Model save
Browse files- README.md +69 -0
- all_results.json +9 -0
- runs/Jan22_20-31-46_ip-172-31-34-75/events.out.tfevents.1737578036.ip-172-31-34-75.4941.0 +2 -2
- train_results.json +9 -0
- trainer_state.json +2262 -0
README.md
ADDED
|
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: philschmid/llama-3-1-8b-math-orca-spectrum-10k-ep1
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: dpo-llama-3-1-8b-math-ep3
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- dpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for dpo-llama-3-1-8b-math-ep3
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [philschmid/llama-3-1-8b-math-orca-spectrum-10k-ep1](https://huggingface.co/philschmid/llama-3-1-8b-math-orca-spectrum-10k-ep1).
|
| 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="philschmid/dpo-llama-3-1-8b-math-ep3", device="cuda")
|
| 24 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 25 |
+
print(output["generated_text"])
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Training procedure
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.14.0.dev0
|
| 38 |
+
- Transformers: 4.48.0
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.1.0
|
| 41 |
+
- Tokenizers: 0.21.0
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite DPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@inproceedings{rafailov2023direct,
|
| 49 |
+
title = {{Direct Preference Optimization: Your Language Model is Secretly a Reward Model}},
|
| 50 |
+
author = {Rafael Rafailov and Archit Sharma and Eric Mitchell and Christopher D. Manning and Stefano Ermon and Chelsea Finn},
|
| 51 |
+
year = 2023,
|
| 52 |
+
booktitle = {Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023},
|
| 53 |
+
url = {http://papers.nips.cc/paper_files/paper/2023/hash/a85b405ed65c6477a4fe8302b5e06ce7-Abstract-Conference.html},
|
| 54 |
+
editor = {Alice Oh and Tristan Naumann and Amir Globerson and Kate Saenko and Moritz Hardt and Sergey Levine},
|
| 55 |
+
}
|
| 56 |
+
```
|
| 57 |
+
|
| 58 |
+
Cite TRL as:
|
| 59 |
+
|
| 60 |
+
```bibtex
|
| 61 |
+
@misc{vonwerra2022trl,
|
| 62 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 63 |
+
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},
|
| 64 |
+
year = 2020,
|
| 65 |
+
journal = {GitHub repository},
|
| 66 |
+
publisher = {GitHub},
|
| 67 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 68 |
+
}
|
| 69 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.989399293286219,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.586526518971653,
|
| 5 |
+
"train_runtime": 26068.2965,
|
| 6 |
+
"train_samples": 1981,
|
| 7 |
+
"train_samples_per_second": 0.228,
|
| 8 |
+
"train_steps_per_second": 0.028
|
| 9 |
+
}
|
runs/Jan22_20-31-46_ip-172-31-34-75/events.out.tfevents.1737578036.ip-172-31-34-75.4941.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5cfed40f4894076b59ba6ee487444284342f8ab3075a061278138549e0ae4de2
|
| 3 |
+
size 108707
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.989399293286219,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.586526518971653,
|
| 5 |
+
"train_runtime": 26068.2965,
|
| 6 |
+
"train_samples": 1981,
|
| 7 |
+
"train_samples_per_second": 0.228,
|
| 8 |
+
"train_steps_per_second": 0.028
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2262 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.989399293286219,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 741,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.020191822311963654,
|
| 13 |
+
"grad_norm": 30.7789249420166,
|
| 14 |
+
"learning_rate": 5e-06,
|
| 15 |
+
"logits/chosen": -1.0492041110992432,
|
| 16 |
+
"logits/rejected": -1.0134608745574951,
|
| 17 |
+
"logps/chosen": -127.44047546386719,
|
| 18 |
+
"logps/rejected": -147.18359375,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 21 |
+
"rewards/chosen": 0.009235936217010021,
|
| 22 |
+
"rewards/margins": 0.0008875288185663521,
|
| 23 |
+
"rewards/rejected": 0.008348408155143261,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04038364462392731,
|
| 28 |
+
"grad_norm": 37.41344451904297,
|
| 29 |
+
"learning_rate": 5e-06,
|
| 30 |
+
"logits/chosen": -1.002793312072754,
|
| 31 |
+
"logits/rejected": -0.9411100149154663,
|
| 32 |
+
"logps/chosen": -126.66816711425781,
|
| 33 |
+
"logps/rejected": -159.05479431152344,
|
| 34 |
+
"loss": 0.6969,
|
| 35 |
+
"rewards/accuracies": 0.375,
|
| 36 |
+
"rewards/chosen": 0.006721611134707928,
|
| 37 |
+
"rewards/margins": -0.006841087248176336,
|
| 38 |
+
"rewards/rejected": 0.013562696985900402,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.06057546693589096,
|
| 43 |
+
"grad_norm": 42.148170471191406,
|
| 44 |
+
"learning_rate": 5e-06,
|
| 45 |
+
"logits/chosen": -1.0353584289550781,
|
| 46 |
+
"logits/rejected": -1.0312868356704712,
|
| 47 |
+
"logps/chosen": -145.601806640625,
|
| 48 |
+
"logps/rejected": -159.65481567382812,
|
| 49 |
+
"loss": 0.6928,
|
| 50 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 51 |
+
"rewards/chosen": 0.02560463175177574,
|
| 52 |
+
"rewards/margins": 0.0021010604687035084,
|
| 53 |
+
"rewards/rejected": 0.023503568023443222,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.08076728924785462,
|
| 58 |
+
"grad_norm": 58.297794342041016,
|
| 59 |
+
"learning_rate": 5e-06,
|
| 60 |
+
"logits/chosen": -1.053727388381958,
|
| 61 |
+
"logits/rejected": -1.0229905843734741,
|
| 62 |
+
"logps/chosen": -121.37943267822266,
|
| 63 |
+
"logps/rejected": -130.5764923095703,
|
| 64 |
+
"loss": 0.6857,
|
| 65 |
+
"rewards/accuracies": 0.574999988079071,
|
| 66 |
+
"rewards/chosen": 0.06899537146091461,
|
| 67 |
+
"rewards/margins": 0.015979710966348648,
|
| 68 |
+
"rewards/rejected": 0.053015656769275665,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.10095911155981828,
|
| 73 |
+
"grad_norm": 158.1420440673828,
|
| 74 |
+
"learning_rate": 5e-06,
|
| 75 |
+
"logits/chosen": -1.097388744354248,
|
| 76 |
+
"logits/rejected": -1.0624990463256836,
|
| 77 |
+
"logps/chosen": -126.20062255859375,
|
| 78 |
+
"logps/rejected": -140.00369262695312,
|
| 79 |
+
"loss": 0.6968,
|
| 80 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 81 |
+
"rewards/chosen": 0.040822066366672516,
|
| 82 |
+
"rewards/margins": -0.004479425959289074,
|
| 83 |
+
"rewards/rejected": 0.04530149698257446,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.12115093387178193,
|
| 88 |
+
"grad_norm": 111.64271545410156,
|
| 89 |
+
"learning_rate": 5e-06,
|
| 90 |
+
"logits/chosen": -0.9921103715896606,
|
| 91 |
+
"logits/rejected": -0.9481407999992371,
|
| 92 |
+
"logps/chosen": -133.66769409179688,
|
| 93 |
+
"logps/rejected": -149.52188110351562,
|
| 94 |
+
"loss": 0.7066,
|
| 95 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 96 |
+
"rewards/chosen": -0.17871007323265076,
|
| 97 |
+
"rewards/margins": -0.001442834734916687,
|
| 98 |
+
"rewards/rejected": -0.17726723849773407,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.1413427561837456,
|
| 103 |
+
"grad_norm": 35.10946273803711,
|
| 104 |
+
"learning_rate": 5e-06,
|
| 105 |
+
"logits/chosen": -0.9996519088745117,
|
| 106 |
+
"logits/rejected": -1.0022809505462646,
|
| 107 |
+
"logps/chosen": -142.79452514648438,
|
| 108 |
+
"logps/rejected": -142.8091583251953,
|
| 109 |
+
"loss": 0.6933,
|
| 110 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 111 |
+
"rewards/chosen": -0.1559412181377411,
|
| 112 |
+
"rewards/margins": 0.018214434385299683,
|
| 113 |
+
"rewards/rejected": -0.17415565252304077,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.16153457849570924,
|
| 118 |
+
"grad_norm": 154.33160400390625,
|
| 119 |
+
"learning_rate": 5e-06,
|
| 120 |
+
"logits/chosen": -1.0090668201446533,
|
| 121 |
+
"logits/rejected": -0.9780625104904175,
|
| 122 |
+
"logps/chosen": -124.8656234741211,
|
| 123 |
+
"logps/rejected": -151.53759765625,
|
| 124 |
+
"loss": 0.6991,
|
| 125 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 126 |
+
"rewards/chosen": -0.0610191710293293,
|
| 127 |
+
"rewards/margins": 0.001720429165288806,
|
| 128 |
+
"rewards/rejected": -0.06273959577083588,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.18172640080767288,
|
| 133 |
+
"grad_norm": 66.54533386230469,
|
| 134 |
+
"learning_rate": 5e-06,
|
| 135 |
+
"logits/chosen": -1.0511558055877686,
|
| 136 |
+
"logits/rejected": -0.9915176630020142,
|
| 137 |
+
"logps/chosen": -125.61368560791016,
|
| 138 |
+
"logps/rejected": -140.61505126953125,
|
| 139 |
+
"loss": 0.6762,
|
| 140 |
+
"rewards/accuracies": 0.5,
|
| 141 |
+
"rewards/chosen": -0.11788681894540787,
|
| 142 |
+
"rewards/margins": 0.049356214702129364,
|
| 143 |
+
"rewards/rejected": -0.16724303364753723,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.20191822311963656,
|
| 148 |
+
"grad_norm": 133.6022491455078,
|
| 149 |
+
"learning_rate": 5e-06,
|
| 150 |
+
"logits/chosen": -1.0434271097183228,
|
| 151 |
+
"logits/rejected": -1.0047496557235718,
|
| 152 |
+
"logps/chosen": -126.78190612792969,
|
| 153 |
+
"logps/rejected": -138.11805725097656,
|
| 154 |
+
"loss": 0.7026,
|
| 155 |
+
"rewards/accuracies": 0.375,
|
| 156 |
+
"rewards/chosen": -0.2226967066526413,
|
| 157 |
+
"rewards/margins": 0.006922038737684488,
|
| 158 |
+
"rewards/rejected": -0.22961874306201935,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.2221100454316002,
|
| 163 |
+
"grad_norm": 61.76374053955078,
|
| 164 |
+
"learning_rate": 5e-06,
|
| 165 |
+
"logits/chosen": -1.026590347290039,
|
| 166 |
+
"logits/rejected": -1.0285896062850952,
|
| 167 |
+
"logps/chosen": -126.12849426269531,
|
| 168 |
+
"logps/rejected": -142.56619262695312,
|
| 169 |
+
"loss": 0.6729,
|
| 170 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 171 |
+
"rewards/chosen": -0.26916319131851196,
|
| 172 |
+
"rewards/margins": 0.06655414402484894,
|
| 173 |
+
"rewards/rejected": -0.3357173502445221,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.24230186774356385,
|
| 178 |
+
"grad_norm": 61.80793380737305,
|
| 179 |
+
"learning_rate": 5e-06,
|
| 180 |
+
"logits/chosen": -1.0803353786468506,
|
| 181 |
+
"logits/rejected": -1.0326142311096191,
|
| 182 |
+
"logps/chosen": -126.38145446777344,
|
| 183 |
+
"logps/rejected": -143.6183319091797,
|
| 184 |
+
"loss": 0.6796,
|
| 185 |
+
"rewards/accuracies": 0.574999988079071,
|
| 186 |
+
"rewards/chosen": -0.36116331815719604,
|
| 187 |
+
"rewards/margins": 0.07047157734632492,
|
| 188 |
+
"rewards/rejected": -0.43163490295410156,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.26249369005552753,
|
| 193 |
+
"grad_norm": 63.28056716918945,
|
| 194 |
+
"learning_rate": 5e-06,
|
| 195 |
+
"logits/chosen": -0.9822731018066406,
|
| 196 |
+
"logits/rejected": -0.9656317830085754,
|
| 197 |
+
"logps/chosen": -127.26124572753906,
|
| 198 |
+
"logps/rejected": -142.61508178710938,
|
| 199 |
+
"loss": 0.6992,
|
| 200 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 201 |
+
"rewards/chosen": -0.36886388063430786,
|
| 202 |
+
"rewards/margins": 0.059492867439985275,
|
| 203 |
+
"rewards/rejected": -0.42835673689842224,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.2826855123674912,
|
| 208 |
+
"grad_norm": 40.66404724121094,
|
| 209 |
+
"learning_rate": 5e-06,
|
| 210 |
+
"logits/chosen": -1.0611032247543335,
|
| 211 |
+
"logits/rejected": -0.9986160397529602,
|
| 212 |
+
"logps/chosen": -119.45987701416016,
|
| 213 |
+
"logps/rejected": -137.6642303466797,
|
| 214 |
+
"loss": 0.635,
|
| 215 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 216 |
+
"rewards/chosen": -0.24067869782447815,
|
| 217 |
+
"rewards/margins": 0.19307085871696472,
|
| 218 |
+
"rewards/rejected": -0.4337495267391205,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.3028773346794548,
|
| 223 |
+
"grad_norm": 141.08375549316406,
|
| 224 |
+
"learning_rate": 5e-06,
|
| 225 |
+
"logits/chosen": -1.0468595027923584,
|
| 226 |
+
"logits/rejected": -1.0398485660552979,
|
| 227 |
+
"logps/chosen": -123.22066497802734,
|
| 228 |
+
"logps/rejected": -137.7001190185547,
|
| 229 |
+
"loss": 0.6822,
|
| 230 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 231 |
+
"rewards/chosen": -0.32747992873191833,
|
| 232 |
+
"rewards/margins": 0.04724354296922684,
|
| 233 |
+
"rewards/rejected": -0.37472349405288696,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.32306915699141847,
|
| 238 |
+
"grad_norm": 76.24655151367188,
|
| 239 |
+
"learning_rate": 5e-06,
|
| 240 |
+
"logits/chosen": -1.0108659267425537,
|
| 241 |
+
"logits/rejected": -0.9696664810180664,
|
| 242 |
+
"logps/chosen": -132.04534912109375,
|
| 243 |
+
"logps/rejected": -134.4063262939453,
|
| 244 |
+
"loss": 0.7582,
|
| 245 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 246 |
+
"rewards/chosen": -0.3479716181755066,
|
| 247 |
+
"rewards/margins": -0.06882990151643753,
|
| 248 |
+
"rewards/rejected": -0.27914175391197205,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.3432609793033821,
|
| 253 |
+
"grad_norm": 294.6365661621094,
|
| 254 |
+
"learning_rate": 5e-06,
|
| 255 |
+
"logits/chosen": -1.0133106708526611,
|
| 256 |
+
"logits/rejected": -1.0042521953582764,
|
| 257 |
+
"logps/chosen": -134.86880493164062,
|
| 258 |
+
"logps/rejected": -145.12660217285156,
|
| 259 |
+
"loss": 0.7659,
|
| 260 |
+
"rewards/accuracies": 0.5,
|
| 261 |
+
"rewards/chosen": -0.3877476453781128,
|
| 262 |
+
"rewards/margins": -0.0979350134730339,
|
| 263 |
+
"rewards/rejected": -0.2898126542568207,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.36345280161534577,
|
| 268 |
+
"grad_norm": 48.36005783081055,
|
| 269 |
+
"learning_rate": 5e-06,
|
| 270 |
+
"logits/chosen": -1.0785777568817139,
|
| 271 |
+
"logits/rejected": -1.0150744915008545,
|
| 272 |
+
"logps/chosen": -125.0882339477539,
|
| 273 |
+
"logps/rejected": -134.69866943359375,
|
| 274 |
+
"loss": 0.7113,
|
| 275 |
+
"rewards/accuracies": 0.5,
|
| 276 |
+
"rewards/chosen": -0.36325889825820923,
|
| 277 |
+
"rewards/margins": 0.047263432294130325,
|
| 278 |
+
"rewards/rejected": -0.41052237153053284,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.3836446239273094,
|
| 283 |
+
"grad_norm": 134.56393432617188,
|
| 284 |
+
"learning_rate": 5e-06,
|
| 285 |
+
"logits/chosen": -1.0220086574554443,
|
| 286 |
+
"logits/rejected": -1.0256017446517944,
|
| 287 |
+
"logps/chosen": -136.06663513183594,
|
| 288 |
+
"logps/rejected": -141.03758239746094,
|
| 289 |
+
"loss": 0.7582,
|
| 290 |
+
"rewards/accuracies": 0.5,
|
| 291 |
+
"rewards/chosen": -0.5725785493850708,
|
| 292 |
+
"rewards/margins": -0.02296329475939274,
|
| 293 |
+
"rewards/rejected": -0.549615204334259,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.4038364462392731,
|
| 298 |
+
"grad_norm": 192.60671997070312,
|
| 299 |
+
"learning_rate": 5e-06,
|
| 300 |
+
"logits/chosen": -1.029658555984497,
|
| 301 |
+
"logits/rejected": -1.015587568283081,
|
| 302 |
+
"logps/chosen": -135.1102752685547,
|
| 303 |
+
"logps/rejected": -132.73965454101562,
|
| 304 |
+
"loss": 0.7576,
|
| 305 |
+
"rewards/accuracies": 0.574999988079071,
|
| 306 |
+
"rewards/chosen": -0.6215102076530457,
|
| 307 |
+
"rewards/margins": -0.022445138543844223,
|
| 308 |
+
"rewards/rejected": -0.5990650653839111,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.42402826855123676,
|
| 313 |
+
"grad_norm": 42.97038269042969,
|
| 314 |
+
"learning_rate": 5e-06,
|
| 315 |
+
"logits/chosen": -1.01835036277771,
|
| 316 |
+
"logits/rejected": -0.9691454768180847,
|
| 317 |
+
"logps/chosen": -127.7636489868164,
|
| 318 |
+
"logps/rejected": -131.01290893554688,
|
| 319 |
+
"loss": 0.8087,
|
| 320 |
+
"rewards/accuracies": 0.550000011920929,
|
| 321 |
+
"rewards/chosen": -0.43060892820358276,
|
| 322 |
+
"rewards/margins": -0.1305285394191742,
|
| 323 |
+
"rewards/rejected": -0.30008038878440857,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.4442200908632004,
|
| 328 |
+
"grad_norm": 73.89049530029297,
|
| 329 |
+
"learning_rate": 5e-06,
|
| 330 |
+
"logits/chosen": -1.0535072088241577,
|
| 331 |
+
"logits/rejected": -1.0358234643936157,
|
| 332 |
+
"logps/chosen": -131.6114959716797,
|
| 333 |
+
"logps/rejected": -148.74351501464844,
|
| 334 |
+
"loss": 0.6739,
|
| 335 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 336 |
+
"rewards/chosen": -0.319467693567276,
|
| 337 |
+
"rewards/margins": 0.11042511463165283,
|
| 338 |
+
"rewards/rejected": -0.4298928380012512,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.46441191317516406,
|
| 343 |
+
"grad_norm": 261.6590881347656,
|
| 344 |
+
"learning_rate": 5e-06,
|
| 345 |
+
"logits/chosen": -1.039872407913208,
|
| 346 |
+
"logits/rejected": -1.061729073524475,
|
| 347 |
+
"logps/chosen": -133.2412567138672,
|
| 348 |
+
"logps/rejected": -141.62680053710938,
|
| 349 |
+
"loss": 0.8194,
|
| 350 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 351 |
+
"rewards/chosen": -0.5433087348937988,
|
| 352 |
+
"rewards/margins": -0.15157613158226013,
|
| 353 |
+
"rewards/rejected": -0.3917326033115387,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.4846037354871277,
|
| 358 |
+
"grad_norm": 153.96530151367188,
|
| 359 |
+
"learning_rate": 5e-06,
|
| 360 |
+
"logits/chosen": -1.0134645700454712,
|
| 361 |
+
"logits/rejected": -1.0050514936447144,
|
| 362 |
+
"logps/chosen": -140.56578063964844,
|
| 363 |
+
"logps/rejected": -146.6461944580078,
|
| 364 |
+
"loss": 0.7222,
|
| 365 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 366 |
+
"rewards/chosen": -0.4193994104862213,
|
| 367 |
+
"rewards/margins": -0.014027280732989311,
|
| 368 |
+
"rewards/rejected": -0.40537214279174805,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.5047955577990914,
|
| 373 |
+
"grad_norm": 92.5348129272461,
|
| 374 |
+
"learning_rate": 5e-06,
|
| 375 |
+
"logits/chosen": -1.0472946166992188,
|
| 376 |
+
"logits/rejected": -1.0149815082550049,
|
| 377 |
+
"logps/chosen": -126.81514739990234,
|
| 378 |
+
"logps/rejected": -138.25100708007812,
|
| 379 |
+
"loss": 0.641,
|
| 380 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 381 |
+
"rewards/chosen": -0.2987712025642395,
|
| 382 |
+
"rewards/margins": 0.14457064867019653,
|
| 383 |
+
"rewards/rejected": -0.44334182143211365,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.5249873801110551,
|
| 388 |
+
"grad_norm": 69.95176696777344,
|
| 389 |
+
"learning_rate": 5e-06,
|
| 390 |
+
"logits/chosen": -0.9726712107658386,
|
| 391 |
+
"logits/rejected": -0.9396511316299438,
|
| 392 |
+
"logps/chosen": -134.06784057617188,
|
| 393 |
+
"logps/rejected": -149.26393127441406,
|
| 394 |
+
"loss": 0.6874,
|
| 395 |
+
"rewards/accuracies": 0.574999988079071,
|
| 396 |
+
"rewards/chosen": -0.4227423071861267,
|
| 397 |
+
"rewards/margins": 0.07707767188549042,
|
| 398 |
+
"rewards/rejected": -0.4998199939727783,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.5451792024230186,
|
| 403 |
+
"grad_norm": 59.02754592895508,
|
| 404 |
+
"learning_rate": 5e-06,
|
| 405 |
+
"logits/chosen": -1.0336484909057617,
|
| 406 |
+
"logits/rejected": -1.0117156505584717,
|
| 407 |
+
"logps/chosen": -123.12812805175781,
|
| 408 |
+
"logps/rejected": -138.68785095214844,
|
| 409 |
+
"loss": 0.6972,
|
| 410 |
+
"rewards/accuracies": 0.574999988079071,
|
| 411 |
+
"rewards/chosen": -0.2850368916988373,
|
| 412 |
+
"rewards/margins": 0.08022015541791916,
|
| 413 |
+
"rewards/rejected": -0.3652570843696594,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.5653710247349824,
|
| 418 |
+
"grad_norm": 164.89817810058594,
|
| 419 |
+
"learning_rate": 5e-06,
|
| 420 |
+
"logits/chosen": -0.9582400321960449,
|
| 421 |
+
"logits/rejected": -0.9342845678329468,
|
| 422 |
+
"logps/chosen": -136.7896270751953,
|
| 423 |
+
"logps/rejected": -144.94981384277344,
|
| 424 |
+
"loss": 0.8658,
|
| 425 |
+
"rewards/accuracies": 0.5,
|
| 426 |
+
"rewards/chosen": -0.5358542799949646,
|
| 427 |
+
"rewards/margins": -0.13312572240829468,
|
| 428 |
+
"rewards/rejected": -0.40272849798202515,
|
| 429 |
+
"step": 140
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.585562847046946,
|
| 433 |
+
"grad_norm": 273.0749816894531,
|
| 434 |
+
"learning_rate": 5e-06,
|
| 435 |
+
"logits/chosen": -1.0799394845962524,
|
| 436 |
+
"logits/rejected": -1.0522867441177368,
|
| 437 |
+
"logps/chosen": -133.341796875,
|
| 438 |
+
"logps/rejected": -137.64601135253906,
|
| 439 |
+
"loss": 0.7,
|
| 440 |
+
"rewards/accuracies": 0.550000011920929,
|
| 441 |
+
"rewards/chosen": -0.5455681681632996,
|
| 442 |
+
"rewards/margins": 0.1389351338148117,
|
| 443 |
+
"rewards/rejected": -0.6845033168792725,
|
| 444 |
+
"step": 145
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.6057546693589096,
|
| 448 |
+
"grad_norm": 58.67772674560547,
|
| 449 |
+
"learning_rate": 5e-06,
|
| 450 |
+
"logits/chosen": -1.0716097354888916,
|
| 451 |
+
"logits/rejected": -1.0067239999771118,
|
| 452 |
+
"logps/chosen": -131.79644775390625,
|
| 453 |
+
"logps/rejected": -145.97897338867188,
|
| 454 |
+
"loss": 0.6711,
|
| 455 |
+
"rewards/accuracies": 0.550000011920929,
|
| 456 |
+
"rewards/chosen": -0.48813509941101074,
|
| 457 |
+
"rewards/margins": 0.15445764362812042,
|
| 458 |
+
"rewards/rejected": -0.64259272813797,
|
| 459 |
+
"step": 150
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.6259464916708734,
|
| 463 |
+
"grad_norm": 153.71763610839844,
|
| 464 |
+
"learning_rate": 5e-06,
|
| 465 |
+
"logits/chosen": -0.9819652438163757,
|
| 466 |
+
"logits/rejected": -0.975143551826477,
|
| 467 |
+
"logps/chosen": -131.63906860351562,
|
| 468 |
+
"logps/rejected": -136.08070373535156,
|
| 469 |
+
"loss": 0.6992,
|
| 470 |
+
"rewards/accuracies": 0.550000011920929,
|
| 471 |
+
"rewards/chosen": -0.511472761631012,
|
| 472 |
+
"rewards/margins": 0.09871827065944672,
|
| 473 |
+
"rewards/rejected": -0.6101909875869751,
|
| 474 |
+
"step": 155
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.6461383139828369,
|
| 478 |
+
"grad_norm": 140.51803588867188,
|
| 479 |
+
"learning_rate": 5e-06,
|
| 480 |
+
"logits/chosen": -0.9384062886238098,
|
| 481 |
+
"logits/rejected": -0.9456030130386353,
|
| 482 |
+
"logps/chosen": -137.96107482910156,
|
| 483 |
+
"logps/rejected": -148.41708374023438,
|
| 484 |
+
"loss": 0.6837,
|
| 485 |
+
"rewards/accuracies": 0.574999988079071,
|
| 486 |
+
"rewards/chosen": -0.5771997570991516,
|
| 487 |
+
"rewards/margins": 0.1888493001461029,
|
| 488 |
+
"rewards/rejected": -0.7660490274429321,
|
| 489 |
+
"step": 160
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.6663301362948006,
|
| 493 |
+
"grad_norm": 115.60352325439453,
|
| 494 |
+
"learning_rate": 5e-06,
|
| 495 |
+
"logits/chosen": -1.032671332359314,
|
| 496 |
+
"logits/rejected": -1.032019853591919,
|
| 497 |
+
"logps/chosen": -136.2175750732422,
|
| 498 |
+
"logps/rejected": -139.32192993164062,
|
| 499 |
+
"loss": 0.7872,
|
| 500 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 501 |
+
"rewards/chosen": -0.6546095609664917,
|
| 502 |
+
"rewards/margins": -0.08018063008785248,
|
| 503 |
+
"rewards/rejected": -0.5744289755821228,
|
| 504 |
+
"step": 165
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.6865219586067642,
|
| 508 |
+
"grad_norm": 79.54869079589844,
|
| 509 |
+
"learning_rate": 5e-06,
|
| 510 |
+
"logits/chosen": -1.0461210012435913,
|
| 511 |
+
"logits/rejected": -1.001206398010254,
|
| 512 |
+
"logps/chosen": -133.0095672607422,
|
| 513 |
+
"logps/rejected": -142.75643920898438,
|
| 514 |
+
"loss": 0.6265,
|
| 515 |
+
"rewards/accuracies": 0.675000011920929,
|
| 516 |
+
"rewards/chosen": -0.38921013474464417,
|
| 517 |
+
"rewards/margins": 0.21428516507148743,
|
| 518 |
+
"rewards/rejected": -0.6034952998161316,
|
| 519 |
+
"step": 170
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.7067137809187279,
|
| 523 |
+
"grad_norm": 124.25992584228516,
|
| 524 |
+
"learning_rate": 5e-06,
|
| 525 |
+
"logits/chosen": -1.0024333000183105,
|
| 526 |
+
"logits/rejected": -1.0128402709960938,
|
| 527 |
+
"logps/chosen": -134.9090576171875,
|
| 528 |
+
"logps/rejected": -141.35768127441406,
|
| 529 |
+
"loss": 0.6573,
|
| 530 |
+
"rewards/accuracies": 0.625,
|
| 531 |
+
"rewards/chosen": -0.43251633644104004,
|
| 532 |
+
"rewards/margins": 0.22649070620536804,
|
| 533 |
+
"rewards/rejected": -0.6590070128440857,
|
| 534 |
+
"step": 175
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.7269056032306915,
|
| 538 |
+
"grad_norm": 90.11141204833984,
|
| 539 |
+
"learning_rate": 5e-06,
|
| 540 |
+
"logits/chosen": -1.0318266153335571,
|
| 541 |
+
"logits/rejected": -0.9781469106674194,
|
| 542 |
+
"logps/chosen": -128.78518676757812,
|
| 543 |
+
"logps/rejected": -147.6506805419922,
|
| 544 |
+
"loss": 0.7165,
|
| 545 |
+
"rewards/accuracies": 0.550000011920929,
|
| 546 |
+
"rewards/chosen": -0.4871068596839905,
|
| 547 |
+
"rewards/margins": 0.07786975055932999,
|
| 548 |
+
"rewards/rejected": -0.5649765729904175,
|
| 549 |
+
"step": 180
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.7470974255426552,
|
| 553 |
+
"grad_norm": 199.0548858642578,
|
| 554 |
+
"learning_rate": 5e-06,
|
| 555 |
+
"logits/chosen": -0.984387993812561,
|
| 556 |
+
"logits/rejected": -0.9830384254455566,
|
| 557 |
+
"logps/chosen": -132.85769653320312,
|
| 558 |
+
"logps/rejected": -133.11199951171875,
|
| 559 |
+
"loss": 0.8698,
|
| 560 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 561 |
+
"rewards/chosen": -0.5072764158248901,
|
| 562 |
+
"rewards/margins": -0.2219184935092926,
|
| 563 |
+
"rewards/rejected": -0.28535789251327515,
|
| 564 |
+
"step": 185
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.7672892478546188,
|
| 568 |
+
"grad_norm": 260.7405090332031,
|
| 569 |
+
"learning_rate": 5e-06,
|
| 570 |
+
"logits/chosen": -1.0293523073196411,
|
| 571 |
+
"logits/rejected": -1.0172580480575562,
|
| 572 |
+
"logps/chosen": -126.0751724243164,
|
| 573 |
+
"logps/rejected": -141.263671875,
|
| 574 |
+
"loss": 0.6471,
|
| 575 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 576 |
+
"rewards/chosen": -0.25228002667427063,
|
| 577 |
+
"rewards/margins": 0.1569499969482422,
|
| 578 |
+
"rewards/rejected": -0.4092300534248352,
|
| 579 |
+
"step": 190
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.7874810701665825,
|
| 583 |
+
"grad_norm": 41.43046569824219,
|
| 584 |
+
"learning_rate": 5e-06,
|
| 585 |
+
"logits/chosen": -1.071702480316162,
|
| 586 |
+
"logits/rejected": -1.0639045238494873,
|
| 587 |
+
"logps/chosen": -124.64006042480469,
|
| 588 |
+
"logps/rejected": -129.1591796875,
|
| 589 |
+
"loss": 0.6881,
|
| 590 |
+
"rewards/accuracies": 0.5,
|
| 591 |
+
"rewards/chosen": -0.2144135981798172,
|
| 592 |
+
"rewards/margins": 0.07804916799068451,
|
| 593 |
+
"rewards/rejected": -0.2924627363681793,
|
| 594 |
+
"step": 195
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.8076728924785462,
|
| 598 |
+
"grad_norm": 35.1358528137207,
|
| 599 |
+
"learning_rate": 5e-06,
|
| 600 |
+
"logits/chosen": -0.9687834978103638,
|
| 601 |
+
"logits/rejected": -0.954856276512146,
|
| 602 |
+
"logps/chosen": -120.9579849243164,
|
| 603 |
+
"logps/rejected": -130.809814453125,
|
| 604 |
+
"loss": 0.7126,
|
| 605 |
+
"rewards/accuracies": 0.550000011920929,
|
| 606 |
+
"rewards/chosen": -0.27170878648757935,
|
| 607 |
+
"rewards/margins": 0.005672874394804239,
|
| 608 |
+
"rewards/rejected": -0.27738165855407715,
|
| 609 |
+
"step": 200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.8278647147905098,
|
| 613 |
+
"grad_norm": 27.78827667236328,
|
| 614 |
+
"learning_rate": 5e-06,
|
| 615 |
+
"logits/chosen": -1.048870325088501,
|
| 616 |
+
"logits/rejected": -1.0687897205352783,
|
| 617 |
+
"logps/chosen": -128.82345581054688,
|
| 618 |
+
"logps/rejected": -132.6963348388672,
|
| 619 |
+
"loss": 0.7058,
|
| 620 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 621 |
+
"rewards/chosen": -0.2189714014530182,
|
| 622 |
+
"rewards/margins": 0.05474821850657463,
|
| 623 |
+
"rewards/rejected": -0.2737196087837219,
|
| 624 |
+
"step": 205
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.8480565371024735,
|
| 628 |
+
"grad_norm": 101.58830261230469,
|
| 629 |
+
"learning_rate": 5e-06,
|
| 630 |
+
"logits/chosen": -1.0325753688812256,
|
| 631 |
+
"logits/rejected": -1.0432530641555786,
|
| 632 |
+
"logps/chosen": -126.06434631347656,
|
| 633 |
+
"logps/rejected": -136.6858673095703,
|
| 634 |
+
"loss": 0.7424,
|
| 635 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 636 |
+
"rewards/chosen": -0.38726571202278137,
|
| 637 |
+
"rewards/margins": -0.062184691429138184,
|
| 638 |
+
"rewards/rejected": -0.3250810503959656,
|
| 639 |
+
"step": 210
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.8682483594144371,
|
| 643 |
+
"grad_norm": 145.8494110107422,
|
| 644 |
+
"learning_rate": 5e-06,
|
| 645 |
+
"logits/chosen": -1.0436766147613525,
|
| 646 |
+
"logits/rejected": -1.001501441001892,
|
| 647 |
+
"logps/chosen": -128.81761169433594,
|
| 648 |
+
"logps/rejected": -146.7082977294922,
|
| 649 |
+
"loss": 0.649,
|
| 650 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 651 |
+
"rewards/chosen": -0.27072566747665405,
|
| 652 |
+
"rewards/margins": 0.1860719621181488,
|
| 653 |
+
"rewards/rejected": -0.45679759979248047,
|
| 654 |
+
"step": 215
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.8884401817264008,
|
| 658 |
+
"grad_norm": 122.89201354980469,
|
| 659 |
+
"learning_rate": 5e-06,
|
| 660 |
+
"logits/chosen": -0.9502097964286804,
|
| 661 |
+
"logits/rejected": -0.9250322580337524,
|
| 662 |
+
"logps/chosen": -136.70542907714844,
|
| 663 |
+
"logps/rejected": -145.50021362304688,
|
| 664 |
+
"loss": 0.7171,
|
| 665 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 666 |
+
"rewards/chosen": -0.4370450973510742,
|
| 667 |
+
"rewards/margins": 0.014479175209999084,
|
| 668 |
+
"rewards/rejected": -0.4515243172645569,
|
| 669 |
+
"step": 220
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.9086320040383644,
|
| 673 |
+
"grad_norm": 136.3062286376953,
|
| 674 |
+
"learning_rate": 5e-06,
|
| 675 |
+
"logits/chosen": -1.0122795104980469,
|
| 676 |
+
"logits/rejected": -0.976104736328125,
|
| 677 |
+
"logps/chosen": -130.60975646972656,
|
| 678 |
+
"logps/rejected": -140.47096252441406,
|
| 679 |
+
"loss": 0.7608,
|
| 680 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 681 |
+
"rewards/chosen": -0.3417353332042694,
|
| 682 |
+
"rewards/margins": -0.07229948043823242,
|
| 683 |
+
"rewards/rejected": -0.269435852766037,
|
| 684 |
+
"step": 225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.9288238263503281,
|
| 688 |
+
"grad_norm": 274.8150329589844,
|
| 689 |
+
"learning_rate": 5e-06,
|
| 690 |
+
"logits/chosen": -0.9934455156326294,
|
| 691 |
+
"logits/rejected": -0.9960210919380188,
|
| 692 |
+
"logps/chosen": -138.07879638671875,
|
| 693 |
+
"logps/rejected": -148.25611877441406,
|
| 694 |
+
"loss": 0.7518,
|
| 695 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 696 |
+
"rewards/chosen": -0.3778955340385437,
|
| 697 |
+
"rewards/margins": -0.06377332657575607,
|
| 698 |
+
"rewards/rejected": -0.3141222298145294,
|
| 699 |
+
"step": 230
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.9490156486622918,
|
| 703 |
+
"grad_norm": 362.4608154296875,
|
| 704 |
+
"learning_rate": 5e-06,
|
| 705 |
+
"logits/chosen": -1.028414249420166,
|
| 706 |
+
"logits/rejected": -1.0202635526657104,
|
| 707 |
+
"logps/chosen": -126.71832275390625,
|
| 708 |
+
"logps/rejected": -131.26434326171875,
|
| 709 |
+
"loss": 0.7528,
|
| 710 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 711 |
+
"rewards/chosen": -0.10429117828607559,
|
| 712 |
+
"rewards/margins": -0.09285052120685577,
|
| 713 |
+
"rewards/rejected": -0.011440658941864967,
|
| 714 |
+
"step": 235
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.9692074709742554,
|
| 718 |
+
"grad_norm": 200.3191375732422,
|
| 719 |
+
"learning_rate": 5e-06,
|
| 720 |
+
"logits/chosen": -1.0229647159576416,
|
| 721 |
+
"logits/rejected": -0.9428025484085083,
|
| 722 |
+
"logps/chosen": -126.55399322509766,
|
| 723 |
+
"logps/rejected": -151.26348876953125,
|
| 724 |
+
"loss": 0.6815,
|
| 725 |
+
"rewards/accuracies": 0.550000011920929,
|
| 726 |
+
"rewards/chosen": 0.17346206307411194,
|
| 727 |
+
"rewards/margins": 0.028347322717308998,
|
| 728 |
+
"rewards/rejected": 0.1451147496700287,
|
| 729 |
+
"step": 240
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.9893992932862191,
|
| 733 |
+
"grad_norm": 34.3711051940918,
|
| 734 |
+
"learning_rate": 5e-06,
|
| 735 |
+
"logits/chosen": -1.0116575956344604,
|
| 736 |
+
"logits/rejected": -1.0087472200393677,
|
| 737 |
+
"logps/chosen": -122.85481262207031,
|
| 738 |
+
"logps/rejected": -133.4599609375,
|
| 739 |
+
"loss": 0.6786,
|
| 740 |
+
"rewards/accuracies": 0.699999988079071,
|
| 741 |
+
"rewards/chosen": 0.08724311739206314,
|
| 742 |
+
"rewards/margins": 0.0308841522783041,
|
| 743 |
+
"rewards/rejected": 0.05635897070169449,
|
| 744 |
+
"step": 245
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.0080767289247854,
|
| 748 |
+
"grad_norm": 27.31838607788086,
|
| 749 |
+
"learning_rate": 5e-06,
|
| 750 |
+
"logits/chosen": -1.0463143587112427,
|
| 751 |
+
"logits/rejected": -0.9973466396331787,
|
| 752 |
+
"logps/chosen": -124.2419204711914,
|
| 753 |
+
"logps/rejected": -139.740966796875,
|
| 754 |
+
"loss": 0.6114,
|
| 755 |
+
"rewards/accuracies": 0.5945945978164673,
|
| 756 |
+
"rewards/chosen": 0.005640762392431498,
|
| 757 |
+
"rewards/margins": 0.07190810143947601,
|
| 758 |
+
"rewards/rejected": -0.06626733392477036,
|
| 759 |
+
"step": 250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.028268551236749,
|
| 763 |
+
"grad_norm": 36.600242614746094,
|
| 764 |
+
"learning_rate": 5e-06,
|
| 765 |
+
"logits/chosen": -0.9850190877914429,
|
| 766 |
+
"logits/rejected": -0.9895715713500977,
|
| 767 |
+
"logps/chosen": -117.77237701416016,
|
| 768 |
+
"logps/rejected": -139.15866088867188,
|
| 769 |
+
"loss": 0.623,
|
| 770 |
+
"rewards/accuracies": 0.800000011920929,
|
| 771 |
+
"rewards/chosen": 0.06336753815412521,
|
| 772 |
+
"rewards/margins": 0.15559127926826477,
|
| 773 |
+
"rewards/rejected": -0.09222372621297836,
|
| 774 |
+
"step": 255
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.0484603735487128,
|
| 778 |
+
"grad_norm": 25.334484100341797,
|
| 779 |
+
"learning_rate": 5e-06,
|
| 780 |
+
"logits/chosen": -0.9737665057182312,
|
| 781 |
+
"logits/rejected": -0.9691070318222046,
|
| 782 |
+
"logps/chosen": -125.85533142089844,
|
| 783 |
+
"logps/rejected": -142.1818084716797,
|
| 784 |
+
"loss": 0.6441,
|
| 785 |
+
"rewards/accuracies": 0.75,
|
| 786 |
+
"rewards/chosen": 0.0201385710388422,
|
| 787 |
+
"rewards/margins": 0.10802127420902252,
|
| 788 |
+
"rewards/rejected": -0.08788268268108368,
|
| 789 |
+
"step": 260
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 1.0686521958606765,
|
| 793 |
+
"grad_norm": 170.07984924316406,
|
| 794 |
+
"learning_rate": 5e-06,
|
| 795 |
+
"logits/chosen": -0.9972125887870789,
|
| 796 |
+
"logits/rejected": -0.9916419982910156,
|
| 797 |
+
"logps/chosen": -128.32261657714844,
|
| 798 |
+
"logps/rejected": -154.72247314453125,
|
| 799 |
+
"loss": 0.6211,
|
| 800 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 801 |
+
"rewards/chosen": 0.003512249793857336,
|
| 802 |
+
"rewards/margins": 0.1796935498714447,
|
| 803 |
+
"rewards/rejected": -0.17618130147457123,
|
| 804 |
+
"step": 265
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.08884401817264,
|
| 808 |
+
"grad_norm": 82.89457702636719,
|
| 809 |
+
"learning_rate": 5e-06,
|
| 810 |
+
"logits/chosen": -0.9838020205497742,
|
| 811 |
+
"logits/rejected": -0.9610651135444641,
|
| 812 |
+
"logps/chosen": -133.44190979003906,
|
| 813 |
+
"logps/rejected": -135.03680419921875,
|
| 814 |
+
"loss": 0.6226,
|
| 815 |
+
"rewards/accuracies": 0.824999988079071,
|
| 816 |
+
"rewards/chosen": 0.2014409303665161,
|
| 817 |
+
"rewards/margins": 0.15444937348365784,
|
| 818 |
+
"rewards/rejected": 0.04699156433343887,
|
| 819 |
+
"step": 270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 1.1090358404846037,
|
| 823 |
+
"grad_norm": 256.55047607421875,
|
| 824 |
+
"learning_rate": 5e-06,
|
| 825 |
+
"logits/chosen": -1.0722827911376953,
|
| 826 |
+
"logits/rejected": -1.0403627157211304,
|
| 827 |
+
"logps/chosen": -129.58506774902344,
|
| 828 |
+
"logps/rejected": -135.889892578125,
|
| 829 |
+
"loss": 0.6594,
|
| 830 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 831 |
+
"rewards/chosen": 0.1928495317697525,
|
| 832 |
+
"rewards/margins": 0.09002382308244705,
|
| 833 |
+
"rewards/rejected": 0.10282570123672485,
|
| 834 |
+
"step": 275
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 1.1292276627965674,
|
| 838 |
+
"grad_norm": 157.06471252441406,
|
| 839 |
+
"learning_rate": 5e-06,
|
| 840 |
+
"logits/chosen": -1.0328292846679688,
|
| 841 |
+
"logits/rejected": -1.0051934719085693,
|
| 842 |
+
"logps/chosen": -126.65974426269531,
|
| 843 |
+
"logps/rejected": -137.65618896484375,
|
| 844 |
+
"loss": 0.6243,
|
| 845 |
+
"rewards/accuracies": 0.675000011920929,
|
| 846 |
+
"rewards/chosen": 0.24083271622657776,
|
| 847 |
+
"rewards/margins": 0.17111970484256744,
|
| 848 |
+
"rewards/rejected": 0.06971298158168793,
|
| 849 |
+
"step": 280
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.149419485108531,
|
| 853 |
+
"grad_norm": 318.4048767089844,
|
| 854 |
+
"learning_rate": 5e-06,
|
| 855 |
+
"logits/chosen": -1.0038608312606812,
|
| 856 |
+
"logits/rejected": -0.9795931577682495,
|
| 857 |
+
"logps/chosen": -119.45623779296875,
|
| 858 |
+
"logps/rejected": -147.02752685546875,
|
| 859 |
+
"loss": 0.5933,
|
| 860 |
+
"rewards/accuracies": 0.75,
|
| 861 |
+
"rewards/chosen": 0.10432057082653046,
|
| 862 |
+
"rewards/margins": 0.235098198056221,
|
| 863 |
+
"rewards/rejected": -0.13077762722969055,
|
| 864 |
+
"step": 285
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 1.1696113074204948,
|
| 868 |
+
"grad_norm": 111.46747589111328,
|
| 869 |
+
"learning_rate": 5e-06,
|
| 870 |
+
"logits/chosen": -1.0481410026550293,
|
| 871 |
+
"logits/rejected": -1.0197895765304565,
|
| 872 |
+
"logps/chosen": -132.78433227539062,
|
| 873 |
+
"logps/rejected": -139.17845153808594,
|
| 874 |
+
"loss": 0.615,
|
| 875 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 876 |
+
"rewards/chosen": -0.008489435538649559,
|
| 877 |
+
"rewards/margins": 0.21404604613780975,
|
| 878 |
+
"rewards/rejected": -0.22253546118736267,
|
| 879 |
+
"step": 290
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.1898031297324583,
|
| 883 |
+
"grad_norm": 118.42349243164062,
|
| 884 |
+
"learning_rate": 5e-06,
|
| 885 |
+
"logits/chosen": -0.9705491065979004,
|
| 886 |
+
"logits/rejected": -0.969982922077179,
|
| 887 |
+
"logps/chosen": -127.90840911865234,
|
| 888 |
+
"logps/rejected": -132.5115509033203,
|
| 889 |
+
"loss": 0.6599,
|
| 890 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 891 |
+
"rewards/chosen": -0.07771094888448715,
|
| 892 |
+
"rewards/margins": 0.09617092460393906,
|
| 893 |
+
"rewards/rejected": -0.1738818734884262,
|
| 894 |
+
"step": 295
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 1.209994952044422,
|
| 898 |
+
"grad_norm": 28.792146682739258,
|
| 899 |
+
"learning_rate": 5e-06,
|
| 900 |
+
"logits/chosen": -1.0062731504440308,
|
| 901 |
+
"logits/rejected": -1.0046422481536865,
|
| 902 |
+
"logps/chosen": -122.89021301269531,
|
| 903 |
+
"logps/rejected": -128.922119140625,
|
| 904 |
+
"loss": 0.6536,
|
| 905 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 906 |
+
"rewards/chosen": -0.21470877528190613,
|
| 907 |
+
"rewards/margins": 0.10167400538921356,
|
| 908 |
+
"rewards/rejected": -0.31638282537460327,
|
| 909 |
+
"step": 300
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1.2301867743563857,
|
| 913 |
+
"grad_norm": 30.531999588012695,
|
| 914 |
+
"learning_rate": 5e-06,
|
| 915 |
+
"logits/chosen": -1.0686832666397095,
|
| 916 |
+
"logits/rejected": -0.9891031384468079,
|
| 917 |
+
"logps/chosen": -131.0304718017578,
|
| 918 |
+
"logps/rejected": -155.66464233398438,
|
| 919 |
+
"loss": 0.6276,
|
| 920 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 921 |
+
"rewards/chosen": -0.22880509495735168,
|
| 922 |
+
"rewards/margins": 0.17476414144039154,
|
| 923 |
+
"rewards/rejected": -0.40356922149658203,
|
| 924 |
+
"step": 305
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 1.2503785966683494,
|
| 928 |
+
"grad_norm": 30.640419006347656,
|
| 929 |
+
"learning_rate": 5e-06,
|
| 930 |
+
"logits/chosen": -1.0535386800765991,
|
| 931 |
+
"logits/rejected": -1.0263631343841553,
|
| 932 |
+
"logps/chosen": -131.8798828125,
|
| 933 |
+
"logps/rejected": -134.03146362304688,
|
| 934 |
+
"loss": 0.6779,
|
| 935 |
+
"rewards/accuracies": 0.550000011920929,
|
| 936 |
+
"rewards/chosen": -0.29857125878334045,
|
| 937 |
+
"rewards/margins": 0.08670302480459213,
|
| 938 |
+
"rewards/rejected": -0.3852742612361908,
|
| 939 |
+
"step": 310
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1.2705704189803129,
|
| 943 |
+
"grad_norm": 66.49352264404297,
|
| 944 |
+
"learning_rate": 5e-06,
|
| 945 |
+
"logits/chosen": -1.0025132894515991,
|
| 946 |
+
"logits/rejected": -1.0053356885910034,
|
| 947 |
+
"logps/chosen": -141.34634399414062,
|
| 948 |
+
"logps/rejected": -155.73513793945312,
|
| 949 |
+
"loss": 0.5846,
|
| 950 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 951 |
+
"rewards/chosen": -0.12694263458251953,
|
| 952 |
+
"rewards/margins": 0.32664644718170166,
|
| 953 |
+
"rewards/rejected": -0.4535890519618988,
|
| 954 |
+
"step": 315
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.2907622412922766,
|
| 958 |
+
"grad_norm": 426.6810302734375,
|
| 959 |
+
"learning_rate": 5e-06,
|
| 960 |
+
"logits/chosen": -0.9757963418960571,
|
| 961 |
+
"logits/rejected": -0.971258819103241,
|
| 962 |
+
"logps/chosen": -133.66746520996094,
|
| 963 |
+
"logps/rejected": -142.61044311523438,
|
| 964 |
+
"loss": 0.6566,
|
| 965 |
+
"rewards/accuracies": 0.675000011920929,
|
| 966 |
+
"rewards/chosen": -0.2488606721162796,
|
| 967 |
+
"rewards/margins": 0.13673368096351624,
|
| 968 |
+
"rewards/rejected": -0.38559430837631226,
|
| 969 |
+
"step": 320
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1.3109540636042403,
|
| 973 |
+
"grad_norm": 24.37104034423828,
|
| 974 |
+
"learning_rate": 5e-06,
|
| 975 |
+
"logits/chosen": -1.0103740692138672,
|
| 976 |
+
"logits/rejected": -0.9810827374458313,
|
| 977 |
+
"logps/chosen": -125.60025787353516,
|
| 978 |
+
"logps/rejected": -135.81178283691406,
|
| 979 |
+
"loss": 0.6212,
|
| 980 |
+
"rewards/accuracies": 0.675000011920929,
|
| 981 |
+
"rewards/chosen": -0.18190403282642365,
|
| 982 |
+
"rewards/margins": 0.23106291890144348,
|
| 983 |
+
"rewards/rejected": -0.4129669666290283,
|
| 984 |
+
"step": 325
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 1.331145885916204,
|
| 988 |
+
"grad_norm": 25.101844787597656,
|
| 989 |
+
"learning_rate": 5e-06,
|
| 990 |
+
"logits/chosen": -0.9837746620178223,
|
| 991 |
+
"logits/rejected": -0.9348167181015015,
|
| 992 |
+
"logps/chosen": -123.2785415649414,
|
| 993 |
+
"logps/rejected": -138.2392578125,
|
| 994 |
+
"loss": 0.608,
|
| 995 |
+
"rewards/accuracies": 0.675000011920929,
|
| 996 |
+
"rewards/chosen": -0.1881856620311737,
|
| 997 |
+
"rewards/margins": 0.23057310283184052,
|
| 998 |
+
"rewards/rejected": -0.4187587797641754,
|
| 999 |
+
"step": 330
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 1.3513377082281677,
|
| 1003 |
+
"grad_norm": 33.728919982910156,
|
| 1004 |
+
"learning_rate": 5e-06,
|
| 1005 |
+
"logits/chosen": -0.9812191724777222,
|
| 1006 |
+
"logits/rejected": -0.960713267326355,
|
| 1007 |
+
"logps/chosen": -135.54893493652344,
|
| 1008 |
+
"logps/rejected": -147.19290161132812,
|
| 1009 |
+
"loss": 0.6661,
|
| 1010 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1011 |
+
"rewards/chosen": -0.39554575085639954,
|
| 1012 |
+
"rewards/margins": 0.18837305903434753,
|
| 1013 |
+
"rewards/rejected": -0.5839187502861023,
|
| 1014 |
+
"step": 335
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.3715295305401312,
|
| 1018 |
+
"grad_norm": 47.88369369506836,
|
| 1019 |
+
"learning_rate": 5e-06,
|
| 1020 |
+
"logits/chosen": -0.9874971508979797,
|
| 1021 |
+
"logits/rejected": -0.9519085884094238,
|
| 1022 |
+
"logps/chosen": -128.935791015625,
|
| 1023 |
+
"logps/rejected": -149.05130004882812,
|
| 1024 |
+
"loss": 0.5798,
|
| 1025 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1026 |
+
"rewards/chosen": -0.3125215470790863,
|
| 1027 |
+
"rewards/margins": 0.3500240445137024,
|
| 1028 |
+
"rewards/rejected": -0.6625455617904663,
|
| 1029 |
+
"step": 340
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.3917213528520949,
|
| 1033 |
+
"grad_norm": 51.914127349853516,
|
| 1034 |
+
"learning_rate": 5e-06,
|
| 1035 |
+
"logits/chosen": -1.0254395008087158,
|
| 1036 |
+
"logits/rejected": -0.9828994870185852,
|
| 1037 |
+
"logps/chosen": -131.8434295654297,
|
| 1038 |
+
"logps/rejected": -147.67422485351562,
|
| 1039 |
+
"loss": 0.5951,
|
| 1040 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1041 |
+
"rewards/chosen": -0.3478095829486847,
|
| 1042 |
+
"rewards/margins": 0.28336477279663086,
|
| 1043 |
+
"rewards/rejected": -0.6311743855476379,
|
| 1044 |
+
"step": 345
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.4119131751640586,
|
| 1048 |
+
"grad_norm": 202.00729370117188,
|
| 1049 |
+
"learning_rate": 5e-06,
|
| 1050 |
+
"logits/chosen": -1.0301780700683594,
|
| 1051 |
+
"logits/rejected": -0.9973276257514954,
|
| 1052 |
+
"logps/chosen": -134.11044311523438,
|
| 1053 |
+
"logps/rejected": -146.69715881347656,
|
| 1054 |
+
"loss": 0.5718,
|
| 1055 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1056 |
+
"rewards/chosen": -0.36468935012817383,
|
| 1057 |
+
"rewards/margins": 0.3158627152442932,
|
| 1058 |
+
"rewards/rejected": -0.680552065372467,
|
| 1059 |
+
"step": 350
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.4321049974760223,
|
| 1063 |
+
"grad_norm": 59.660091400146484,
|
| 1064 |
+
"learning_rate": 5e-06,
|
| 1065 |
+
"logits/chosen": -0.9530469179153442,
|
| 1066 |
+
"logits/rejected": -0.9394823312759399,
|
| 1067 |
+
"logps/chosen": -129.4771270751953,
|
| 1068 |
+
"logps/rejected": -139.2011260986328,
|
| 1069 |
+
"loss": 0.6579,
|
| 1070 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1071 |
+
"rewards/chosen": -0.3907158672809601,
|
| 1072 |
+
"rewards/margins": 0.22507838904857635,
|
| 1073 |
+
"rewards/rejected": -0.61579430103302,
|
| 1074 |
+
"step": 355
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 1.452296819787986,
|
| 1078 |
+
"grad_norm": 37.25257110595703,
|
| 1079 |
+
"learning_rate": 5e-06,
|
| 1080 |
+
"logits/chosen": -0.9922634363174438,
|
| 1081 |
+
"logits/rejected": -0.960089385509491,
|
| 1082 |
+
"logps/chosen": -132.1666717529297,
|
| 1083 |
+
"logps/rejected": -146.88284301757812,
|
| 1084 |
+
"loss": 0.5824,
|
| 1085 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1086 |
+
"rewards/chosen": -0.30794504284858704,
|
| 1087 |
+
"rewards/margins": 0.31921353936195374,
|
| 1088 |
+
"rewards/rejected": -0.627158522605896,
|
| 1089 |
+
"step": 360
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 1.4724886420999495,
|
| 1093 |
+
"grad_norm": 32.722537994384766,
|
| 1094 |
+
"learning_rate": 5e-06,
|
| 1095 |
+
"logits/chosen": -0.9966095685958862,
|
| 1096 |
+
"logits/rejected": -0.990027904510498,
|
| 1097 |
+
"logps/chosen": -130.92294311523438,
|
| 1098 |
+
"logps/rejected": -150.19712829589844,
|
| 1099 |
+
"loss": 0.6163,
|
| 1100 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1101 |
+
"rewards/chosen": -0.5703646540641785,
|
| 1102 |
+
"rewards/margins": 0.3038683533668518,
|
| 1103 |
+
"rewards/rejected": -0.874233067035675,
|
| 1104 |
+
"step": 365
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 1.4926804644119132,
|
| 1108 |
+
"grad_norm": 25.522695541381836,
|
| 1109 |
+
"learning_rate": 5e-06,
|
| 1110 |
+
"logits/chosen": -0.9670052528381348,
|
| 1111 |
+
"logits/rejected": -0.9508792757987976,
|
| 1112 |
+
"logps/chosen": -126.89973449707031,
|
| 1113 |
+
"logps/rejected": -139.86825561523438,
|
| 1114 |
+
"loss": 0.6495,
|
| 1115 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1116 |
+
"rewards/chosen": -0.5382200479507446,
|
| 1117 |
+
"rewards/margins": 0.1835038661956787,
|
| 1118 |
+
"rewards/rejected": -0.7217239141464233,
|
| 1119 |
+
"step": 370
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 1.5128722867238769,
|
| 1123 |
+
"grad_norm": 90.71711730957031,
|
| 1124 |
+
"learning_rate": 5e-06,
|
| 1125 |
+
"logits/chosen": -0.9002586603164673,
|
| 1126 |
+
"logits/rejected": -0.8389655947685242,
|
| 1127 |
+
"logps/chosen": -139.4822235107422,
|
| 1128 |
+
"logps/rejected": -149.20494079589844,
|
| 1129 |
+
"loss": 0.6875,
|
| 1130 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1131 |
+
"rewards/chosen": -0.6395633816719055,
|
| 1132 |
+
"rewards/margins": 0.2545303702354431,
|
| 1133 |
+
"rewards/rejected": -0.8940938115119934,
|
| 1134 |
+
"step": 375
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.5330641090358403,
|
| 1138 |
+
"grad_norm": 63.06035232543945,
|
| 1139 |
+
"learning_rate": 5e-06,
|
| 1140 |
+
"logits/chosen": -0.969630241394043,
|
| 1141 |
+
"logits/rejected": -0.9403184056282043,
|
| 1142 |
+
"logps/chosen": -128.59902954101562,
|
| 1143 |
+
"logps/rejected": -152.47572326660156,
|
| 1144 |
+
"loss": 0.568,
|
| 1145 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1146 |
+
"rewards/chosen": -0.4838544726371765,
|
| 1147 |
+
"rewards/margins": 0.41895875334739685,
|
| 1148 |
+
"rewards/rejected": -0.9028133153915405,
|
| 1149 |
+
"step": 380
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.5532559313478043,
|
| 1153 |
+
"grad_norm": 31.95730972290039,
|
| 1154 |
+
"learning_rate": 5e-06,
|
| 1155 |
+
"logits/chosen": -0.9574828147888184,
|
| 1156 |
+
"logits/rejected": -0.9362820386886597,
|
| 1157 |
+
"logps/chosen": -133.1338348388672,
|
| 1158 |
+
"logps/rejected": -137.119873046875,
|
| 1159 |
+
"loss": 0.6919,
|
| 1160 |
+
"rewards/accuracies": 0.625,
|
| 1161 |
+
"rewards/chosen": -0.6652856469154358,
|
| 1162 |
+
"rewards/margins": 0.1158902496099472,
|
| 1163 |
+
"rewards/rejected": -0.7811757922172546,
|
| 1164 |
+
"step": 385
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.5734477536597677,
|
| 1168 |
+
"grad_norm": 82.7573471069336,
|
| 1169 |
+
"learning_rate": 5e-06,
|
| 1170 |
+
"logits/chosen": -0.9827107191085815,
|
| 1171 |
+
"logits/rejected": -0.9468755722045898,
|
| 1172 |
+
"logps/chosen": -133.9182891845703,
|
| 1173 |
+
"logps/rejected": -147.95758056640625,
|
| 1174 |
+
"loss": 0.6258,
|
| 1175 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1176 |
+
"rewards/chosen": -0.508975088596344,
|
| 1177 |
+
"rewards/margins": 0.27125072479248047,
|
| 1178 |
+
"rewards/rejected": -0.7802258133888245,
|
| 1179 |
+
"step": 390
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.5936395759717314,
|
| 1183 |
+
"grad_norm": 44.66698455810547,
|
| 1184 |
+
"learning_rate": 5e-06,
|
| 1185 |
+
"logits/chosen": -0.9529491662979126,
|
| 1186 |
+
"logits/rejected": -0.9102163314819336,
|
| 1187 |
+
"logps/chosen": -127.85603332519531,
|
| 1188 |
+
"logps/rejected": -165.6941375732422,
|
| 1189 |
+
"loss": 0.5216,
|
| 1190 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1191 |
+
"rewards/chosen": -0.4891042709350586,
|
| 1192 |
+
"rewards/margins": 0.5509933233261108,
|
| 1193 |
+
"rewards/rejected": -1.0400975942611694,
|
| 1194 |
+
"step": 395
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.6138313982836952,
|
| 1198 |
+
"grad_norm": 28.743886947631836,
|
| 1199 |
+
"learning_rate": 5e-06,
|
| 1200 |
+
"logits/chosen": -0.9382497072219849,
|
| 1201 |
+
"logits/rejected": -0.9725173115730286,
|
| 1202 |
+
"logps/chosen": -125.57157135009766,
|
| 1203 |
+
"logps/rejected": -139.066650390625,
|
| 1204 |
+
"loss": 0.6589,
|
| 1205 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1206 |
+
"rewards/chosen": -0.4889567792415619,
|
| 1207 |
+
"rewards/margins": 0.13421431183815002,
|
| 1208 |
+
"rewards/rejected": -0.6231711506843567,
|
| 1209 |
+
"step": 400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 1.6340232205956586,
|
| 1213 |
+
"grad_norm": 26.614904403686523,
|
| 1214 |
+
"learning_rate": 5e-06,
|
| 1215 |
+
"logits/chosen": -0.9904226064682007,
|
| 1216 |
+
"logits/rejected": -0.9899235963821411,
|
| 1217 |
+
"logps/chosen": -130.1407470703125,
|
| 1218 |
+
"logps/rejected": -148.65634155273438,
|
| 1219 |
+
"loss": 0.568,
|
| 1220 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1221 |
+
"rewards/chosen": -0.3976411819458008,
|
| 1222 |
+
"rewards/margins": 0.3590780794620514,
|
| 1223 |
+
"rewards/rejected": -0.7567192912101746,
|
| 1224 |
+
"step": 405
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 1.6542150429076226,
|
| 1228 |
+
"grad_norm": 63.328731536865234,
|
| 1229 |
+
"learning_rate": 5e-06,
|
| 1230 |
+
"logits/chosen": -0.9512359499931335,
|
| 1231 |
+
"logits/rejected": -0.9244702458381653,
|
| 1232 |
+
"logps/chosen": -142.51605224609375,
|
| 1233 |
+
"logps/rejected": -146.30654907226562,
|
| 1234 |
+
"loss": 0.6478,
|
| 1235 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1236 |
+
"rewards/chosen": -0.611981213092804,
|
| 1237 |
+
"rewards/margins": 0.2888861298561096,
|
| 1238 |
+
"rewards/rejected": -0.9008673429489136,
|
| 1239 |
+
"step": 410
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 1.674406865219586,
|
| 1243 |
+
"grad_norm": 77.01255798339844,
|
| 1244 |
+
"learning_rate": 5e-06,
|
| 1245 |
+
"logits/chosen": -0.957229495048523,
|
| 1246 |
+
"logits/rejected": -0.9442984461784363,
|
| 1247 |
+
"logps/chosen": -129.83509826660156,
|
| 1248 |
+
"logps/rejected": -140.24246215820312,
|
| 1249 |
+
"loss": 0.6499,
|
| 1250 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1251 |
+
"rewards/chosen": -0.3882976770401001,
|
| 1252 |
+
"rewards/margins": 0.2001279890537262,
|
| 1253 |
+
"rewards/rejected": -0.5884256958961487,
|
| 1254 |
+
"step": 415
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.6945986875315497,
|
| 1258 |
+
"grad_norm": 64.73177337646484,
|
| 1259 |
+
"learning_rate": 5e-06,
|
| 1260 |
+
"logits/chosen": -0.9330608248710632,
|
| 1261 |
+
"logits/rejected": -0.9302207827568054,
|
| 1262 |
+
"logps/chosen": -136.36305236816406,
|
| 1263 |
+
"logps/rejected": -156.6847381591797,
|
| 1264 |
+
"loss": 0.5401,
|
| 1265 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1266 |
+
"rewards/chosen": -0.5075548887252808,
|
| 1267 |
+
"rewards/margins": 0.4187861382961273,
|
| 1268 |
+
"rewards/rejected": -0.9263409376144409,
|
| 1269 |
+
"step": 420
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.7147905098435134,
|
| 1273 |
+
"grad_norm": 33.53404235839844,
|
| 1274 |
+
"learning_rate": 5e-06,
|
| 1275 |
+
"logits/chosen": -0.9966014623641968,
|
| 1276 |
+
"logits/rejected": -0.9909483790397644,
|
| 1277 |
+
"logps/chosen": -131.6799774169922,
|
| 1278 |
+
"logps/rejected": -132.3755340576172,
|
| 1279 |
+
"loss": 0.613,
|
| 1280 |
+
"rewards/accuracies": 0.75,
|
| 1281 |
+
"rewards/chosen": -0.41457399725914,
|
| 1282 |
+
"rewards/margins": 0.2104615867137909,
|
| 1283 |
+
"rewards/rejected": -0.6250355839729309,
|
| 1284 |
+
"step": 425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.734982332155477,
|
| 1288 |
+
"grad_norm": 22.384197235107422,
|
| 1289 |
+
"learning_rate": 5e-06,
|
| 1290 |
+
"logits/chosen": -0.9789875149726868,
|
| 1291 |
+
"logits/rejected": -0.9263578653335571,
|
| 1292 |
+
"logps/chosen": -135.25543212890625,
|
| 1293 |
+
"logps/rejected": -143.0823211669922,
|
| 1294 |
+
"loss": 0.6018,
|
| 1295 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1296 |
+
"rewards/chosen": -0.4446907639503479,
|
| 1297 |
+
"rewards/margins": 0.33062392473220825,
|
| 1298 |
+
"rewards/rejected": -0.7753147482872009,
|
| 1299 |
+
"step": 430
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.7551741544674409,
|
| 1303 |
+
"grad_norm": 37.48057174682617,
|
| 1304 |
+
"learning_rate": 5e-06,
|
| 1305 |
+
"logits/chosen": -1.0088233947753906,
|
| 1306 |
+
"logits/rejected": -1.007187843322754,
|
| 1307 |
+
"logps/chosen": -129.21136474609375,
|
| 1308 |
+
"logps/rejected": -147.24827575683594,
|
| 1309 |
+
"loss": 0.6201,
|
| 1310 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1311 |
+
"rewards/chosen": -0.4158124029636383,
|
| 1312 |
+
"rewards/margins": 0.2677377462387085,
|
| 1313 |
+
"rewards/rejected": -0.683550238609314,
|
| 1314 |
+
"step": 435
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 1.7753659767794043,
|
| 1318 |
+
"grad_norm": 82.85381317138672,
|
| 1319 |
+
"learning_rate": 5e-06,
|
| 1320 |
+
"logits/chosen": -0.9587488174438477,
|
| 1321 |
+
"logits/rejected": -0.9475493431091309,
|
| 1322 |
+
"logps/chosen": -129.45645141601562,
|
| 1323 |
+
"logps/rejected": -136.0018310546875,
|
| 1324 |
+
"loss": 0.6111,
|
| 1325 |
+
"rewards/accuracies": 0.625,
|
| 1326 |
+
"rewards/chosen": -0.42559027671813965,
|
| 1327 |
+
"rewards/margins": 0.2766748368740082,
|
| 1328 |
+
"rewards/rejected": -0.7022650837898254,
|
| 1329 |
+
"step": 440
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.795557799091368,
|
| 1333 |
+
"grad_norm": 386.5701904296875,
|
| 1334 |
+
"learning_rate": 5e-06,
|
| 1335 |
+
"logits/chosen": -0.9646188616752625,
|
| 1336 |
+
"logits/rejected": -0.922524631023407,
|
| 1337 |
+
"logps/chosen": -126.2032699584961,
|
| 1338 |
+
"logps/rejected": -161.66268920898438,
|
| 1339 |
+
"loss": 0.5059,
|
| 1340 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1341 |
+
"rewards/chosen": -0.5310975313186646,
|
| 1342 |
+
"rewards/margins": 0.5927624702453613,
|
| 1343 |
+
"rewards/rejected": -1.1238601207733154,
|
| 1344 |
+
"step": 445
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 1.8157496214033317,
|
| 1348 |
+
"grad_norm": 33.02116012573242,
|
| 1349 |
+
"learning_rate": 5e-06,
|
| 1350 |
+
"logits/chosen": -0.9980605840682983,
|
| 1351 |
+
"logits/rejected": -0.9731265902519226,
|
| 1352 |
+
"logps/chosen": -124.5853500366211,
|
| 1353 |
+
"logps/rejected": -143.31353759765625,
|
| 1354 |
+
"loss": 0.5554,
|
| 1355 |
+
"rewards/accuracies": 0.75,
|
| 1356 |
+
"rewards/chosen": -0.6029568314552307,
|
| 1357 |
+
"rewards/margins": 0.38439229130744934,
|
| 1358 |
+
"rewards/rejected": -0.9873491525650024,
|
| 1359 |
+
"step": 450
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.8359414437152952,
|
| 1363 |
+
"grad_norm": 72.07835388183594,
|
| 1364 |
+
"learning_rate": 5e-06,
|
| 1365 |
+
"logits/chosen": -0.9654960632324219,
|
| 1366 |
+
"logits/rejected": -0.9319403767585754,
|
| 1367 |
+
"logps/chosen": -139.59352111816406,
|
| 1368 |
+
"logps/rejected": -148.42514038085938,
|
| 1369 |
+
"loss": 0.598,
|
| 1370 |
+
"rewards/accuracies": 0.75,
|
| 1371 |
+
"rewards/chosen": -0.7650180459022522,
|
| 1372 |
+
"rewards/margins": 0.3671188950538635,
|
| 1373 |
+
"rewards/rejected": -1.1321367025375366,
|
| 1374 |
+
"step": 455
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.856133266027259,
|
| 1378 |
+
"grad_norm": 37.11824417114258,
|
| 1379 |
+
"learning_rate": 5e-06,
|
| 1380 |
+
"logits/chosen": -0.9613730311393738,
|
| 1381 |
+
"logits/rejected": -0.9452563524246216,
|
| 1382 |
+
"logps/chosen": -131.92416381835938,
|
| 1383 |
+
"logps/rejected": -146.32144165039062,
|
| 1384 |
+
"loss": 0.684,
|
| 1385 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1386 |
+
"rewards/chosen": -0.8234866857528687,
|
| 1387 |
+
"rewards/margins": 0.22403354942798615,
|
| 1388 |
+
"rewards/rejected": -1.0475202798843384,
|
| 1389 |
+
"step": 460
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.8763250883392226,
|
| 1393 |
+
"grad_norm": 102.86495208740234,
|
| 1394 |
+
"learning_rate": 5e-06,
|
| 1395 |
+
"logits/chosen": -0.9701520800590515,
|
| 1396 |
+
"logits/rejected": -0.9547005891799927,
|
| 1397 |
+
"logps/chosen": -140.27639770507812,
|
| 1398 |
+
"logps/rejected": -140.81581115722656,
|
| 1399 |
+
"loss": 0.6644,
|
| 1400 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1401 |
+
"rewards/chosen": -0.8090511560440063,
|
| 1402 |
+
"rewards/margins": 0.21369917690753937,
|
| 1403 |
+
"rewards/rejected": -1.0227503776550293,
|
| 1404 |
+
"step": 465
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 1.896516910651186,
|
| 1408 |
+
"grad_norm": 38.73020553588867,
|
| 1409 |
+
"learning_rate": 5e-06,
|
| 1410 |
+
"logits/chosen": -0.9636425971984863,
|
| 1411 |
+
"logits/rejected": -0.8940737843513489,
|
| 1412 |
+
"logps/chosen": -128.97019958496094,
|
| 1413 |
+
"logps/rejected": -144.57125854492188,
|
| 1414 |
+
"loss": 0.5636,
|
| 1415 |
+
"rewards/accuracies": 0.75,
|
| 1416 |
+
"rewards/chosen": -0.7400966882705688,
|
| 1417 |
+
"rewards/margins": 0.3483065962791443,
|
| 1418 |
+
"rewards/rejected": -1.0884032249450684,
|
| 1419 |
+
"step": 470
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 1.91670873296315,
|
| 1423 |
+
"grad_norm": 50.96113204956055,
|
| 1424 |
+
"learning_rate": 5e-06,
|
| 1425 |
+
"logits/chosen": -1.0122615098953247,
|
| 1426 |
+
"logits/rejected": -0.9771004915237427,
|
| 1427 |
+
"logps/chosen": -130.7502899169922,
|
| 1428 |
+
"logps/rejected": -147.36337280273438,
|
| 1429 |
+
"loss": 0.55,
|
| 1430 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1431 |
+
"rewards/chosen": -0.5859622359275818,
|
| 1432 |
+
"rewards/margins": 0.4816177785396576,
|
| 1433 |
+
"rewards/rejected": -1.0675798654556274,
|
| 1434 |
+
"step": 475
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 1.9369005552751135,
|
| 1438 |
+
"grad_norm": 50.336578369140625,
|
| 1439 |
+
"learning_rate": 5e-06,
|
| 1440 |
+
"logits/chosen": -1.009649395942688,
|
| 1441 |
+
"logits/rejected": -0.9977648854255676,
|
| 1442 |
+
"logps/chosen": -138.2186279296875,
|
| 1443 |
+
"logps/rejected": -146.3845977783203,
|
| 1444 |
+
"loss": 0.6772,
|
| 1445 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1446 |
+
"rewards/chosen": -0.772314190864563,
|
| 1447 |
+
"rewards/margins": 0.2373802661895752,
|
| 1448 |
+
"rewards/rejected": -1.0096943378448486,
|
| 1449 |
+
"step": 480
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.9570923775870772,
|
| 1453 |
+
"grad_norm": 68.63349151611328,
|
| 1454 |
+
"learning_rate": 5e-06,
|
| 1455 |
+
"logits/chosen": -0.9773464202880859,
|
| 1456 |
+
"logits/rejected": -0.9308035969734192,
|
| 1457 |
+
"logps/chosen": -139.3035888671875,
|
| 1458 |
+
"logps/rejected": -161.8266143798828,
|
| 1459 |
+
"loss": 0.5776,
|
| 1460 |
+
"rewards/accuracies": 0.75,
|
| 1461 |
+
"rewards/chosen": -0.8935576677322388,
|
| 1462 |
+
"rewards/margins": 0.4999828338623047,
|
| 1463 |
+
"rewards/rejected": -1.393540382385254,
|
| 1464 |
+
"step": 485
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.977284199899041,
|
| 1468 |
+
"grad_norm": 88.56826782226562,
|
| 1469 |
+
"learning_rate": 5e-06,
|
| 1470 |
+
"logits/chosen": -1.0616611242294312,
|
| 1471 |
+
"logits/rejected": -1.0203070640563965,
|
| 1472 |
+
"logps/chosen": -130.81297302246094,
|
| 1473 |
+
"logps/rejected": -141.33740234375,
|
| 1474 |
+
"loss": 0.6037,
|
| 1475 |
+
"rewards/accuracies": 0.75,
|
| 1476 |
+
"rewards/chosen": -0.8136008977890015,
|
| 1477 |
+
"rewards/margins": 0.2671222984790802,
|
| 1478 |
+
"rewards/rejected": -1.0807230472564697,
|
| 1479 |
+
"step": 490
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 1.9974760222110044,
|
| 1483 |
+
"grad_norm": 126.10334777832031,
|
| 1484 |
+
"learning_rate": 5e-06,
|
| 1485 |
+
"logits/chosen": -1.0034191608428955,
|
| 1486 |
+
"logits/rejected": -1.010565161705017,
|
| 1487 |
+
"logps/chosen": -135.72381591796875,
|
| 1488 |
+
"logps/rejected": -133.8328399658203,
|
| 1489 |
+
"loss": 0.6832,
|
| 1490 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1491 |
+
"rewards/chosen": -0.7619636654853821,
|
| 1492 |
+
"rewards/margins": 0.17222894728183746,
|
| 1493 |
+
"rewards/rejected": -0.9341924786567688,
|
| 1494 |
+
"step": 495
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 2.016153457849571,
|
| 1498 |
+
"grad_norm": 48.9210205078125,
|
| 1499 |
+
"learning_rate": 5e-06,
|
| 1500 |
+
"logits/chosen": -1.023393988609314,
|
| 1501 |
+
"logits/rejected": -0.9911313652992249,
|
| 1502 |
+
"logps/chosen": -122.88961029052734,
|
| 1503 |
+
"logps/rejected": -149.01161193847656,
|
| 1504 |
+
"loss": 0.3935,
|
| 1505 |
+
"rewards/accuracies": 0.8648648858070374,
|
| 1506 |
+
"rewards/chosen": -0.35930752754211426,
|
| 1507 |
+
"rewards/margins": 0.8153496980667114,
|
| 1508 |
+
"rewards/rejected": -1.1746571063995361,
|
| 1509 |
+
"step": 500
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 2.0363452801615347,
|
| 1513 |
+
"grad_norm": 39.74098587036133,
|
| 1514 |
+
"learning_rate": 5e-06,
|
| 1515 |
+
"logits/chosen": -1.0313756465911865,
|
| 1516 |
+
"logits/rejected": -0.9942135810852051,
|
| 1517 |
+
"logps/chosen": -135.91714477539062,
|
| 1518 |
+
"logps/rejected": -147.81710815429688,
|
| 1519 |
+
"loss": 0.4991,
|
| 1520 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1521 |
+
"rewards/chosen": -0.5435918569564819,
|
| 1522 |
+
"rewards/margins": 0.6506127715110779,
|
| 1523 |
+
"rewards/rejected": -1.194204568862915,
|
| 1524 |
+
"step": 505
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 2.056537102473498,
|
| 1528 |
+
"grad_norm": 49.587120056152344,
|
| 1529 |
+
"learning_rate": 5e-06,
|
| 1530 |
+
"logits/chosen": -1.0160324573516846,
|
| 1531 |
+
"logits/rejected": -1.0125465393066406,
|
| 1532 |
+
"logps/chosen": -131.9548797607422,
|
| 1533 |
+
"logps/rejected": -152.05862426757812,
|
| 1534 |
+
"loss": 0.5047,
|
| 1535 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1536 |
+
"rewards/chosen": -0.6925584673881531,
|
| 1537 |
+
"rewards/margins": 0.5985544919967651,
|
| 1538 |
+
"rewards/rejected": -1.2911128997802734,
|
| 1539 |
+
"step": 510
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 2.0767289247854617,
|
| 1543 |
+
"grad_norm": 17.234088897705078,
|
| 1544 |
+
"learning_rate": 5e-06,
|
| 1545 |
+
"logits/chosen": -0.9918926358222961,
|
| 1546 |
+
"logits/rejected": -0.9893625378608704,
|
| 1547 |
+
"logps/chosen": -132.64453125,
|
| 1548 |
+
"logps/rejected": -158.79873657226562,
|
| 1549 |
+
"loss": 0.42,
|
| 1550 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1551 |
+
"rewards/chosen": -0.5537530779838562,
|
| 1552 |
+
"rewards/margins": 0.938607394695282,
|
| 1553 |
+
"rewards/rejected": -1.4923605918884277,
|
| 1554 |
+
"step": 515
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 2.0969207470974256,
|
| 1558 |
+
"grad_norm": 162.84471130371094,
|
| 1559 |
+
"learning_rate": 5e-06,
|
| 1560 |
+
"logits/chosen": -1.008296012878418,
|
| 1561 |
+
"logits/rejected": -1.0098919868469238,
|
| 1562 |
+
"logps/chosen": -139.8673095703125,
|
| 1563 |
+
"logps/rejected": -160.15660095214844,
|
| 1564 |
+
"loss": 0.4181,
|
| 1565 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1566 |
+
"rewards/chosen": -0.6410013437271118,
|
| 1567 |
+
"rewards/margins": 0.9079731106758118,
|
| 1568 |
+
"rewards/rejected": -1.5489743947982788,
|
| 1569 |
+
"step": 520
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 2.117112569409389,
|
| 1573 |
+
"grad_norm": 29.040531158447266,
|
| 1574 |
+
"learning_rate": 5e-06,
|
| 1575 |
+
"logits/chosen": -1.0176864862442017,
|
| 1576 |
+
"logits/rejected": -1.007263422012329,
|
| 1577 |
+
"logps/chosen": -131.76019287109375,
|
| 1578 |
+
"logps/rejected": -149.74111938476562,
|
| 1579 |
+
"loss": 0.4928,
|
| 1580 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1581 |
+
"rewards/chosen": -0.6737687587738037,
|
| 1582 |
+
"rewards/margins": 0.6350966691970825,
|
| 1583 |
+
"rewards/rejected": -1.3088653087615967,
|
| 1584 |
+
"step": 525
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 2.137304391721353,
|
| 1588 |
+
"grad_norm": 97.38697052001953,
|
| 1589 |
+
"learning_rate": 5e-06,
|
| 1590 |
+
"logits/chosen": -1.0380053520202637,
|
| 1591 |
+
"logits/rejected": -0.9718503952026367,
|
| 1592 |
+
"logps/chosen": -132.33323669433594,
|
| 1593 |
+
"logps/rejected": -152.97665405273438,
|
| 1594 |
+
"loss": 0.4573,
|
| 1595 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1596 |
+
"rewards/chosen": -0.6338863372802734,
|
| 1597 |
+
"rewards/margins": 0.8563046455383301,
|
| 1598 |
+
"rewards/rejected": -1.4901909828186035,
|
| 1599 |
+
"step": 530
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 2.1574962140333165,
|
| 1603 |
+
"grad_norm": 77.62568664550781,
|
| 1604 |
+
"learning_rate": 5e-06,
|
| 1605 |
+
"logits/chosen": -1.0812318325042725,
|
| 1606 |
+
"logits/rejected": -1.0583606958389282,
|
| 1607 |
+
"logps/chosen": -135.34068298339844,
|
| 1608 |
+
"logps/rejected": -151.8748016357422,
|
| 1609 |
+
"loss": 0.4322,
|
| 1610 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1611 |
+
"rewards/chosen": -0.6975581049919128,
|
| 1612 |
+
"rewards/margins": 0.8094134330749512,
|
| 1613 |
+
"rewards/rejected": -1.5069715976715088,
|
| 1614 |
+
"step": 535
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 2.17768803634528,
|
| 1618 |
+
"grad_norm": 67.6436767578125,
|
| 1619 |
+
"learning_rate": 5e-06,
|
| 1620 |
+
"logits/chosen": -1.0246284008026123,
|
| 1621 |
+
"logits/rejected": -1.014534831047058,
|
| 1622 |
+
"logps/chosen": -137.48171997070312,
|
| 1623 |
+
"logps/rejected": -149.2544708251953,
|
| 1624 |
+
"loss": 0.4063,
|
| 1625 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1626 |
+
"rewards/chosen": -0.7033039331436157,
|
| 1627 |
+
"rewards/margins": 0.945264458656311,
|
| 1628 |
+
"rewards/rejected": -1.6485683917999268,
|
| 1629 |
+
"step": 540
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 2.197879858657244,
|
| 1633 |
+
"grad_norm": 63.45933532714844,
|
| 1634 |
+
"learning_rate": 5e-06,
|
| 1635 |
+
"logits/chosen": -1.0668962001800537,
|
| 1636 |
+
"logits/rejected": -1.0191361904144287,
|
| 1637 |
+
"logps/chosen": -136.38905334472656,
|
| 1638 |
+
"logps/rejected": -156.759521484375,
|
| 1639 |
+
"loss": 0.4042,
|
| 1640 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1641 |
+
"rewards/chosen": -0.7140289545059204,
|
| 1642 |
+
"rewards/margins": 0.8608022928237915,
|
| 1643 |
+
"rewards/rejected": -1.574831247329712,
|
| 1644 |
+
"step": 545
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 2.2180716809692074,
|
| 1648 |
+
"grad_norm": 24.154354095458984,
|
| 1649 |
+
"learning_rate": 5e-06,
|
| 1650 |
+
"logits/chosen": -1.034576654434204,
|
| 1651 |
+
"logits/rejected": -1.0220065116882324,
|
| 1652 |
+
"logps/chosen": -126.9083251953125,
|
| 1653 |
+
"logps/rejected": -147.03082275390625,
|
| 1654 |
+
"loss": 0.4812,
|
| 1655 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1656 |
+
"rewards/chosen": -0.6011126041412354,
|
| 1657 |
+
"rewards/margins": 0.7494941353797913,
|
| 1658 |
+
"rewards/rejected": -1.3506066799163818,
|
| 1659 |
+
"step": 550
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 2.2382635032811713,
|
| 1663 |
+
"grad_norm": 51.32468032836914,
|
| 1664 |
+
"learning_rate": 5e-06,
|
| 1665 |
+
"logits/chosen": -1.0417560338974,
|
| 1666 |
+
"logits/rejected": -0.995162844657898,
|
| 1667 |
+
"logps/chosen": -134.18875122070312,
|
| 1668 |
+
"logps/rejected": -160.58987426757812,
|
| 1669 |
+
"loss": 0.4263,
|
| 1670 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1671 |
+
"rewards/chosen": -0.7680782079696655,
|
| 1672 |
+
"rewards/margins": 0.9993975758552551,
|
| 1673 |
+
"rewards/rejected": -1.7674758434295654,
|
| 1674 |
+
"step": 555
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 2.2584553255931348,
|
| 1678 |
+
"grad_norm": 31.85598373413086,
|
| 1679 |
+
"learning_rate": 5e-06,
|
| 1680 |
+
"logits/chosen": -1.091623306274414,
|
| 1681 |
+
"logits/rejected": -1.065324068069458,
|
| 1682 |
+
"logps/chosen": -147.77536010742188,
|
| 1683 |
+
"logps/rejected": -161.46456909179688,
|
| 1684 |
+
"loss": 0.4047,
|
| 1685 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1686 |
+
"rewards/chosen": -0.9492090940475464,
|
| 1687 |
+
"rewards/margins": 0.9160175323486328,
|
| 1688 |
+
"rewards/rejected": -1.8652265071868896,
|
| 1689 |
+
"step": 560
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 2.2786471479050983,
|
| 1693 |
+
"grad_norm": 40.57142639160156,
|
| 1694 |
+
"learning_rate": 5e-06,
|
| 1695 |
+
"logits/chosen": -1.0318306684494019,
|
| 1696 |
+
"logits/rejected": -1.0202347040176392,
|
| 1697 |
+
"logps/chosen": -133.6524658203125,
|
| 1698 |
+
"logps/rejected": -154.2473602294922,
|
| 1699 |
+
"loss": 0.4398,
|
| 1700 |
+
"rewards/accuracies": 0.875,
|
| 1701 |
+
"rewards/chosen": -0.774714469909668,
|
| 1702 |
+
"rewards/margins": 0.8600656390190125,
|
| 1703 |
+
"rewards/rejected": -1.6347801685333252,
|
| 1704 |
+
"step": 565
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 2.298838970217062,
|
| 1708 |
+
"grad_norm": 310.98028564453125,
|
| 1709 |
+
"learning_rate": 5e-06,
|
| 1710 |
+
"logits/chosen": -1.0679352283477783,
|
| 1711 |
+
"logits/rejected": -1.0488876104354858,
|
| 1712 |
+
"logps/chosen": -143.3402557373047,
|
| 1713 |
+
"logps/rejected": -157.1344757080078,
|
| 1714 |
+
"loss": 0.4709,
|
| 1715 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1716 |
+
"rewards/chosen": -0.8914315104484558,
|
| 1717 |
+
"rewards/margins": 0.8738172650337219,
|
| 1718 |
+
"rewards/rejected": -1.7652488946914673,
|
| 1719 |
+
"step": 570
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 2.3190307925290257,
|
| 1723 |
+
"grad_norm": 32.55216979980469,
|
| 1724 |
+
"learning_rate": 5e-06,
|
| 1725 |
+
"logits/chosen": -1.0195152759552002,
|
| 1726 |
+
"logits/rejected": -1.0584851503372192,
|
| 1727 |
+
"logps/chosen": -140.6759796142578,
|
| 1728 |
+
"logps/rejected": -161.9089813232422,
|
| 1729 |
+
"loss": 0.5658,
|
| 1730 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1731 |
+
"rewards/chosen": -1.0393126010894775,
|
| 1732 |
+
"rewards/margins": 0.8234487771987915,
|
| 1733 |
+
"rewards/rejected": -1.8627614974975586,
|
| 1734 |
+
"step": 575
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 2.3392226148409896,
|
| 1738 |
+
"grad_norm": 68.31248474121094,
|
| 1739 |
+
"learning_rate": 5e-06,
|
| 1740 |
+
"logits/chosen": -1.0342488288879395,
|
| 1741 |
+
"logits/rejected": -1.0230834484100342,
|
| 1742 |
+
"logps/chosen": -129.74427795410156,
|
| 1743 |
+
"logps/rejected": -151.37411499023438,
|
| 1744 |
+
"loss": 0.434,
|
| 1745 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1746 |
+
"rewards/chosen": -0.707876443862915,
|
| 1747 |
+
"rewards/margins": 1.0140485763549805,
|
| 1748 |
+
"rewards/rejected": -1.721925139427185,
|
| 1749 |
+
"step": 580
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 2.359414437152953,
|
| 1753 |
+
"grad_norm": 89.6629638671875,
|
| 1754 |
+
"learning_rate": 5e-06,
|
| 1755 |
+
"logits/chosen": -1.1206090450286865,
|
| 1756 |
+
"logits/rejected": -1.0528461933135986,
|
| 1757 |
+
"logps/chosen": -128.90150451660156,
|
| 1758 |
+
"logps/rejected": -145.35079956054688,
|
| 1759 |
+
"loss": 0.4469,
|
| 1760 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1761 |
+
"rewards/chosen": -0.7581853270530701,
|
| 1762 |
+
"rewards/margins": 0.7499567270278931,
|
| 1763 |
+
"rewards/rejected": -1.5081422328948975,
|
| 1764 |
+
"step": 585
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 2.3796062594649166,
|
| 1768 |
+
"grad_norm": 43.05556869506836,
|
| 1769 |
+
"learning_rate": 5e-06,
|
| 1770 |
+
"logits/chosen": -1.065771460533142,
|
| 1771 |
+
"logits/rejected": -1.0271015167236328,
|
| 1772 |
+
"logps/chosen": -130.0537872314453,
|
| 1773 |
+
"logps/rejected": -149.29409790039062,
|
| 1774 |
+
"loss": 0.4797,
|
| 1775 |
+
"rewards/accuracies": 0.875,
|
| 1776 |
+
"rewards/chosen": -0.8601150512695312,
|
| 1777 |
+
"rewards/margins": 0.7315414547920227,
|
| 1778 |
+
"rewards/rejected": -1.5916566848754883,
|
| 1779 |
+
"step": 590
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 2.3997980817768805,
|
| 1783 |
+
"grad_norm": 66.487060546875,
|
| 1784 |
+
"learning_rate": 5e-06,
|
| 1785 |
+
"logits/chosen": -1.1071510314941406,
|
| 1786 |
+
"logits/rejected": -1.0782781839370728,
|
| 1787 |
+
"logps/chosen": -138.64755249023438,
|
| 1788 |
+
"logps/rejected": -162.64129638671875,
|
| 1789 |
+
"loss": 0.4263,
|
| 1790 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1791 |
+
"rewards/chosen": -0.9284160733222961,
|
| 1792 |
+
"rewards/margins": 1.0049647092819214,
|
| 1793 |
+
"rewards/rejected": -1.9333808422088623,
|
| 1794 |
+
"step": 595
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 2.419989904088844,
|
| 1798 |
+
"grad_norm": 70.14442443847656,
|
| 1799 |
+
"learning_rate": 5e-06,
|
| 1800 |
+
"logits/chosen": -1.0711498260498047,
|
| 1801 |
+
"logits/rejected": -1.0283135175704956,
|
| 1802 |
+
"logps/chosen": -139.4008026123047,
|
| 1803 |
+
"logps/rejected": -173.4601287841797,
|
| 1804 |
+
"loss": 0.4815,
|
| 1805 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1806 |
+
"rewards/chosen": -1.1457339525222778,
|
| 1807 |
+
"rewards/margins": 0.9765189290046692,
|
| 1808 |
+
"rewards/rejected": -2.122252941131592,
|
| 1809 |
+
"step": 600
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 2.440181726400808,
|
| 1813 |
+
"grad_norm": 39.94354248046875,
|
| 1814 |
+
"learning_rate": 5e-06,
|
| 1815 |
+
"logits/chosen": -1.0897289514541626,
|
| 1816 |
+
"logits/rejected": -1.0542991161346436,
|
| 1817 |
+
"logps/chosen": -140.98178100585938,
|
| 1818 |
+
"logps/rejected": -154.21890258789062,
|
| 1819 |
+
"loss": 0.3809,
|
| 1820 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1821 |
+
"rewards/chosen": -0.9919537305831909,
|
| 1822 |
+
"rewards/margins": 1.0165085792541504,
|
| 1823 |
+
"rewards/rejected": -2.008462429046631,
|
| 1824 |
+
"step": 605
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 2.4603735487127714,
|
| 1828 |
+
"grad_norm": 45.401206970214844,
|
| 1829 |
+
"learning_rate": 5e-06,
|
| 1830 |
+
"logits/chosen": -1.088404893875122,
|
| 1831 |
+
"logits/rejected": -1.0834258794784546,
|
| 1832 |
+
"logps/chosen": -132.83590698242188,
|
| 1833 |
+
"logps/rejected": -148.06394958496094,
|
| 1834 |
+
"loss": 0.456,
|
| 1835 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1836 |
+
"rewards/chosen": -0.9679088592529297,
|
| 1837 |
+
"rewards/margins": 0.8238266110420227,
|
| 1838 |
+
"rewards/rejected": -1.7917354106903076,
|
| 1839 |
+
"step": 610
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 2.480565371024735,
|
| 1843 |
+
"grad_norm": 70.00696563720703,
|
| 1844 |
+
"learning_rate": 5e-06,
|
| 1845 |
+
"logits/chosen": -1.0881659984588623,
|
| 1846 |
+
"logits/rejected": -1.0491617918014526,
|
| 1847 |
+
"logps/chosen": -126.14036560058594,
|
| 1848 |
+
"logps/rejected": -139.42196655273438,
|
| 1849 |
+
"loss": 0.4363,
|
| 1850 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1851 |
+
"rewards/chosen": -0.7842862010002136,
|
| 1852 |
+
"rewards/margins": 0.8065754771232605,
|
| 1853 |
+
"rewards/rejected": -1.5908616781234741,
|
| 1854 |
+
"step": 615
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 2.5007571933366988,
|
| 1858 |
+
"grad_norm": 93.69404602050781,
|
| 1859 |
+
"learning_rate": 5e-06,
|
| 1860 |
+
"logits/chosen": -1.088850975036621,
|
| 1861 |
+
"logits/rejected": -1.061447262763977,
|
| 1862 |
+
"logps/chosen": -144.85092163085938,
|
| 1863 |
+
"logps/rejected": -187.40280151367188,
|
| 1864 |
+
"loss": 0.334,
|
| 1865 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1866 |
+
"rewards/chosen": -1.2291361093521118,
|
| 1867 |
+
"rewards/margins": 1.6782220602035522,
|
| 1868 |
+
"rewards/rejected": -2.907358169555664,
|
| 1869 |
+
"step": 620
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 2.5209490156486623,
|
| 1873 |
+
"grad_norm": 24.65225601196289,
|
| 1874 |
+
"learning_rate": 5e-06,
|
| 1875 |
+
"logits/chosen": -1.1317861080169678,
|
| 1876 |
+
"logits/rejected": -1.0619679689407349,
|
| 1877 |
+
"logps/chosen": -150.44094848632812,
|
| 1878 |
+
"logps/rejected": -178.79635620117188,
|
| 1879 |
+
"loss": 0.4855,
|
| 1880 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1881 |
+
"rewards/chosen": -1.2844429016113281,
|
| 1882 |
+
"rewards/margins": 1.1455484628677368,
|
| 1883 |
+
"rewards/rejected": -2.4299912452697754,
|
| 1884 |
+
"step": 625
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 2.5411408379606257,
|
| 1888 |
+
"grad_norm": 81.89047241210938,
|
| 1889 |
+
"learning_rate": 5e-06,
|
| 1890 |
+
"logits/chosen": -1.108019471168518,
|
| 1891 |
+
"logits/rejected": -1.1051324605941772,
|
| 1892 |
+
"logps/chosen": -144.0532989501953,
|
| 1893 |
+
"logps/rejected": -162.6993865966797,
|
| 1894 |
+
"loss": 0.4032,
|
| 1895 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1896 |
+
"rewards/chosen": -1.3340096473693848,
|
| 1897 |
+
"rewards/margins": 1.0980908870697021,
|
| 1898 |
+
"rewards/rejected": -2.432100772857666,
|
| 1899 |
+
"step": 630
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 2.5613326602725897,
|
| 1903 |
+
"grad_norm": 143.68875122070312,
|
| 1904 |
+
"learning_rate": 5e-06,
|
| 1905 |
+
"logits/chosen": -1.0815242528915405,
|
| 1906 |
+
"logits/rejected": -1.0449540615081787,
|
| 1907 |
+
"logps/chosen": -138.04147338867188,
|
| 1908 |
+
"logps/rejected": -156.7818603515625,
|
| 1909 |
+
"loss": 0.4695,
|
| 1910 |
+
"rewards/accuracies": 0.75,
|
| 1911 |
+
"rewards/chosen": -1.1757011413574219,
|
| 1912 |
+
"rewards/margins": 1.038955807685852,
|
| 1913 |
+
"rewards/rejected": -2.2146568298339844,
|
| 1914 |
+
"step": 635
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 2.581524482584553,
|
| 1918 |
+
"grad_norm": 118.63373565673828,
|
| 1919 |
+
"learning_rate": 5e-06,
|
| 1920 |
+
"logits/chosen": -1.1306525468826294,
|
| 1921 |
+
"logits/rejected": -1.1150437593460083,
|
| 1922 |
+
"logps/chosen": -152.9892578125,
|
| 1923 |
+
"logps/rejected": -168.2422332763672,
|
| 1924 |
+
"loss": 0.5398,
|
| 1925 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1926 |
+
"rewards/chosen": -1.6195358037948608,
|
| 1927 |
+
"rewards/margins": 1.2459442615509033,
|
| 1928 |
+
"rewards/rejected": -2.8654799461364746,
|
| 1929 |
+
"step": 640
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 2.601716304896517,
|
| 1933 |
+
"grad_norm": 85.33989715576172,
|
| 1934 |
+
"learning_rate": 5e-06,
|
| 1935 |
+
"logits/chosen": -1.1470824480056763,
|
| 1936 |
+
"logits/rejected": -1.1052191257476807,
|
| 1937 |
+
"logps/chosen": -144.22109985351562,
|
| 1938 |
+
"logps/rejected": -171.24172973632812,
|
| 1939 |
+
"loss": 0.3907,
|
| 1940 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1941 |
+
"rewards/chosen": -1.3878965377807617,
|
| 1942 |
+
"rewards/margins": 1.330134630203247,
|
| 1943 |
+
"rewards/rejected": -2.718031167984009,
|
| 1944 |
+
"step": 645
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 2.6219081272084805,
|
| 1948 |
+
"grad_norm": 28.526046752929688,
|
| 1949 |
+
"learning_rate": 5e-06,
|
| 1950 |
+
"logits/chosen": -1.1499074697494507,
|
| 1951 |
+
"logits/rejected": -1.1181514263153076,
|
| 1952 |
+
"logps/chosen": -141.45933532714844,
|
| 1953 |
+
"logps/rejected": -165.86814880371094,
|
| 1954 |
+
"loss": 0.3265,
|
| 1955 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1956 |
+
"rewards/chosen": -1.2811788320541382,
|
| 1957 |
+
"rewards/margins": 1.2787883281707764,
|
| 1958 |
+
"rewards/rejected": -2.559967279434204,
|
| 1959 |
+
"step": 650
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 2.6420999495204445,
|
| 1963 |
+
"grad_norm": 50.67562484741211,
|
| 1964 |
+
"learning_rate": 5e-06,
|
| 1965 |
+
"logits/chosen": -1.0879682302474976,
|
| 1966 |
+
"logits/rejected": -1.0264991521835327,
|
| 1967 |
+
"logps/chosen": -140.26417541503906,
|
| 1968 |
+
"logps/rejected": -162.7800750732422,
|
| 1969 |
+
"loss": 0.3963,
|
| 1970 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1971 |
+
"rewards/chosen": -1.3028833866119385,
|
| 1972 |
+
"rewards/margins": 1.1044894456863403,
|
| 1973 |
+
"rewards/rejected": -2.4073727130889893,
|
| 1974 |
+
"step": 655
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 2.662291771832408,
|
| 1978 |
+
"grad_norm": 71.36746215820312,
|
| 1979 |
+
"learning_rate": 5e-06,
|
| 1980 |
+
"logits/chosen": -1.12251877784729,
|
| 1981 |
+
"logits/rejected": -1.0848548412322998,
|
| 1982 |
+
"logps/chosen": -137.73040771484375,
|
| 1983 |
+
"logps/rejected": -154.519287109375,
|
| 1984 |
+
"loss": 0.45,
|
| 1985 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1986 |
+
"rewards/chosen": -1.2700433731079102,
|
| 1987 |
+
"rewards/margins": 0.8259071111679077,
|
| 1988 |
+
"rewards/rejected": -2.095950126647949,
|
| 1989 |
+
"step": 660
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 2.6824835941443714,
|
| 1993 |
+
"grad_norm": 145.13929748535156,
|
| 1994 |
+
"learning_rate": 5e-06,
|
| 1995 |
+
"logits/chosen": -1.1168526411056519,
|
| 1996 |
+
"logits/rejected": -1.0883421897888184,
|
| 1997 |
+
"logps/chosen": -138.43850708007812,
|
| 1998 |
+
"logps/rejected": -161.28207397460938,
|
| 1999 |
+
"loss": 0.4334,
|
| 2000 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 2001 |
+
"rewards/chosen": -1.4023481607437134,
|
| 2002 |
+
"rewards/margins": 1.0314326286315918,
|
| 2003 |
+
"rewards/rejected": -2.4337806701660156,
|
| 2004 |
+
"step": 665
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 2.7026754164563354,
|
| 2008 |
+
"grad_norm": 101.96822357177734,
|
| 2009 |
+
"learning_rate": 5e-06,
|
| 2010 |
+
"logits/chosen": -1.1315155029296875,
|
| 2011 |
+
"logits/rejected": -1.1191498041152954,
|
| 2012 |
+
"logps/chosen": -142.5680694580078,
|
| 2013 |
+
"logps/rejected": -147.30868530273438,
|
| 2014 |
+
"loss": 0.4604,
|
| 2015 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 2016 |
+
"rewards/chosen": -1.2466862201690674,
|
| 2017 |
+
"rewards/margins": 0.8339215517044067,
|
| 2018 |
+
"rewards/rejected": -2.0806076526641846,
|
| 2019 |
+
"step": 670
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 2.722867238768299,
|
| 2023 |
+
"grad_norm": 35.541744232177734,
|
| 2024 |
+
"learning_rate": 5e-06,
|
| 2025 |
+
"logits/chosen": -1.1304982900619507,
|
| 2026 |
+
"logits/rejected": -1.096704363822937,
|
| 2027 |
+
"logps/chosen": -140.98458862304688,
|
| 2028 |
+
"logps/rejected": -153.99526977539062,
|
| 2029 |
+
"loss": 0.3437,
|
| 2030 |
+
"rewards/accuracies": 0.875,
|
| 2031 |
+
"rewards/chosen": -1.1470431089401245,
|
| 2032 |
+
"rewards/margins": 1.193063735961914,
|
| 2033 |
+
"rewards/rejected": -2.340106964111328,
|
| 2034 |
+
"step": 675
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 2.7430590610802623,
|
| 2038 |
+
"grad_norm": 119.16114807128906,
|
| 2039 |
+
"learning_rate": 5e-06,
|
| 2040 |
+
"logits/chosen": -1.1810123920440674,
|
| 2041 |
+
"logits/rejected": -1.1393203735351562,
|
| 2042 |
+
"logps/chosen": -143.31198120117188,
|
| 2043 |
+
"logps/rejected": -155.2026824951172,
|
| 2044 |
+
"loss": 0.4346,
|
| 2045 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 2046 |
+
"rewards/chosen": -1.3686244487762451,
|
| 2047 |
+
"rewards/margins": 0.9831773638725281,
|
| 2048 |
+
"rewards/rejected": -2.351801633834839,
|
| 2049 |
+
"step": 680
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 2.7632508833922262,
|
| 2053 |
+
"grad_norm": 112.19788360595703,
|
| 2054 |
+
"learning_rate": 5e-06,
|
| 2055 |
+
"logits/chosen": -1.1026071310043335,
|
| 2056 |
+
"logits/rejected": -1.115681767463684,
|
| 2057 |
+
"logps/chosen": -145.30560302734375,
|
| 2058 |
+
"logps/rejected": -155.7900848388672,
|
| 2059 |
+
"loss": 0.5291,
|
| 2060 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 2061 |
+
"rewards/chosen": -1.4804359674453735,
|
| 2062 |
+
"rewards/margins": 0.9093244671821594,
|
| 2063 |
+
"rewards/rejected": -2.3897604942321777,
|
| 2064 |
+
"step": 685
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 2.7834427057041897,
|
| 2068 |
+
"grad_norm": 33.10359191894531,
|
| 2069 |
+
"learning_rate": 5e-06,
|
| 2070 |
+
"logits/chosen": -1.1413536071777344,
|
| 2071 |
+
"logits/rejected": -1.1235370635986328,
|
| 2072 |
+
"logps/chosen": -137.90969848632812,
|
| 2073 |
+
"logps/rejected": -170.1824188232422,
|
| 2074 |
+
"loss": 0.3372,
|
| 2075 |
+
"rewards/accuracies": 0.875,
|
| 2076 |
+
"rewards/chosen": -1.2158892154693604,
|
| 2077 |
+
"rewards/margins": 1.584923505783081,
|
| 2078 |
+
"rewards/rejected": -2.8008127212524414,
|
| 2079 |
+
"step": 690
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 2.803634528016153,
|
| 2083 |
+
"grad_norm": 150.4509735107422,
|
| 2084 |
+
"learning_rate": 5e-06,
|
| 2085 |
+
"logits/chosen": -1.1553657054901123,
|
| 2086 |
+
"logits/rejected": -1.1448795795440674,
|
| 2087 |
+
"logps/chosen": -137.4596405029297,
|
| 2088 |
+
"logps/rejected": -151.9920196533203,
|
| 2089 |
+
"loss": 0.4594,
|
| 2090 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2091 |
+
"rewards/chosen": -1.3479053974151611,
|
| 2092 |
+
"rewards/margins": 0.890475869178772,
|
| 2093 |
+
"rewards/rejected": -2.2383813858032227,
|
| 2094 |
+
"step": 695
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 2.823826350328117,
|
| 2098 |
+
"grad_norm": 287.857666015625,
|
| 2099 |
+
"learning_rate": 5e-06,
|
| 2100 |
+
"logits/chosen": -1.1810104846954346,
|
| 2101 |
+
"logits/rejected": -1.1226354837417603,
|
| 2102 |
+
"logps/chosen": -143.69639587402344,
|
| 2103 |
+
"logps/rejected": -163.6051483154297,
|
| 2104 |
+
"loss": 0.4457,
|
| 2105 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2106 |
+
"rewards/chosen": -1.513098955154419,
|
| 2107 |
+
"rewards/margins": 1.0407955646514893,
|
| 2108 |
+
"rewards/rejected": -2.5538947582244873,
|
| 2109 |
+
"step": 700
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 2.8440181726400806,
|
| 2113 |
+
"grad_norm": 50.71148681640625,
|
| 2114 |
+
"learning_rate": 5e-06,
|
| 2115 |
+
"logits/chosen": -1.1496613025665283,
|
| 2116 |
+
"logits/rejected": -1.12812340259552,
|
| 2117 |
+
"logps/chosen": -140.67123413085938,
|
| 2118 |
+
"logps/rejected": -169.8626708984375,
|
| 2119 |
+
"loss": 0.3212,
|
| 2120 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2121 |
+
"rewards/chosen": -1.3559614419937134,
|
| 2122 |
+
"rewards/margins": 1.4378584623336792,
|
| 2123 |
+
"rewards/rejected": -2.7938199043273926,
|
| 2124 |
+
"step": 705
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 2.8642099949520445,
|
| 2128 |
+
"grad_norm": 26.175968170166016,
|
| 2129 |
+
"learning_rate": 5e-06,
|
| 2130 |
+
"logits/chosen": -1.2208508253097534,
|
| 2131 |
+
"logits/rejected": -1.1588385105133057,
|
| 2132 |
+
"logps/chosen": -129.68093872070312,
|
| 2133 |
+
"logps/rejected": -161.8412322998047,
|
| 2134 |
+
"loss": 0.3767,
|
| 2135 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 2136 |
+
"rewards/chosen": -1.222773790359497,
|
| 2137 |
+
"rewards/margins": 1.361515998840332,
|
| 2138 |
+
"rewards/rejected": -2.58428955078125,
|
| 2139 |
+
"step": 710
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 2.884401817264008,
|
| 2143 |
+
"grad_norm": 51.06369400024414,
|
| 2144 |
+
"learning_rate": 5e-06,
|
| 2145 |
+
"logits/chosen": -1.1848516464233398,
|
| 2146 |
+
"logits/rejected": -1.1376698017120361,
|
| 2147 |
+
"logps/chosen": -131.61441040039062,
|
| 2148 |
+
"logps/rejected": -154.9497528076172,
|
| 2149 |
+
"loss": 0.3265,
|
| 2150 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2151 |
+
"rewards/chosen": -1.088381290435791,
|
| 2152 |
+
"rewards/margins": 1.4161834716796875,
|
| 2153 |
+
"rewards/rejected": -2.5045647621154785,
|
| 2154 |
+
"step": 715
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 2.904593639575972,
|
| 2158 |
+
"grad_norm": 35.54250717163086,
|
| 2159 |
+
"learning_rate": 5e-06,
|
| 2160 |
+
"logits/chosen": -1.1926292181015015,
|
| 2161 |
+
"logits/rejected": -1.157402753829956,
|
| 2162 |
+
"logps/chosen": -136.55596923828125,
|
| 2163 |
+
"logps/rejected": -163.8975372314453,
|
| 2164 |
+
"loss": 0.2961,
|
| 2165 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 2166 |
+
"rewards/chosen": -1.1022846698760986,
|
| 2167 |
+
"rewards/margins": 1.461648941040039,
|
| 2168 |
+
"rewards/rejected": -2.563933849334717,
|
| 2169 |
+
"step": 720
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 2.9247854618879354,
|
| 2173 |
+
"grad_norm": 155.65060424804688,
|
| 2174 |
+
"learning_rate": 5e-06,
|
| 2175 |
+
"logits/chosen": -1.1305577754974365,
|
| 2176 |
+
"logits/rejected": -1.085231065750122,
|
| 2177 |
+
"logps/chosen": -148.7545928955078,
|
| 2178 |
+
"logps/rejected": -176.36215209960938,
|
| 2179 |
+
"loss": 0.426,
|
| 2180 |
+
"rewards/accuracies": 0.824999988079071,
|
| 2181 |
+
"rewards/chosen": -1.75044846534729,
|
| 2182 |
+
"rewards/margins": 1.306915521621704,
|
| 2183 |
+
"rewards/rejected": -3.057363986968994,
|
| 2184 |
+
"step": 725
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 2.944977284199899,
|
| 2188 |
+
"grad_norm": 83.46701049804688,
|
| 2189 |
+
"learning_rate": 5e-06,
|
| 2190 |
+
"logits/chosen": -1.1913352012634277,
|
| 2191 |
+
"logits/rejected": -1.1349766254425049,
|
| 2192 |
+
"logps/chosen": -132.3568115234375,
|
| 2193 |
+
"logps/rejected": -161.4365234375,
|
| 2194 |
+
"loss": 0.4037,
|
| 2195 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2196 |
+
"rewards/chosen": -1.3909555673599243,
|
| 2197 |
+
"rewards/margins": 1.264525055885315,
|
| 2198 |
+
"rewards/rejected": -2.6554808616638184,
|
| 2199 |
+
"step": 730
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 2.965169106511863,
|
| 2203 |
+
"grad_norm": 29.703113555908203,
|
| 2204 |
+
"learning_rate": 5e-06,
|
| 2205 |
+
"logits/chosen": -1.1975009441375732,
|
| 2206 |
+
"logits/rejected": -1.1907942295074463,
|
| 2207 |
+
"logps/chosen": -126.33477783203125,
|
| 2208 |
+
"logps/rejected": -158.81204223632812,
|
| 2209 |
+
"loss": 0.4499,
|
| 2210 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2211 |
+
"rewards/chosen": -1.3245980739593506,
|
| 2212 |
+
"rewards/margins": 1.180716872215271,
|
| 2213 |
+
"rewards/rejected": -2.5053153038024902,
|
| 2214 |
+
"step": 735
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 2.9853609288238263,
|
| 2218 |
+
"grad_norm": 41.88002014160156,
|
| 2219 |
+
"learning_rate": 5e-06,
|
| 2220 |
+
"logits/chosen": -1.211860179901123,
|
| 2221 |
+
"logits/rejected": -1.1881868839263916,
|
| 2222 |
+
"logps/chosen": -141.8226776123047,
|
| 2223 |
+
"logps/rejected": -175.9602508544922,
|
| 2224 |
+
"loss": 0.3985,
|
| 2225 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 2226 |
+
"rewards/chosen": -1.6403604745864868,
|
| 2227 |
+
"rewards/margins": 1.4641687870025635,
|
| 2228 |
+
"rewards/rejected": -3.1045291423797607,
|
| 2229 |
+
"step": 740
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 2.989399293286219,
|
| 2233 |
+
"step": 741,
|
| 2234 |
+
"total_flos": 0.0,
|
| 2235 |
+
"train_loss": 0.586526518971653,
|
| 2236 |
+
"train_runtime": 26068.2965,
|
| 2237 |
+
"train_samples_per_second": 0.228,
|
| 2238 |
+
"train_steps_per_second": 0.028
|
| 2239 |
+
}
|
| 2240 |
+
],
|
| 2241 |
+
"logging_steps": 5,
|
| 2242 |
+
"max_steps": 741,
|
| 2243 |
+
"num_input_tokens_seen": 0,
|
| 2244 |
+
"num_train_epochs": 3,
|
| 2245 |
+
"save_steps": 500,
|
| 2246 |
+
"stateful_callbacks": {
|
| 2247 |
+
"TrainerControl": {
|
| 2248 |
+
"args": {
|
| 2249 |
+
"should_epoch_stop": false,
|
| 2250 |
+
"should_evaluate": false,
|
| 2251 |
+
"should_log": false,
|
| 2252 |
+
"should_save": true,
|
| 2253 |
+
"should_training_stop": true
|
| 2254 |
+
},
|
| 2255 |
+
"attributes": {}
|
| 2256 |
+
}
|
| 2257 |
+
},
|
| 2258 |
+
"total_flos": 0.0,
|
| 2259 |
+
"train_batch_size": 1,
|
| 2260 |
+
"trial_name": null,
|
| 2261 |
+
"trial_params": null
|
| 2262 |
+
}
|