 |
+
"rewards/chosen": NaN,
|
| 2023 |
+
"rewards/margins": NaN,
|
| 2024 |
+
"rewards/rejected": NaN,
|
| 2025 |
+
"step": 1120
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 2.505543237250554,
|
| 2029 |
+
"grad_norm": NaN,
|
| 2030 |
+
"learning_rate": 5.568368832927742e-07,
|
| 2031 |
+
"log_odds_chosen": NaN,
|
| 2032 |
+
"log_odds_ratio": NaN,
|
| 2033 |
+
"logits/chosen": -2.5909504890441895,
|
| 2034 |
+
"logits/rejected": -2.596946954727173,
|
| 2035 |
+
"logps/chosen": NaN,
|
| 2036 |
+
"logps/rejected": NaN,
|
| 2037 |
+
"loss": 1.3932,
|
| 2038 |
+
"nll_loss": 0.9140310287475586,
|
| 2039 |
+
"rewards/accuracies": 0.625,
|
| 2040 |
+
"rewards/chosen": NaN,
|
| 2041 |
+
"rewards/margins": NaN,
|
| 2042 |
+
"rewards/rejected": NaN,
|
| 2043 |
+
"step": 1130
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 2.5277161862527717,
|
| 2047 |
+
"grad_norm": NaN,
|
| 2048 |
+
"learning_rate": 5.090803665238872e-07,
|
| 2049 |
+
"log_odds_chosen": NaN,
|
| 2050 |
+
"log_odds_ratio": NaN,
|
| 2051 |
+
"logits/chosen": -2.5009472370147705,
|
| 2052 |
+
"logits/rejected": -2.5562808513641357,
|
| 2053 |
+
"logps/chosen": NaN,
|
| 2054 |
+
"logps/rejected": NaN,
|
| 2055 |
+
"loss": 1.2941,
|
| 2056 |
+
"nll_loss": 0.8312563896179199,
|
| 2057 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 2058 |
+
"rewards/chosen": NaN,
|
| 2059 |
+
"rewards/margins": NaN,
|
| 2060 |
+
"rewards/rejected": NaN,
|
| 2061 |
+
"step": 1140
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 2.549889135254989,
|
| 2065 |
+
"grad_norm": NaN,
|
| 2066 |
+
"learning_rate": 4.633253293883679e-07,
|
| 2067 |
+
"log_odds_chosen": NaN,
|
| 2068 |
+
"log_odds_ratio": NaN,
|
| 2069 |
+
"logits/chosen": -2.6316781044006348,
|
| 2070 |
+
"logits/rejected": -2.635262966156006,
|
| 2071 |
+
"logps/chosen": NaN,
|
| 2072 |
+
"logps/rejected": NaN,
|
| 2073 |
+
"loss": 1.4821,
|
| 2074 |
+
"nll_loss": 0.8716381192207336,
|
| 2075 |
+
"rewards/accuracies": 0.5625,
|
| 2076 |
+
"rewards/chosen": NaN,
|
| 2077 |
+
"rewards/margins": NaN,
|
| 2078 |
+
"rewards/rejected": NaN,
|
| 2079 |
+
"step": 1150
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 2.5720620842572064,
|
| 2083 |
+
"grad_norm": NaN,
|
| 2084 |
+
"learning_rate": 4.19598005023774e-07,
|
| 2085 |
+
"log_odds_chosen": NaN,
|
| 2086 |
+
"log_odds_ratio": NaN,
|
| 2087 |
+
"logits/chosen": -2.6068694591522217,
|
| 2088 |
+
"logits/rejected": -2.650502920150757,
|
| 2089 |
+
"logps/chosen": NaN,
|
| 2090 |
+
"logps/rejected": NaN,
|
| 2091 |
+
"loss": 1.5451,
|
| 2092 |
+
"nll_loss": 0.9065351486206055,
|
| 2093 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 2094 |
+
"rewards/chosen": NaN,
|
| 2095 |
+
"rewards/margins": NaN,
|
| 2096 |
+
"rewards/rejected": NaN,
|
| 2097 |
+
"step": 1160
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 2.5942350332594235,
|
| 2101 |
+
"grad_norm": NaN,
|
| 2102 |
+
"learning_rate": 3.7792346400128183e-07,
|
| 2103 |
+
"log_odds_chosen": NaN,
|
| 2104 |
+
"log_odds_ratio": NaN,
|
| 2105 |
+
"logits/chosen": -2.579982280731201,
|
| 2106 |
+
"logits/rejected": -2.643984317779541,
|
| 2107 |
+
"logps/chosen": NaN,
|
| 2108 |
+
"logps/rejected": NaN,
|
| 2109 |
+
"loss": 1.3209,
|
| 2110 |
+
"nll_loss": 0.8348051905632019,
|
| 2111 |
+
"rewards/accuracies": 0.612500011920929,
|
| 2112 |
+
"rewards/chosen": NaN,
|
| 2113 |
+
"rewards/margins": NaN,
|
| 2114 |
+
"rewards/rejected": NaN,
|
| 2115 |
+
"step": 1170
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 2.6164079822616406,
|
| 2119 |
+
"grad_norm": NaN,
|
| 2120 |
+
"learning_rate": 3.3832559995175116e-07,
|
| 2121 |
+
"log_odds_chosen": NaN,
|
| 2122 |
+
"log_odds_ratio": NaN,
|
| 2123 |
+
"logits/chosen": -2.5105443000793457,
|
| 2124 |
+
"logits/rejected": -2.57656192779541,
|
| 2125 |
+
"logps/chosen": NaN,
|
| 2126 |
+
"logps/rejected": NaN,
|
| 2127 |
+
"loss": 1.4288,
|
| 2128 |
+
"nll_loss": 0.997296154499054,
|
| 2129 |
+
"rewards/accuracies": 0.637499988079071,
|
| 2130 |
+
"rewards/chosen": NaN,
|
| 2131 |
+
"rewards/margins": NaN,
|
| 2132 |
+
"rewards/rejected": NaN,
|
| 2133 |
+
"step": 1180
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 2.638580931263858,
|
| 2137 |
+
"grad_norm": NaN,
|
| 2138 |
+
"learning_rate": 3.0082711586658336e-07,
|
| 2139 |
+
"log_odds_chosen": NaN,
|
| 2140 |
+
"log_odds_ratio": NaN,
|
| 2141 |
+
"logits/chosen": -2.5682618618011475,
|
| 2142 |
+
"logits/rejected": -2.6190028190612793,
|
| 2143 |
+
"logps/chosen": NaN,
|
| 2144 |
+
"logps/rejected": NaN,
|
| 2145 |
+
"loss": 1.3811,
|
| 2146 |
+
"nll_loss": 0.840679943561554,
|
| 2147 |
+
"rewards/accuracies": 0.574999988079071,
|
| 2148 |
+
"rewards/chosen": NaN,
|
| 2149 |
+
"rewards/margins": NaN,
|
| 2150 |
+
"rewards/rejected": NaN,
|
| 2151 |
+
"step": 1190
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 2.6607538802660753,
|
| 2155 |
+
"grad_norm": NaN,
|
| 2156 |
+
"learning_rate": 2.654495110812136e-07,
|
| 2157 |
+
"log_odds_chosen": NaN,
|
| 2158 |
+
"log_odds_ratio": NaN,
|
| 2159 |
+
"logits/chosen": -2.534421920776367,
|
| 2160 |
+
"logits/rejected": -2.5627925395965576,
|
| 2161 |
+
"logps/chosen": NaN,
|
| 2162 |
+
"logps/rejected": NaN,
|
| 2163 |
+
"loss": 1.4835,
|
| 2164 |
+
"nll_loss": 0.9845026135444641,
|
| 2165 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 2166 |
+
"rewards/chosen": NaN,
|
| 2167 |
+
"rewards/margins": NaN,
|
| 2168 |
+
"rewards/rejected": NaN,
|
| 2169 |
+
"step": 1200
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 2.682926829268293,
|
| 2173 |
+
"grad_norm": NaN,
|
| 2174 |
+
"learning_rate": 2.3221306894870962e-07,
|
| 2175 |
+
"log_odds_chosen": NaN,
|
| 2176 |
+
"log_odds_ratio": NaN,
|
| 2177 |
+
"logits/chosen": -2.597075939178467,
|
| 2178 |
+
"logits/rejected": -2.605375051498413,
|
| 2179 |
+
"logps/chosen": NaN,
|
| 2180 |
+
"logps/rejected": NaN,
|
| 2181 |
+
"loss": 1.9092,
|
| 2182 |
+
"nll_loss": 0.8531631231307983,
|
| 2183 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 2184 |
+
"rewards/chosen": NaN,
|
| 2185 |
+
"rewards/margins": NaN,
|
| 2186 |
+
"rewards/rejected": NaN,
|
| 2187 |
+
"step": 1210
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 2.70509977827051,
|
| 2191 |
+
"grad_norm": NaN,
|
| 2192 |
+
"learning_rate": 2.0113684521053663e-07,
|
| 2193 |
+
"log_odds_chosen": NaN,
|
| 2194 |
+
"log_odds_ratio": NaN,
|
| 2195 |
+
"logits/chosen": -2.5460121631622314,
|
| 2196 |
+
"logits/rejected": -2.6030726432800293,
|
| 2197 |
+
"logps/chosen": NaN,
|
| 2198 |
+
"logps/rejected": NaN,
|
| 2199 |
+
"loss": 1.5883,
|
| 2200 |
+
"nll_loss": 0.8551692962646484,
|
| 2201 |
+
"rewards/accuracies": 0.5625,
|
| 2202 |
+
"rewards/chosen": NaN,
|
| 2203 |
+
"rewards/margins": NaN,
|
| 2204 |
+
"rewards/rejected": NaN,
|
| 2205 |
+
"step": 1220
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 2.7272727272727275,
|
| 2209 |
+
"grad_norm": NaN,
|
| 2210 |
+
"learning_rate": 1.722386570711647e-07,
|
| 2211 |
+
"log_odds_chosen": NaN,
|
| 2212 |
+
"log_odds_ratio": NaN,
|
| 2213 |
+
"logits/chosen": -2.559178352355957,
|
| 2214 |
+
"logits/rejected": -2.601423740386963,
|
| 2215 |
+
"logps/chosen": NaN,
|
| 2216 |
+
"logps/rejected": NaN,
|
| 2217 |
+
"loss": 1.5836,
|
| 2218 |
+
"nll_loss": 0.8850408792495728,
|
| 2219 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 2220 |
+
"rewards/chosen": NaN,
|
| 2221 |
+
"rewards/margins": NaN,
|
| 2222 |
+
"rewards/rejected": NaN,
|
| 2223 |
+
"step": 1230
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 2.7494456762749446,
|
| 2227 |
+
"grad_norm": 0.6112498641014099,
|
| 2228 |
+
"learning_rate": 1.455350729827698e-07,
|
| 2229 |
+
"log_odds_chosen": NaN,
|
| 2230 |
+
"log_odds_ratio": NaN,
|
| 2231 |
+
"logits/chosen": -2.579529285430908,
|
| 2232 |
+
"logits/rejected": -2.5950798988342285,
|
| 2233 |
+
"logps/chosen": NaN,
|
| 2234 |
+
"logps/rejected": NaN,
|
| 2235 |
+
"loss": 1.3608,
|
| 2236 |
+
"nll_loss": 0.8255780339241028,
|
| 2237 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 2238 |
+
"rewards/chosen": NaN,
|
| 2239 |
+
"rewards/margins": NaN,
|
| 2240 |
+
"rewards/rejected": NaN,
|
| 2241 |
+
"step": 1240
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 2.7716186252771617,
|
| 2245 |
+
"grad_norm": NaN,
|
| 2246 |
+
"learning_rate": 1.2104140314590194e-07,
|
| 2247 |
+
"log_odds_chosen": NaN,
|
| 2248 |
+
"log_odds_ratio": NaN,
|
| 2249 |
+
"logits/chosen": -2.525857448577881,
|
| 2250 |
+
"logits/rejected": -2.587132215499878,
|
| 2251 |
+
"logps/chosen": NaN,
|
| 2252 |
+
"logps/rejected": NaN,
|
| 2253 |
+
"loss": 1.5238,
|
| 2254 |
+
"nll_loss": 0.9289102554321289,
|
| 2255 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 2256 |
+
"rewards/chosen": NaN,
|
| 2257 |
+
"rewards/margins": NaN,
|
| 2258 |
+
"rewards/rejected": NaN,
|
| 2259 |
+
"step": 1250
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 2.7937915742793793,
|
| 2263 |
+
"grad_norm": NaN,
|
| 2264 |
+
"learning_rate": 9.877169073155167e-08,
|
| 2265 |
+
"log_odds_chosen": NaN,
|
| 2266 |
+
"log_odds_ratio": NaN,
|
| 2267 |
+
"logits/chosen": -2.533351182937622,
|
| 2268 |
+
"logits/rejected": -2.584251880645752,
|
| 2269 |
+
"logps/chosen": NaN,
|
| 2270 |
+
"logps/rejected": NaN,
|
| 2271 |
+
"loss": 1.4373,
|
| 2272 |
+
"nll_loss": 0.9016596674919128,
|
| 2273 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 2274 |
+
"rewards/chosen": NaN,
|
| 2275 |
+
"rewards/margins": NaN,
|
| 2276 |
+
"rewards/rejected": NaN,
|
| 2277 |
+
"step": 1260
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 2.8159645232815964,
|
| 2281 |
+
"grad_norm": NaN,
|
| 2282 |
+
"learning_rate": 7.873870382965364e-08,
|
| 2283 |
+
"log_odds_chosen": NaN,
|
| 2284 |
+
"log_odds_ratio": NaN,
|
| 2285 |
+
"logits/chosen": -2.5861144065856934,
|
| 2286 |
+
"logits/rejected": -2.612295150756836,
|
| 2287 |
+
"logps/chosen": NaN,
|
| 2288 |
+
"logps/rejected": NaN,
|
| 2289 |
+
"loss": 1.5352,
|
| 2290 |
+
"nll_loss": 0.8941007852554321,
|
| 2291 |
+
"rewards/accuracies": 0.612500011920929,
|
| 2292 |
+
"rewards/chosen": NaN,
|
| 2293 |
+
"rewards/margins": NaN,
|
| 2294 |
+
"rewards/rejected": NaN,
|
| 2295 |
+
"step": 1270
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 2.8381374722838135,
|
| 2299 |
+
"grad_norm": 0.6927753686904907,
|
| 2300 |
+
"learning_rate": 6.095392812864863e-08,
|
| 2301 |
+
"log_odds_chosen": NaN,
|
| 2302 |
+
"log_odds_ratio": NaN,
|
| 2303 |
+
"logits/chosen": -2.6299588680267334,
|
| 2304 |
+
"logits/rejected": -2.654484272003174,
|
| 2305 |
+
"logps/chosen": NaN,
|
| 2306 |
+
"logps/rejected": NaN,
|
| 2307 |
+
"loss": 1.1739,
|
| 2308 |
+
"nll_loss": 0.9128586053848267,
|
| 2309 |
+
"rewards/accuracies": 0.5,
|
| 2310 |
+
"rewards/chosen": NaN,
|
| 2311 |
+
"rewards/margins": NaN,
|
| 2312 |
+
"rewards/rejected": NaN,
|
| 2313 |
+
"step": 1280
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 2.860310421286031,
|
| 2317 |
+
"grad_norm": 0.6138539910316467,
|
| 2318 |
+
"learning_rate": 4.5427560330289824e-08,
|
| 2319 |
+
"log_odds_chosen": NaN,
|
| 2320 |
+
"log_odds_ratio": NaN,
|
| 2321 |
+
"logits/chosen": -2.5274271965026855,
|
| 2322 |
+
"logits/rejected": -2.5881905555725098,
|
| 2323 |
+
"logps/chosen": NaN,
|
| 2324 |
+
"logps/rejected": NaN,
|
| 2325 |
+
"loss": 1.0278,
|
| 2326 |
+
"nll_loss": 0.8638602495193481,
|
| 2327 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 2328 |
+
"rewards/chosen": NaN,
|
| 2329 |
+
"rewards/margins": NaN,
|
| 2330 |
+
"rewards/rejected": NaN,
|
| 2331 |
+
"step": 1290
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 2.882483370288248,
|
| 2335 |
+
"grad_norm": NaN,
|
| 2336 |
+
"learning_rate": 3.216850230348145e-08,
|
| 2337 |
+
"log_odds_chosen": NaN,
|
| 2338 |
+
"log_odds_ratio": NaN,
|
| 2339 |
+
"logits/chosen": -2.6004536151885986,
|
| 2340 |
+
"logits/rejected": -2.639181613922119,
|
| 2341 |
+
"logps/chosen": NaN,
|
| 2342 |
+
"logps/rejected": NaN,
|
| 2343 |
+
"loss": 1.3678,
|
| 2344 |
+
"nll_loss": 0.9433631896972656,
|
| 2345 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 2346 |
+
"rewards/chosen": NaN,
|
| 2347 |
+
"rewards/margins": NaN,
|
| 2348 |
+
"rewards/rejected": NaN,
|
| 2349 |
+
"step": 1300
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 2.9046563192904657,
|
| 2353 |
+
"grad_norm": NaN,
|
| 2354 |
+
"learning_rate": 2.1184355980488067e-08,
|
| 2355 |
+
"log_odds_chosen": NaN,
|
| 2356 |
+
"log_odds_ratio": NaN,
|
| 2357 |
+
"logits/chosen": -2.6405842304229736,
|
| 2358 |
+
"logits/rejected": -2.632291316986084,
|
| 2359 |
+
"logps/chosen": NaN,
|
| 2360 |
+
"logps/rejected": NaN,
|
| 2361 |
+
"loss": 1.7057,
|
| 2362 |
+
"nll_loss": 0.9545801281929016,
|
| 2363 |
+
"rewards/accuracies": 0.612500011920929,
|
| 2364 |
+
"rewards/chosen": NaN,
|
| 2365 |
+
"rewards/margins": NaN,
|
| 2366 |
+
"rewards/rejected": NaN,
|
| 2367 |
+
"step": 1310
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 2.926829268292683,
|
| 2371 |
+
"grad_norm": NaN,
|
| 2372 |
+
"learning_rate": 1.2481418998456118e-08,
|
| 2373 |
+
"log_odds_chosen": NaN,
|
| 2374 |
+
"log_odds_ratio": NaN,
|
| 2375 |
+
"logits/chosen": -2.552712917327881,
|
| 2376 |
+
"logits/rejected": -2.603447437286377,
|
| 2377 |
+
"logps/chosen": NaN,
|
| 2378 |
+
"logps/rejected": NaN,
|
| 2379 |
+
"loss": 1.1937,
|
| 2380 |
+
"nll_loss": 0.885978102684021,
|
| 2381 |
+
"rewards/accuracies": 0.612500011920929,
|
| 2382 |
+
"rewards/chosen": NaN,
|
| 2383 |
+
"rewards/margins": NaN,
|
| 2384 |
+
"rewards/rejected": NaN,
|
| 2385 |
+
"step": 1320
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 2.9490022172949004,
|
| 2389 |
+
"grad_norm": NaN,
|
| 2390 |
+
"learning_rate": 6.064681088730151e-09,
|
| 2391 |
+
"log_odds_chosen": NaN,
|
| 2392 |
+
"log_odds_ratio": NaN,
|
| 2393 |
+
"logits/chosen": -2.49711275100708,
|
| 2394 |
+
"logits/rejected": -2.5226452350616455,
|
| 2395 |
+
"logps/chosen": NaN,
|
| 2396 |
+
"logps/rejected": NaN,
|
| 2397 |
+
"loss": 1.2368,
|
| 2398 |
+
"nll_loss": 0.8924044370651245,
|
| 2399 |
+
"rewards/accuracies": 0.574999988079071,
|
| 2400 |
+
"rewards/chosen": NaN,
|
| 2401 |
+
"rewards/margins": NaN,
|
| 2402 |
+
"rewards/rejected": NaN,
|
| 2403 |
+
"step": 1330
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 2.9711751662971175,
|
| 2407 |
+
"grad_norm": NaN,
|
| 2408 |
+
"learning_rate": 1.9378212160501285e-09,
|
| 2409 |
+
"log_odds_chosen": NaN,
|
| 2410 |
+
"log_odds_ratio": NaN,
|
| 2411 |
+
"logits/chosen": -2.568042755126953,
|
| 2412 |
+
"logits/rejected": -2.608767032623291,
|
| 2413 |
+
"logps/chosen": NaN,
|
| 2414 |
+
"logps/rejected": NaN,
|
| 2415 |
+
"loss": 1.4421,
|
| 2416 |
+
"nll_loss": 0.8176005482673645,
|
| 2417 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 2418 |
+
"rewards/chosen": NaN,
|
| 2419 |
+
"rewards/margins": NaN,
|
| 2420 |
+
"rewards/rejected": NaN,
|
| 2421 |
+
"step": 1340
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 2.9933481152993346,
|
| 2425 |
+
"grad_norm": NaN,
|
| 2426 |
+
"learning_rate": 1.0320546925512985e-10,
|
| 2427 |
+
"log_odds_chosen": NaN,
|
| 2428 |
+
"log_odds_ratio": NaN,
|
| 2429 |
+
"logits/chosen": -2.606173276901245,
|
| 2430 |
+
"logits/rejected": -2.6094250679016113,
|
| 2431 |
+
"logps/chosen": NaN,
|
| 2432 |
+
"logps/rejected": NaN,
|
| 2433 |
+
"loss": 1.9638,
|
| 2434 |
+
"nll_loss": 0.9152927398681641,
|
| 2435 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 2436 |
+
"rewards/chosen": NaN,
|
| 2437 |
+
"rewards/margins": NaN,
|
| 2438 |
+
"rewards/rejected": NaN,
|
| 2439 |
+
"step": 1350
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 3.0,
|
| 2443 |
+
"step": 1353,
|
| 2444 |
+
"total_flos": 0.0,
|
| 2445 |
+
"train_loss": 1.4294498003771459,
|
| 2446 |
+
"train_runtime": 7808.9602,
|
| 2447 |
+
"train_samples_per_second": 2.77,
|
| 2448 |
+
"train_steps_per_second": 0.173
|
| 2449 |
+
}
|
| 2450 |
+
],
|
| 2451 |
+
"logging_steps": 10,
|
| 2452 |
+
"max_steps": 1353,
|
| 2453 |
+
"num_input_tokens_seen": 0,
|
| 2454 |
+
"num_train_epochs": 3,
|
| 2455 |
+
"save_steps": 200,
|
| 2456 |
+
"stateful_callbacks": {
|
| 2457 |
+
"TrainerControl": {
|
| 2458 |
+
"args": {
|
| 2459 |
+
"should_epoch_stop": false,
|
| 2460 |
+
"should_evaluate": false,
|
| 2461 |
+
"should_log": false,
|
| 2462 |
+
"should_save": true,
|
| 2463 |
+
"should_training_stop": true
|
| 2464 |
+
},
|
| 2465 |
+
"attributes": {}
|
| 2466 |
+
}
|
| 2467 |
+
},
|
| 2468 |
+
"total_flos": 0.0,
|
| 2469 |
+
"train_batch_size": 4,
|
| 2470 |
+
"trial_name": null,
|
| 2471 |
+
"trial_params": null
|
| 2472 |
+
}
|