Model save
Browse files- README.md +67 -0
- adapter_model.safetensors +1 -1
- all_results.json +9 -0
- runs/May05_14-53-20_593bafed1c64/events.out.tfevents.1746456856.593bafed1c64.17067.0 +2 -2
- train_results.json +9 -0
- trainer_state.json +1034 -0
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-7b
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: gemma-7b-orpo-noisy-5e-5-005-v141
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- orpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for gemma-7b-orpo-noisy-5e-5-005-v141
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="silviasapora/gemma-7b-orpo-noisy-5e-5-005-v141", device="cuda")
|
| 24 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 25 |
+
print(output["generated_text"])
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Training procedure
|
| 29 |
+
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/silvias/huggingface/runs/jnv8pbtn)
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
This model was trained with ORPO, a method introduced in [ORPO: Monolithic Preference Optimization without Reference Model](https://huggingface.co/papers/2403.07691).
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.15.2
|
| 38 |
+
- Transformers: 4.49.0
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.2.0
|
| 41 |
+
- Tokenizers: 0.21.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite ORPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{hong2024orpo,
|
| 49 |
+
title = {{ORPO: Monolithic Preference Optimization without Reference Model}},
|
| 50 |
+
author = {Jiwoo Hong and Noah Lee and James Thorne},
|
| 51 |
+
year = 2024,
|
| 52 |
+
eprint = {arXiv:2403.07691}
|
| 53 |
+
}
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
Cite TRL as:
|
| 57 |
+
|
| 58 |
+
```bibtex
|
| 59 |
+
@misc{vonwerra2022trl,
|
| 60 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 61 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 62 |
+
year = 2020,
|
| 63 |
+
journal = {GitHub repository},
|
| 64 |
+
publisher = {GitHub},
|
| 65 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 66 |
+
}
|
| 67 |
+
```
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 800117240
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9634c66a3182d945380a1f3665a0e8fb0bf1a02cbd3060643d90efa5c628525e
|
| 3 |
size 800117240
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.9881235154394297,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 25.76104594499637,
|
| 5 |
+
"train_runtime": 3640.7426,
|
| 6 |
+
"train_samples": 5050,
|
| 7 |
+
"train_samples_per_second": 5.548,
|
| 8 |
+
"train_steps_per_second": 0.086
|
| 9 |
+
}
|
runs/May05_14-53-20_593bafed1c64/events.out.tfevents.1746456856.593bafed1c64.17067.0
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:a3cb97308e8cf6abe5b68ef350182a8b64c8c71fa836e28d9e28950055b69a17
|
| 3 |
+
size 51652
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.9881235154394297,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 25.76104594499637,
|
| 5 |
+
"train_runtime": 3640.7426,
|
| 6 |
+
"train_samples": 5050,
|
| 7 |
+
"train_samples_per_second": 5.548,
|
| 8 |
+
"train_steps_per_second": 0.086
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1034 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.9881235154394297,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 312,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.06334125098970704,
|
| 13 |
+
"grad_norm": 440.0,
|
| 14 |
+
"learning_rate": 7.8125e-06,
|
| 15 |
+
"log_odds_chosen": 3.1748764514923096,
|
| 16 |
+
"log_odds_ratio": -9.884220123291016,
|
| 17 |
+
"logps/chosen": -23.121124267578125,
|
| 18 |
+
"logps/rejected": -26.29541015625,
|
| 19 |
+
"loss": 178.8174,
|
| 20 |
+
"nll_loss": 10.681877136230469,
|
| 21 |
+
"rewards/accuracies": 0.543749988079071,
|
| 22 |
+
"rewards/chosen": -1.1560561656951904,
|
| 23 |
+
"rewards/margins": 0.1587144136428833,
|
| 24 |
+
"rewards/rejected": -1.3147705793380737,
|
| 25 |
+
"step": 5
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.12668250197941408,
|
| 29 |
+
"grad_norm": 340.0,
|
| 30 |
+
"learning_rate": 1.5625e-05,
|
| 31 |
+
"log_odds_chosen": 6.453858852386475,
|
| 32 |
+
"log_odds_ratio": -9.807241439819336,
|
| 33 |
+
"logps/chosen": -22.413124084472656,
|
| 34 |
+
"logps/rejected": -28.864864349365234,
|
| 35 |
+
"loss": 178.3701,
|
| 36 |
+
"nll_loss": 10.657767295837402,
|
| 37 |
+
"rewards/accuracies": 0.578125,
|
| 38 |
+
"rewards/chosen": -1.1206562519073486,
|
| 39 |
+
"rewards/margins": 0.32258713245391846,
|
| 40 |
+
"rewards/rejected": -1.443243384361267,
|
| 41 |
+
"step": 10
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 0.19002375296912113,
|
| 45 |
+
"grad_norm": 276.0,
|
| 46 |
+
"learning_rate": 2.34375e-05,
|
| 47 |
+
"log_odds_chosen": 6.200798988342285,
|
| 48 |
+
"log_odds_ratio": -6.756936073303223,
|
| 49 |
+
"logps/chosen": -19.177919387817383,
|
| 50 |
+
"logps/rejected": -25.377355575561523,
|
| 51 |
+
"loss": 148.0439,
|
| 52 |
+
"nll_loss": 8.914896011352539,
|
| 53 |
+
"rewards/accuracies": 0.559374988079071,
|
| 54 |
+
"rewards/chosen": -0.9588959813117981,
|
| 55 |
+
"rewards/margins": 0.3099718987941742,
|
| 56 |
+
"rewards/rejected": -1.26886785030365,
|
| 57 |
+
"step": 15
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.25336500395882816,
|
| 61 |
+
"grad_norm": 900.0,
|
| 62 |
+
"learning_rate": 3.125e-05,
|
| 63 |
+
"log_odds_chosen": 2.6450843811035156,
|
| 64 |
+
"log_odds_ratio": -5.436456203460693,
|
| 65 |
+
"logps/chosen": -13.422134399414062,
|
| 66 |
+
"logps/rejected": -16.063411712646484,
|
| 67 |
+
"loss": 101.9164,
|
| 68 |
+
"nll_loss": 6.097954750061035,
|
| 69 |
+
"rewards/accuracies": 0.59375,
|
| 70 |
+
"rewards/chosen": -0.6711066961288452,
|
| 71 |
+
"rewards/margins": 0.1320638358592987,
|
| 72 |
+
"rewards/rejected": -0.8031705617904663,
|
| 73 |
+
"step": 20
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.3167062549485352,
|
| 77 |
+
"grad_norm": 93.5,
|
| 78 |
+
"learning_rate": 3.90625e-05,
|
| 79 |
+
"log_odds_chosen": 0.6309150457382202,
|
| 80 |
+
"log_odds_ratio": -1.286942481994629,
|
| 81 |
+
"logps/chosen": -3.097271680831909,
|
| 82 |
+
"logps/rejected": -3.7033658027648926,
|
| 83 |
+
"loss": 42.6767,
|
| 84 |
+
"nll_loss": 2.6029443740844727,
|
| 85 |
+
"rewards/accuracies": 0.5625,
|
| 86 |
+
"rewards/chosen": -0.15486359596252441,
|
| 87 |
+
"rewards/margins": 0.03030467964708805,
|
| 88 |
+
"rewards/rejected": -0.18516826629638672,
|
| 89 |
+
"step": 25
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.38004750593824227,
|
| 93 |
+
"grad_norm": 152.0,
|
| 94 |
+
"learning_rate": 4.6875e-05,
|
| 95 |
+
"log_odds_chosen": 0.14747688174247742,
|
| 96 |
+
"log_odds_ratio": -0.863192081451416,
|
| 97 |
+
"logps/chosen": -1.753333330154419,
|
| 98 |
+
"logps/rejected": -1.8908967971801758,
|
| 99 |
+
"loss": 33.7883,
|
| 100 |
+
"nll_loss": 2.068608283996582,
|
| 101 |
+
"rewards/accuracies": 0.550000011920929,
|
| 102 |
+
"rewards/chosen": -0.08766667544841766,
|
| 103 |
+
"rewards/margins": 0.006878170184791088,
|
| 104 |
+
"rewards/rejected": -0.09454484283924103,
|
| 105 |
+
"step": 30
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.4433887569279493,
|
| 109 |
+
"grad_norm": 320.0,
|
| 110 |
+
"learning_rate": 4.998583895833834e-05,
|
| 111 |
+
"log_odds_chosen": 0.0781320109963417,
|
| 112 |
+
"log_odds_ratio": -0.8847883939743042,
|
| 113 |
+
"logps/chosen": -1.601119041442871,
|
| 114 |
+
"logps/rejected": -1.677050232887268,
|
| 115 |
+
"loss": 30.4438,
|
| 116 |
+
"nll_loss": 1.8584957122802734,
|
| 117 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 118 |
+
"rewards/chosen": -0.08005593717098236,
|
| 119 |
+
"rewards/margins": 0.003796567674726248,
|
| 120 |
+
"rewards/rejected": -0.08385250717401505,
|
| 121 |
+
"step": 35
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.5067300079176563,
|
| 125 |
+
"grad_norm": 79.5,
|
| 126 |
+
"learning_rate": 4.989935734988098e-05,
|
| 127 |
+
"log_odds_chosen": 0.17777779698371887,
|
| 128 |
+
"log_odds_ratio": -0.7818454504013062,
|
| 129 |
+
"logps/chosen": -1.434004545211792,
|
| 130 |
+
"logps/rejected": -1.5908782482147217,
|
| 131 |
+
"loss": 29.0365,
|
| 132 |
+
"nll_loss": 1.7756904363632202,
|
| 133 |
+
"rewards/accuracies": 0.5625,
|
| 134 |
+
"rewards/chosen": -0.07170023769140244,
|
| 135 |
+
"rewards/margins": 0.00784367136657238,
|
| 136 |
+
"rewards/rejected": -0.07954390347003937,
|
| 137 |
+
"step": 40
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.5700712589073634,
|
| 141 |
+
"grad_norm": 46.0,
|
| 142 |
+
"learning_rate": 4.973453315132592e-05,
|
| 143 |
+
"log_odds_chosen": 0.12557587027549744,
|
| 144 |
+
"log_odds_ratio": -0.7924664616584778,
|
| 145 |
+
"logps/chosen": -1.366229772567749,
|
| 146 |
+
"logps/rejected": -1.46891450881958,
|
| 147 |
+
"loss": 27.2712,
|
| 148 |
+
"nll_loss": 1.6648271083831787,
|
| 149 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 150 |
+
"rewards/chosen": -0.06831149756908417,
|
| 151 |
+
"rewards/margins": 0.00513423141092062,
|
| 152 |
+
"rewards/rejected": -0.07344572246074677,
|
| 153 |
+
"step": 45
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.6334125098970704,
|
| 157 |
+
"grad_norm": 322.0,
|
| 158 |
+
"learning_rate": 4.9491884960580894e-05,
|
| 159 |
+
"log_odds_chosen": 0.2877430319786072,
|
| 160 |
+
"log_odds_ratio": -0.712502121925354,
|
| 161 |
+
"logps/chosen": -1.2252671718597412,
|
| 162 |
+
"logps/rejected": -1.4585528373718262,
|
| 163 |
+
"loss": 25.6522,
|
| 164 |
+
"nll_loss": 1.5676389932632446,
|
| 165 |
+
"rewards/accuracies": 0.59375,
|
| 166 |
+
"rewards/chosen": -0.06126336008310318,
|
| 167 |
+
"rewards/margins": 0.011664283461868763,
|
| 168 |
+
"rewards/rejected": -0.07292764633893967,
|
| 169 |
+
"step": 50
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 0.6967537608867775,
|
| 173 |
+
"grad_norm": 74.0,
|
| 174 |
+
"learning_rate": 4.9172176238625164e-05,
|
| 175 |
+
"log_odds_chosen": 0.12641319632530212,
|
| 176 |
+
"log_odds_ratio": -0.7623997926712036,
|
| 177 |
+
"logps/chosen": -1.275046467781067,
|
| 178 |
+
"logps/rejected": -1.3824188709259033,
|
| 179 |
+
"loss": 25.7811,
|
| 180 |
+
"nll_loss": 1.5732007026672363,
|
| 181 |
+
"rewards/accuracies": 0.5218750238418579,
|
| 182 |
+
"rewards/chosen": -0.06375232338905334,
|
| 183 |
+
"rewards/margins": 0.005368619225919247,
|
| 184 |
+
"rewards/rejected": -0.06912095099687576,
|
| 185 |
+
"step": 55
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.7600950118764845,
|
| 189 |
+
"grad_norm": 49.25,
|
| 190 |
+
"learning_rate": 4.877641290737884e-05,
|
| 191 |
+
"log_odds_chosen": 0.19897113740444183,
|
| 192 |
+
"log_odds_ratio": -0.7030541300773621,
|
| 193 |
+
"logps/chosen": -1.1849174499511719,
|
| 194 |
+
"logps/rejected": -1.3446745872497559,
|
| 195 |
+
"loss": 25.5335,
|
| 196 |
+
"nll_loss": 1.5606908798217773,
|
| 197 |
+
"rewards/accuracies": 0.581250011920929,
|
| 198 |
+
"rewards/chosen": -0.05924587324261665,
|
| 199 |
+
"rewards/margins": 0.007987860590219498,
|
| 200 |
+
"rewards/rejected": -0.06723372638225555,
|
| 201 |
+
"step": 60
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.8234362628661916,
|
| 205 |
+
"grad_norm": 73.0,
|
| 206 |
+
"learning_rate": 4.8305840184700356e-05,
|
| 207 |
+
"log_odds_chosen": 0.2491961419582367,
|
| 208 |
+
"log_odds_ratio": -0.693637490272522,
|
| 209 |
+
"logps/chosen": -1.1569511890411377,
|
| 210 |
+
"logps/rejected": -1.3474953174591064,
|
| 211 |
+
"loss": 25.4002,
|
| 212 |
+
"nll_loss": 1.5528333187103271,
|
| 213 |
+
"rewards/accuracies": 0.574999988079071,
|
| 214 |
+
"rewards/chosen": -0.057847559452056885,
|
| 215 |
+
"rewards/margins": 0.009527203626930714,
|
| 216 |
+
"rewards/rejected": -0.06737476587295532,
|
| 217 |
+
"step": 65
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 0.8867775138558986,
|
| 221 |
+
"grad_norm": 62.75,
|
| 222 |
+
"learning_rate": 4.7761938666470403e-05,
|
| 223 |
+
"log_odds_chosen": 0.1128716841340065,
|
| 224 |
+
"log_odds_ratio": -0.76786869764328,
|
| 225 |
+
"logps/chosen": -1.2037312984466553,
|
| 226 |
+
"logps/rejected": -1.294019103050232,
|
| 227 |
+
"loss": 25.6293,
|
| 228 |
+
"nll_loss": 1.5634351968765259,
|
| 229 |
+
"rewards/accuracies": 0.518750011920929,
|
| 230 |
+
"rewards/chosen": -0.060186564922332764,
|
| 231 |
+
"rewards/margins": 0.004514397121965885,
|
| 232 |
+
"rewards/rejected": -0.06470096111297607,
|
| 233 |
+
"step": 70
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.9501187648456056,
|
| 237 |
+
"grad_norm": 43.25,
|
| 238 |
+
"learning_rate": 4.71464196680895e-05,
|
| 239 |
+
"log_odds_chosen": 0.15465858578681946,
|
| 240 |
+
"log_odds_ratio": -0.7334991693496704,
|
| 241 |
+
"logps/chosen": -1.1338040828704834,
|
| 242 |
+
"logps/rejected": -1.2355806827545166,
|
| 243 |
+
"loss": 23.9048,
|
| 244 |
+
"nll_loss": 1.457373023033142,
|
| 245 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 246 |
+
"rewards/chosen": -0.05669020861387253,
|
| 247 |
+
"rewards/margins": 0.005088826175779104,
|
| 248 |
+
"rewards/rejected": -0.06177903339266777,
|
| 249 |
+
"step": 75
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.0253365003958828,
|
| 253 |
+
"grad_norm": 39.25,
|
| 254 |
+
"learning_rate": 4.6461219840046654e-05,
|
| 255 |
+
"log_odds_chosen": 0.339913547039032,
|
| 256 |
+
"log_odds_ratio": -0.6567232608795166,
|
| 257 |
+
"logps/chosen": -1.0351102352142334,
|
| 258 |
+
"logps/rejected": -1.276901364326477,
|
| 259 |
+
"loss": 26.0349,
|
| 260 |
+
"nll_loss": 1.4135476350784302,
|
| 261 |
+
"rewards/accuracies": 0.6111111044883728,
|
| 262 |
+
"rewards/chosen": -0.05175551399588585,
|
| 263 |
+
"rewards/margins": 0.012089554220438004,
|
| 264 |
+
"rewards/rejected": -0.06384507566690445,
|
| 265 |
+
"step": 80
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.0886777513855899,
|
| 269 |
+
"grad_norm": 39.5,
|
| 270 |
+
"learning_rate": 4.5708495074500414e-05,
|
| 271 |
+
"log_odds_chosen": 0.5737000703811646,
|
| 272 |
+
"log_odds_ratio": -0.5414241552352905,
|
| 273 |
+
"logps/chosen": -0.8744705319404602,
|
| 274 |
+
"logps/rejected": -1.2483097314834595,
|
| 275 |
+
"loss": 19.751,
|
| 276 |
+
"nll_loss": 1.2073694467544556,
|
| 277 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 278 |
+
"rewards/chosen": -0.04372352734208107,
|
| 279 |
+
"rewards/margins": 0.018691958859562874,
|
| 280 |
+
"rewards/rejected": -0.062415480613708496,
|
| 281 |
+
"step": 85
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.152019002375297,
|
| 285 |
+
"grad_norm": 41.25,
|
| 286 |
+
"learning_rate": 4.489061372204453e-05,
|
| 287 |
+
"log_odds_chosen": 0.6578646302223206,
|
| 288 |
+
"log_odds_ratio": -0.5210095047950745,
|
| 289 |
+
"logps/chosen": -0.8167413473129272,
|
| 290 |
+
"logps/rejected": -1.2592405080795288,
|
| 291 |
+
"loss": 19.3149,
|
| 292 |
+
"nll_loss": 1.181132197380066,
|
| 293 |
+
"rewards/accuracies": 0.734375,
|
| 294 |
+
"rewards/chosen": -0.04083707183599472,
|
| 295 |
+
"rewards/margins": 0.02212495729327202,
|
| 296 |
+
"rewards/rejected": -0.06296203285455704,
|
| 297 |
+
"step": 90
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 1.2153602533650039,
|
| 301 |
+
"grad_norm": 27.0,
|
| 302 |
+
"learning_rate": 4.401014914000078e-05,
|
| 303 |
+
"log_odds_chosen": 0.5433515310287476,
|
| 304 |
+
"log_odds_ratio": -0.5697063207626343,
|
| 305 |
+
"logps/chosen": -0.8254748582839966,
|
| 306 |
+
"logps/rejected": -1.1539433002471924,
|
| 307 |
+
"loss": 19.3225,
|
| 308 |
+
"nll_loss": 1.179168701171875,
|
| 309 |
+
"rewards/accuracies": 0.690625011920929,
|
| 310 |
+
"rewards/chosen": -0.04127374291419983,
|
| 311 |
+
"rewards/margins": 0.016423424705863,
|
| 312 |
+
"rewards/rejected": -0.05769716575741768,
|
| 313 |
+
"step": 95
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 1.278701504354711,
|
| 317 |
+
"grad_norm": 79.5,
|
| 318 |
+
"learning_rate": 4.306987159568479e-05,
|
| 319 |
+
"log_odds_chosen": 0.592140793800354,
|
| 320 |
+
"log_odds_ratio": -0.5446491241455078,
|
| 321 |
+
"logps/chosen": -0.8890258073806763,
|
| 322 |
+
"logps/rejected": -1.2678449153900146,
|
| 323 |
+
"loss": 19.6595,
|
| 324 |
+
"nll_loss": 1.2014886140823364,
|
| 325 |
+
"rewards/accuracies": 0.703125,
|
| 326 |
+
"rewards/chosen": -0.044451288878917694,
|
| 327 |
+
"rewards/margins": 0.018940959125757217,
|
| 328 |
+
"rewards/rejected": -0.06339225172996521,
|
| 329 |
+
"step": 100
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.342042755344418,
|
| 333 |
+
"grad_norm": 26.25,
|
| 334 |
+
"learning_rate": 4.207273955012018e-05,
|
| 335 |
+
"log_odds_chosen": 0.5168221592903137,
|
| 336 |
+
"log_odds_ratio": -0.5695356130599976,
|
| 337 |
+
"logps/chosen": -0.8439987897872925,
|
| 338 |
+
"logps/rejected": -1.158204197883606,
|
| 339 |
+
"loss": 18.8408,
|
| 340 |
+
"nll_loss": 1.1490731239318848,
|
| 341 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 342 |
+
"rewards/chosen": -0.04219993203878403,
|
| 343 |
+
"rewards/margins": 0.01571027934551239,
|
| 344 |
+
"rewards/rejected": -0.05791021138429642,
|
| 345 |
+
"step": 105
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.405384006334125,
|
| 349 |
+
"grad_norm": 32.0,
|
| 350 |
+
"learning_rate": 4.10218903496256e-05,
|
| 351 |
+
"log_odds_chosen": 0.5098962187767029,
|
| 352 |
+
"log_odds_ratio": -0.5828264355659485,
|
| 353 |
+
"logps/chosen": -0.8689939379692078,
|
| 354 |
+
"logps/rejected": -1.1937323808670044,
|
| 355 |
+
"loss": 20.1778,
|
| 356 |
+
"nll_loss": 1.231968641281128,
|
| 357 |
+
"rewards/accuracies": 0.6781250238418579,
|
| 358 |
+
"rewards/chosen": -0.04344969242811203,
|
| 359 |
+
"rewards/margins": 0.0162369217723608,
|
| 360 |
+
"rewards/rejected": -0.05968661978840828,
|
| 361 |
+
"step": 110
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 1.468725257323832,
|
| 365 |
+
"grad_norm": 34.0,
|
| 366 |
+
"learning_rate": 3.992063035456259e-05,
|
| 367 |
+
"log_odds_chosen": 0.6211134195327759,
|
| 368 |
+
"log_odds_ratio": -0.5309170484542847,
|
| 369 |
+
"logps/chosen": -0.7933008074760437,
|
| 370 |
+
"logps/rejected": -1.1834070682525635,
|
| 371 |
+
"loss": 18.7885,
|
| 372 |
+
"nll_loss": 1.14773428440094,
|
| 373 |
+
"rewards/accuracies": 0.734375,
|
| 374 |
+
"rewards/chosen": -0.039665043354034424,
|
| 375 |
+
"rewards/margins": 0.01950531266629696,
|
| 376 |
+
"rewards/rejected": -0.059170354157686234,
|
| 377 |
+
"step": 115
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.5320665083135392,
|
| 381 |
+
"grad_norm": 28.5,
|
| 382 |
+
"learning_rate": 3.8772424536302564e-05,
|
| 383 |
+
"log_odds_chosen": 0.6670337915420532,
|
| 384 |
+
"log_odds_ratio": -0.5194506049156189,
|
| 385 |
+
"logps/chosen": -0.7793987989425659,
|
| 386 |
+
"logps/rejected": -1.1679680347442627,
|
| 387 |
+
"loss": 18.5035,
|
| 388 |
+
"nll_loss": 1.1304987668991089,
|
| 389 |
+
"rewards/accuracies": 0.734375,
|
| 390 |
+
"rewards/chosen": -0.038969945162534714,
|
| 391 |
+
"rewards/margins": 0.01942845992743969,
|
| 392 |
+
"rewards/rejected": -0.058398403227329254,
|
| 393 |
+
"step": 120
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 1.5954077593032463,
|
| 397 |
+
"grad_norm": 24.375,
|
| 398 |
+
"learning_rate": 3.758088557514501e-05,
|
| 399 |
+
"log_odds_chosen": 0.5610660314559937,
|
| 400 |
+
"log_odds_ratio": -0.552033543586731,
|
| 401 |
+
"logps/chosen": -0.795187771320343,
|
| 402 |
+
"logps/rejected": -1.1401742696762085,
|
| 403 |
+
"loss": 18.7339,
|
| 404 |
+
"nll_loss": 1.1432664394378662,
|
| 405 |
+
"rewards/accuracies": 0.684374988079071,
|
| 406 |
+
"rewards/chosen": -0.03975939005613327,
|
| 407 |
+
"rewards/margins": 0.017249327152967453,
|
| 408 |
+
"rewards/rejected": -0.05700872093439102,
|
| 409 |
+
"step": 125
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.6587490102929534,
|
| 413 |
+
"grad_norm": 29.625,
|
| 414 |
+
"learning_rate": 3.634976249348867e-05,
|
| 415 |
+
"log_odds_chosen": 0.4703447222709656,
|
| 416 |
+
"log_odds_ratio": -0.5870831608772278,
|
| 417 |
+
"logps/chosen": -0.7940428256988525,
|
| 418 |
+
"logps/rejected": -1.0778682231903076,
|
| 419 |
+
"loss": 18.7796,
|
| 420 |
+
"nll_loss": 1.1443703174591064,
|
| 421 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 422 |
+
"rewards/chosen": -0.039702143520116806,
|
| 423 |
+
"rewards/margins": 0.014191272668540478,
|
| 424 |
+
"rewards/rejected": -0.05389341711997986,
|
| 425 |
+
"step": 130
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.7220902612826603,
|
| 429 |
+
"grad_norm": 28.875,
|
| 430 |
+
"learning_rate": 3.508292886002013e-05,
|
| 431 |
+
"log_odds_chosen": 0.5749887228012085,
|
| 432 |
+
"log_odds_ratio": -0.5368548631668091,
|
| 433 |
+
"logps/chosen": -0.8258606195449829,
|
| 434 |
+
"logps/rejected": -1.1885448694229126,
|
| 435 |
+
"loss": 18.9446,
|
| 436 |
+
"nll_loss": 1.1571968793869019,
|
| 437 |
+
"rewards/accuracies": 0.71875,
|
| 438 |
+
"rewards/chosen": -0.04129302501678467,
|
| 439 |
+
"rewards/margins": 0.018134210258722305,
|
| 440 |
+
"rewards/rejected": -0.05942724272608757,
|
| 441 |
+
"step": 135
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.7854315122723674,
|
| 445 |
+
"grad_norm": 26.75,
|
| 446 |
+
"learning_rate": 3.378437060203357e-05,
|
| 447 |
+
"log_odds_chosen": 0.5498477220535278,
|
| 448 |
+
"log_odds_ratio": -0.5687417984008789,
|
| 449 |
+
"logps/chosen": -0.81671142578125,
|
| 450 |
+
"logps/rejected": -1.1441940069198608,
|
| 451 |
+
"loss": 18.5467,
|
| 452 |
+
"nll_loss": 1.130732774734497,
|
| 453 |
+
"rewards/accuracies": 0.6781250238418579,
|
| 454 |
+
"rewards/chosen": -0.04083556681871414,
|
| 455 |
+
"rewards/margins": 0.0163741298019886,
|
| 456 |
+
"rewards/rejected": -0.05720970779657364,
|
| 457 |
+
"step": 140
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.8487727632620743,
|
| 461 |
+
"grad_norm": 36.5,
|
| 462 |
+
"learning_rate": 3.24581734642289e-05,
|
| 463 |
+
"log_odds_chosen": 0.7345950603485107,
|
| 464 |
+
"log_odds_ratio": -0.5215600728988647,
|
| 465 |
+
"logps/chosen": -0.7965916991233826,
|
| 466 |
+
"logps/rejected": -1.2689704895019531,
|
| 467 |
+
"loss": 18.424,
|
| 468 |
+
"nll_loss": 1.125421404838562,
|
| 469 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 470 |
+
"rewards/chosen": -0.03982958942651749,
|
| 471 |
+
"rewards/margins": 0.02361893653869629,
|
| 472 |
+
"rewards/rejected": -0.06344853341579437,
|
| 473 |
+
"step": 145
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.9121140142517814,
|
| 477 |
+
"grad_norm": 30.5,
|
| 478 |
+
"learning_rate": 3.110851015344735e-05,
|
| 479 |
+
"log_odds_chosen": 0.5858422517776489,
|
| 480 |
+
"log_odds_ratio": -0.5331650376319885,
|
| 481 |
+
"logps/chosen": -0.8209154009819031,
|
| 482 |
+
"logps/rejected": -1.1717817783355713,
|
| 483 |
+
"loss": 18.0192,
|
| 484 |
+
"nll_loss": 1.0995434522628784,
|
| 485 |
+
"rewards/accuracies": 0.7281249761581421,
|
| 486 |
+
"rewards/chosen": -0.04104577749967575,
|
| 487 |
+
"rewards/margins": 0.017543310299515724,
|
| 488 |
+
"rewards/rejected": -0.058589089661836624,
|
| 489 |
+
"step": 150
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.9754552652414885,
|
| 493 |
+
"grad_norm": 31.5,
|
| 494 |
+
"learning_rate": 2.9739627209791965e-05,
|
| 495 |
+
"log_odds_chosen": 0.6027151346206665,
|
| 496 |
+
"log_odds_ratio": -0.5478641390800476,
|
| 497 |
+
"logps/chosen": -0.8092836141586304,
|
| 498 |
+
"logps/rejected": -1.1755163669586182,
|
| 499 |
+
"loss": 18.8905,
|
| 500 |
+
"nll_loss": 1.1532632112503052,
|
| 501 |
+
"rewards/accuracies": 0.699999988079071,
|
| 502 |
+
"rewards/chosen": -0.04046417772769928,
|
| 503 |
+
"rewards/margins": 0.0183116365224123,
|
| 504 |
+
"rewards/rejected": -0.05877581983804703,
|
| 505 |
+
"step": 155
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 2.0506730007917655,
|
| 509 |
+
"grad_norm": 26.625,
|
| 510 |
+
"learning_rate": 2.8355831645441388e-05,
|
| 511 |
+
"log_odds_chosen": 0.9976516962051392,
|
| 512 |
+
"log_odds_ratio": -0.412594735622406,
|
| 513 |
+
"logps/chosen": -0.6453574299812317,
|
| 514 |
+
"logps/rejected": -1.2032675743103027,
|
| 515 |
+
"loss": 17.6716,
|
| 516 |
+
"nll_loss": 0.9611276984214783,
|
| 517 |
+
"rewards/accuracies": 0.8333333134651184,
|
| 518 |
+
"rewards/chosen": -0.032267872244119644,
|
| 519 |
+
"rewards/margins": 0.027895506471395493,
|
| 520 |
+
"rewards/rejected": -0.06016337499022484,
|
| 521 |
+
"step": 160
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 2.1140142517814726,
|
| 525 |
+
"grad_norm": 25.125,
|
| 526 |
+
"learning_rate": 2.6961477393196126e-05,
|
| 527 |
+
"log_odds_chosen": 1.592689871788025,
|
| 528 |
+
"log_odds_ratio": -0.27941542863845825,
|
| 529 |
+
"logps/chosen": -0.5471819639205933,
|
| 530 |
+
"logps/rejected": -1.4586117267608643,
|
| 531 |
+
"loss": 13.9962,
|
| 532 |
+
"nll_loss": 0.8607938885688782,
|
| 533 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 534 |
+
"rewards/chosen": -0.027359094470739365,
|
| 535 |
+
"rewards/margins": 0.04557149484753609,
|
| 536 |
+
"rewards/rejected": -0.07293058931827545,
|
| 537 |
+
"step": 165
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 2.1773555027711797,
|
| 541 |
+
"grad_norm": 23.125,
|
| 542 |
+
"learning_rate": 2.556095160739513e-05,
|
| 543 |
+
"log_odds_chosen": 1.3663856983184814,
|
| 544 |
+
"log_odds_ratio": -0.3143647313117981,
|
| 545 |
+
"logps/chosen": -0.5476871728897095,
|
| 546 |
+
"logps/rejected": -1.304485559463501,
|
| 547 |
+
"loss": 13.0544,
|
| 548 |
+
"nll_loss": 0.800182044506073,
|
| 549 |
+
"rewards/accuracies": 0.925000011920929,
|
| 550 |
+
"rewards/chosen": -0.027384359389543533,
|
| 551 |
+
"rewards/margins": 0.037839919328689575,
|
| 552 |
+
"rewards/rejected": -0.06522427499294281,
|
| 553 |
+
"step": 170
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 2.240696753760887,
|
| 557 |
+
"grad_norm": 26.75,
|
| 558 |
+
"learning_rate": 2.415866086030516e-05,
|
| 559 |
+
"log_odds_chosen": 1.4820417165756226,
|
| 560 |
+
"log_odds_ratio": -0.29762935638427734,
|
| 561 |
+
"logps/chosen": -0.5468381643295288,
|
| 562 |
+
"logps/rejected": -1.3758976459503174,
|
| 563 |
+
"loss": 13.3913,
|
| 564 |
+
"nll_loss": 0.8220744132995605,
|
| 565 |
+
"rewards/accuracies": 0.921875,
|
| 566 |
+
"rewards/chosen": -0.02734190784394741,
|
| 567 |
+
"rewards/margins": 0.04145296663045883,
|
| 568 |
+
"rewards/rejected": -0.06879488378763199,
|
| 569 |
+
"step": 175
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 2.304038004750594,
|
| 573 |
+
"grad_norm": 26.25,
|
| 574 |
+
"learning_rate": 2.2759017277414166e-05,
|
| 575 |
+
"log_odds_chosen": 1.4541139602661133,
|
| 576 |
+
"log_odds_ratio": -0.30560511350631714,
|
| 577 |
+
"logps/chosen": -0.5400351285934448,
|
| 578 |
+
"logps/rejected": -1.3224079608917236,
|
| 579 |
+
"loss": 13.3401,
|
| 580 |
+
"nll_loss": 0.818473219871521,
|
| 581 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 582 |
+
"rewards/chosen": -0.0270017571747303,
|
| 583 |
+
"rewards/margins": 0.03911864012479782,
|
| 584 |
+
"rewards/rejected": -0.06612040102481842,
|
| 585 |
+
"step": 180
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 2.367379255740301,
|
| 589 |
+
"grad_norm": 45.0,
|
| 590 |
+
"learning_rate": 2.136642465525219e-05,
|
| 591 |
+
"log_odds_chosen": 1.3566925525665283,
|
| 592 |
+
"log_odds_ratio": -0.312085896730423,
|
| 593 |
+
"logps/chosen": -0.5589757561683655,
|
| 594 |
+
"logps/rejected": -1.2790095806121826,
|
| 595 |
+
"loss": 13.2166,
|
| 596 |
+
"nll_loss": 0.8104313015937805,
|
| 597 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 598 |
+
"rewards/chosen": -0.027948787435889244,
|
| 599 |
+
"rewards/margins": 0.03600168973207474,
|
| 600 |
+
"rewards/rejected": -0.06395047903060913,
|
| 601 |
+
"step": 185
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 2.4307205067300077,
|
| 605 |
+
"grad_norm": 31.0,
|
| 606 |
+
"learning_rate": 1.9985264605418183e-05,
|
| 607 |
+
"log_odds_chosen": 1.4590914249420166,
|
| 608 |
+
"log_odds_ratio": -0.3126595616340637,
|
| 609 |
+
"logps/chosen": -0.5508120059967041,
|
| 610 |
+
"logps/rejected": -1.3448957204818726,
|
| 611 |
+
"loss": 13.3402,
|
| 612 |
+
"nll_loss": 0.8181284666061401,
|
| 613 |
+
"rewards/accuracies": 0.890625,
|
| 614 |
+
"rewards/chosen": -0.027540603652596474,
|
| 615 |
+
"rewards/margins": 0.03970418870449066,
|
| 616 |
+
"rewards/rejected": -0.06724478304386139,
|
| 617 |
+
"step": 190
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 2.494061757719715,
|
| 621 |
+
"grad_norm": 25.375,
|
| 622 |
+
"learning_rate": 1.8619882768408935e-05,
|
| 623 |
+
"log_odds_chosen": 1.5256664752960205,
|
| 624 |
+
"log_odds_ratio": -0.29186081886291504,
|
| 625 |
+
"logps/chosen": -0.5439263582229614,
|
| 626 |
+
"logps/rejected": -1.4009654521942139,
|
| 627 |
+
"loss": 13.2504,
|
| 628 |
+
"nll_loss": 0.8135560750961304,
|
| 629 |
+
"rewards/accuracies": 0.918749988079071,
|
| 630 |
+
"rewards/chosen": -0.02719631791114807,
|
| 631 |
+
"rewards/margins": 0.04285195842385292,
|
| 632 |
+
"rewards/rejected": -0.07004828751087189,
|
| 633 |
+
"step": 195
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.557403008709422,
|
| 637 |
+
"grad_norm": 24.25,
|
| 638 |
+
"learning_rate": 1.7274575140626318e-05,
|
| 639 |
+
"log_odds_chosen": 1.4801175594329834,
|
| 640 |
+
"log_odds_ratio": -0.3182724118232727,
|
| 641 |
+
"logps/chosen": -0.5556176900863647,
|
| 642 |
+
"logps/rejected": -1.355858564376831,
|
| 643 |
+
"loss": 13.7318,
|
| 644 |
+
"nll_loss": 0.8423222303390503,
|
| 645 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 646 |
+
"rewards/chosen": -0.027780886739492416,
|
| 647 |
+
"rewards/margins": 0.04001203924417496,
|
| 648 |
+
"rewards/rejected": -0.06779292970895767,
|
| 649 |
+
"step": 200
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 2.620744259699129,
|
| 653 |
+
"grad_norm": 21.25,
|
| 654 |
+
"learning_rate": 1.59535745575832e-05,
|
| 655 |
+
"log_odds_chosen": 1.538774847984314,
|
| 656 |
+
"log_odds_ratio": -0.2739660441875458,
|
| 657 |
+
"logps/chosen": -0.5107089281082153,
|
| 658 |
+
"logps/rejected": -1.3470090627670288,
|
| 659 |
+
"loss": 12.7697,
|
| 660 |
+
"nll_loss": 0.7844074368476868,
|
| 661 |
+
"rewards/accuracies": 0.9375,
|
| 662 |
+
"rewards/chosen": -0.025535445660352707,
|
| 663 |
+
"rewards/margins": 0.041815001517534256,
|
| 664 |
+
"rewards/rejected": -0.06735045462846756,
|
| 665 |
+
"step": 205
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 2.684085510688836,
|
| 669 |
+
"grad_norm": 26.5,
|
| 670 |
+
"learning_rate": 1.466103737583699e-05,
|
| 671 |
+
"log_odds_chosen": 1.6077247858047485,
|
| 672 |
+
"log_odds_ratio": -0.27898281812667847,
|
| 673 |
+
"logps/chosen": -0.5174364447593689,
|
| 674 |
+
"logps/rejected": -1.3971433639526367,
|
| 675 |
+
"loss": 12.9226,
|
| 676 |
+
"nll_loss": 0.7937139272689819,
|
| 677 |
+
"rewards/accuracies": 0.909375011920929,
|
| 678 |
+
"rewards/chosen": -0.025871824473142624,
|
| 679 |
+
"rewards/margins": 0.04398534819483757,
|
| 680 |
+
"rewards/rejected": -0.0698571726679802,
|
| 681 |
+
"step": 210
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 2.7474267616785433,
|
| 685 |
+
"grad_norm": 37.5,
|
| 686 |
+
"learning_rate": 1.340103039555415e-05,
|
| 687 |
+
"log_odds_chosen": 1.4663047790527344,
|
| 688 |
+
"log_odds_ratio": -0.3185238242149353,
|
| 689 |
+
"logps/chosen": -0.5488572716712952,
|
| 690 |
+
"logps/rejected": -1.336905837059021,
|
| 691 |
+
"loss": 13.052,
|
| 692 |
+
"nll_loss": 0.7998241782188416,
|
| 693 |
+
"rewards/accuracies": 0.903124988079071,
|
| 694 |
+
"rewards/chosen": -0.02744286321103573,
|
| 695 |
+
"rewards/margins": 0.03940243273973465,
|
| 696 |
+
"rewards/rejected": -0.06684528291225433,
|
| 697 |
+
"step": 215
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 2.81076801266825,
|
| 701 |
+
"grad_norm": 33.25,
|
| 702 |
+
"learning_rate": 1.217751806485235e-05,
|
| 703 |
+
"log_odds_chosen": 1.548749327659607,
|
| 704 |
+
"log_odds_ratio": -0.2925991415977478,
|
| 705 |
+
"logps/chosen": -0.5320712327957153,
|
| 706 |
+
"logps/rejected": -1.400617241859436,
|
| 707 |
+
"loss": 13.2408,
|
| 708 |
+
"nll_loss": 0.8129196166992188,
|
| 709 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 710 |
+
"rewards/chosen": -0.026603559032082558,
|
| 711 |
+
"rewards/margins": 0.043427299708127975,
|
| 712 |
+
"rewards/rejected": -0.07003085315227509,
|
| 713 |
+
"step": 220
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 2.8741092636579575,
|
| 717 |
+
"grad_norm": 28.25,
|
| 718 |
+
"learning_rate": 1.0994350006179931e-05,
|
| 719 |
+
"log_odds_chosen": 1.4858758449554443,
|
| 720 |
+
"log_odds_ratio": -0.29021698236465454,
|
| 721 |
+
"logps/chosen": -0.5439127683639526,
|
| 722 |
+
"logps/rejected": -1.3659403324127197,
|
| 723 |
+
"loss": 13.1169,
|
| 724 |
+
"nll_loss": 0.8052948117256165,
|
| 725 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 726 |
+
"rewards/chosen": -0.027195638045668602,
|
| 727 |
+
"rewards/margins": 0.04110138118267059,
|
| 728 |
+
"rewards/rejected": -0.06829701364040375,
|
| 729 |
+
"step": 225
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 2.937450514647664,
|
| 733 |
+
"grad_norm": 23.125,
|
| 734 |
+
"learning_rate": 9.855248903979506e-06,
|
| 735 |
+
"log_odds_chosen": 1.4533650875091553,
|
| 736 |
+
"log_odds_ratio": -0.2951867878437042,
|
| 737 |
+
"logps/chosen": -0.5295075178146362,
|
| 738 |
+
"logps/rejected": -1.2960364818572998,
|
| 739 |
+
"loss": 13.1057,
|
| 740 |
+
"nll_loss": 0.8043462038040161,
|
| 741 |
+
"rewards/accuracies": 0.918749988079071,
|
| 742 |
+
"rewards/chosen": -0.026475373655557632,
|
| 743 |
+
"rewards/margins": 0.0383264534175396,
|
| 744 |
+
"rewards/rejected": -0.06480182707309723,
|
| 745 |
+
"step": 230
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 3.0126682501979416,
|
| 749 |
+
"grad_norm": 29.375,
|
| 750 |
+
"learning_rate": 8.763798791745411e-06,
|
| 751 |
+
"log_odds_chosen": 1.5862672328948975,
|
| 752 |
+
"log_odds_ratio": -0.2758994698524475,
|
| 753 |
+
"logps/chosen": -0.5054166316986084,
|
| 754 |
+
"logps/rejected": -1.356021761894226,
|
| 755 |
+
"loss": 14.3167,
|
| 756 |
+
"nll_loss": 0.7815780639648438,
|
| 757 |
+
"rewards/accuracies": 0.9222221970558167,
|
| 758 |
+
"rewards/chosen": -0.02527083270251751,
|
| 759 |
+
"rewards/margins": 0.042530253529548645,
|
| 760 |
+
"rewards/rejected": -0.0678010806441307,
|
| 761 |
+
"step": 235
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 3.0760095011876483,
|
| 765 |
+
"grad_norm": 21.25,
|
| 766 |
+
"learning_rate": 7.723433775328384e-06,
|
| 767 |
+
"log_odds_chosen": 2.0705292224884033,
|
| 768 |
+
"log_odds_ratio": -0.19949007034301758,
|
| 769 |
+
"logps/chosen": -0.4086482524871826,
|
| 770 |
+
"logps/rejected": -1.4631153345108032,
|
| 771 |
+
"loss": 10.5377,
|
| 772 |
+
"nll_loss": 0.6486285924911499,
|
| 773 |
+
"rewards/accuracies": 0.965624988079071,
|
| 774 |
+
"rewards/chosen": -0.02043241262435913,
|
| 775 |
+
"rewards/margins": 0.05272335559129715,
|
| 776 |
+
"rewards/rejected": -0.07315576076507568,
|
| 777 |
+
"step": 240
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 3.1393507521773554,
|
| 781 |
+
"grad_norm": 30.0,
|
| 782 |
+
"learning_rate": 6.737427227968063e-06,
|
| 783 |
+
"log_odds_chosen": 2.158939838409424,
|
| 784 |
+
"log_odds_ratio": -0.17655470967292786,
|
| 785 |
+
"logps/chosen": -0.36677083373069763,
|
| 786 |
+
"logps/rejected": -1.4107964038848877,
|
| 787 |
+
"loss": 10.2002,
|
| 788 |
+
"nll_loss": 0.6286830902099609,
|
| 789 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 790 |
+
"rewards/chosen": -0.01833854243159294,
|
| 791 |
+
"rewards/margins": 0.0522012785077095,
|
| 792 |
+
"rewards/rejected": -0.07053982466459274,
|
| 793 |
+
"step": 245
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 3.2026920031670625,
|
| 797 |
+
"grad_norm": 24.375,
|
| 798 |
+
"learning_rate": 5.808881491049723e-06,
|
| 799 |
+
"log_odds_chosen": 2.1467604637145996,
|
| 800 |
+
"log_odds_ratio": -0.21470721065998077,
|
| 801 |
+
"logps/chosen": -0.45862817764282227,
|
| 802 |
+
"logps/rejected": -1.5745422840118408,
|
| 803 |
+
"loss": 10.6424,
|
| 804 |
+
"nll_loss": 0.6544127464294434,
|
| 805 |
+
"rewards/accuracies": 0.953125,
|
| 806 |
+
"rewards/chosen": -0.022931408137083054,
|
| 807 |
+
"rewards/margins": 0.05579570680856705,
|
| 808 |
+
"rewards/rejected": -0.0787271112203598,
|
| 809 |
+
"step": 250
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.2660332541567696,
|
| 813 |
+
"grad_norm": 28.375,
|
| 814 |
+
"learning_rate": 4.940718112990553e-06,
|
| 815 |
+
"log_odds_chosen": 2.224365711212158,
|
| 816 |
+
"log_odds_ratio": -0.18038669228553772,
|
| 817 |
+
"logps/chosen": -0.38552433252334595,
|
| 818 |
+
"logps/rejected": -1.5169192552566528,
|
| 819 |
+
"loss": 10.064,
|
| 820 |
+
"nll_loss": 0.6199790835380554,
|
| 821 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 822 |
+
"rewards/chosen": -0.019276214763522148,
|
| 823 |
+
"rewards/margins": 0.056569743901491165,
|
| 824 |
+
"rewards/rejected": -0.07584595680236816,
|
| 825 |
+
"step": 255
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 3.3293745051464767,
|
| 829 |
+
"grad_norm": 22.875,
|
| 830 |
+
"learning_rate": 4.135668656967434e-06,
|
| 831 |
+
"log_odds_chosen": 2.1139309406280518,
|
| 832 |
+
"log_odds_ratio": -0.18317027390003204,
|
| 833 |
+
"logps/chosen": -0.42459002137184143,
|
| 834 |
+
"logps/rejected": -1.4972463846206665,
|
| 835 |
+
"loss": 10.2835,
|
| 836 |
+
"nll_loss": 0.6335573792457581,
|
| 837 |
+
"rewards/accuracies": 0.965624988079071,
|
| 838 |
+
"rewards/chosen": -0.02122950181365013,
|
| 839 |
+
"rewards/margins": 0.053632818162441254,
|
| 840 |
+
"rewards/rejected": -0.07486231625080109,
|
| 841 |
+
"step": 260
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 3.392715756136184,
|
| 845 |
+
"grad_norm": 26.5,
|
| 846 |
+
"learning_rate": 3.3962661064086354e-06,
|
| 847 |
+
"log_odds_chosen": 2.135136842727661,
|
| 848 |
+
"log_odds_ratio": -0.18204157054424286,
|
| 849 |
+
"logps/chosen": -0.40990591049194336,
|
| 850 |
+
"logps/rejected": -1.4916514158248901,
|
| 851 |
+
"loss": 10.287,
|
| 852 |
+
"nll_loss": 0.6338369250297546,
|
| 853 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 854 |
+
"rewards/chosen": -0.020495299249887466,
|
| 855 |
+
"rewards/margins": 0.05408727377653122,
|
| 856 |
+
"rewards/rejected": -0.07458256930112839,
|
| 857 |
+
"step": 265
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 3.456057007125891,
|
| 861 |
+
"grad_norm": 22.625,
|
| 862 |
+
"learning_rate": 2.7248368952908053e-06,
|
| 863 |
+
"log_odds_chosen": 2.2748658657073975,
|
| 864 |
+
"log_odds_ratio": -0.17176181077957153,
|
| 865 |
+
"logps/chosen": -0.41875600814819336,
|
| 866 |
+
"logps/rejected": -1.62033212184906,
|
| 867 |
+
"loss": 10.5217,
|
| 868 |
+
"nll_loss": 0.6490176916122437,
|
| 869 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 870 |
+
"rewards/chosen": -0.020937800407409668,
|
| 871 |
+
"rewards/margins": 0.060078807175159454,
|
| 872 |
+
"rewards/rejected": -0.08101661503314972,
|
| 873 |
+
"step": 270
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 3.519398258115598,
|
| 877 |
+
"grad_norm": 21.25,
|
| 878 |
+
"learning_rate": 2.1234935883170047e-06,
|
| 879 |
+
"log_odds_chosen": 2.311600685119629,
|
| 880 |
+
"log_odds_ratio": -0.17423459887504578,
|
| 881 |
+
"logps/chosen": -0.3958762288093567,
|
| 882 |
+
"logps/rejected": -1.5846991539001465,
|
| 883 |
+
"loss": 9.5479,
|
| 884 |
+
"nll_loss": 0.5880299806594849,
|
| 885 |
+
"rewards/accuracies": 0.96875,
|
| 886 |
+
"rewards/chosen": -0.019793810322880745,
|
| 887 |
+
"rewards/margins": 0.05944114923477173,
|
| 888 |
+
"rewards/rejected": -0.07923495769500732,
|
| 889 |
+
"step": 275
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 3.5827395091053047,
|
| 893 |
+
"grad_norm": 25.875,
|
| 894 |
+
"learning_rate": 1.59412823400657e-06,
|
| 895 |
+
"log_odds_chosen": 2.3392279148101807,
|
| 896 |
+
"log_odds_ratio": -0.17188401520252228,
|
| 897 |
+
"logps/chosen": -0.40492624044418335,
|
| 898 |
+
"logps/rejected": -1.6255334615707397,
|
| 899 |
+
"loss": 10.1858,
|
| 900 |
+
"nll_loss": 0.628017783164978,
|
| 901 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 902 |
+
"rewards/chosen": -0.020246315747499466,
|
| 903 |
+
"rewards/margins": 0.061030369251966476,
|
| 904 |
+
"rewards/rejected": -0.08127669245004654,
|
| 905 |
+
"step": 280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 3.646080760095012,
|
| 909 |
+
"grad_norm": 26.625,
|
| 910 |
+
"learning_rate": 1.138406411610482e-06,
|
| 911 |
+
"log_odds_chosen": 2.3245720863342285,
|
| 912 |
+
"log_odds_ratio": -0.16916075348854065,
|
| 913 |
+
"logps/chosen": -0.37623462080955505,
|
| 914 |
+
"logps/rejected": -1.5502853393554688,
|
| 915 |
+
"loss": 9.6503,
|
| 916 |
+
"nll_loss": 0.5946874618530273,
|
| 917 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 918 |
+
"rewards/chosen": -0.018811732530593872,
|
| 919 |
+
"rewards/margins": 0.058702535927295685,
|
| 920 |
+
"rewards/rejected": -0.07751427590847015,
|
| 921 |
+
"step": 285
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 3.709422011084719,
|
| 925 |
+
"grad_norm": 21.875,
|
| 926 |
+
"learning_rate": 7.577619905828282e-07,
|
| 927 |
+
"log_odds_chosen": 2.351839065551758,
|
| 928 |
+
"log_odds_ratio": -0.16314415633678436,
|
| 929 |
+
"logps/chosen": -0.3871734142303467,
|
| 930 |
+
"logps/rejected": -1.6283880472183228,
|
| 931 |
+
"loss": 9.8769,
|
| 932 |
+
"nll_loss": 0.6091512441635132,
|
| 933 |
+
"rewards/accuracies": 0.971875011920929,
|
| 934 |
+
"rewards/chosen": -0.019358674064278603,
|
| 935 |
+
"rewards/margins": 0.06206073611974716,
|
| 936 |
+
"rewards/rejected": -0.08141940087080002,
|
| 937 |
+
"step": 290
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 3.772763262074426,
|
| 941 |
+
"grad_norm": 23.0,
|
| 942 |
+
"learning_rate": 4.5339261909704e-07,
|
| 943 |
+
"log_odds_chosen": 2.405986785888672,
|
| 944 |
+
"log_odds_ratio": -0.16327905654907227,
|
| 945 |
+
"logps/chosen": -0.4028860628604889,
|
| 946 |
+
"logps/rejected": -1.711242914199829,
|
| 947 |
+
"loss": 10.0405,
|
| 948 |
+
"nll_loss": 0.6193649172782898,
|
| 949 |
+
"rewards/accuracies": 0.987500011920929,
|
| 950 |
+
"rewards/chosen": -0.020144306123256683,
|
| 951 |
+
"rewards/margins": 0.06541784107685089,
|
| 952 |
+
"rewards/rejected": -0.08556214720010757,
|
| 953 |
+
"step": 295
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 3.836104513064133,
|
| 957 |
+
"grad_norm": 23.375,
|
| 958 |
+
"learning_rate": 2.262559558016325e-07,
|
| 959 |
+
"log_odds_chosen": 2.3016839027404785,
|
| 960 |
+
"log_odds_ratio": -0.16708031296730042,
|
| 961 |
+
"logps/chosen": -0.38780677318573,
|
| 962 |
+
"logps/rejected": -1.5517771244049072,
|
| 963 |
+
"loss": 9.6048,
|
| 964 |
+
"nll_loss": 0.5919488072395325,
|
| 965 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 966 |
+
"rewards/chosen": -0.01939033903181553,
|
| 967 |
+
"rewards/margins": 0.058198511600494385,
|
| 968 |
+
"rewards/rejected": -0.07758884876966476,
|
| 969 |
+
"step": 300
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 3.89944576405384,
|
| 973 |
+
"grad_norm": 24.75,
|
| 974 |
+
"learning_rate": 7.706665667180091e-08,
|
| 975 |
+
"log_odds_chosen": 2.3498542308807373,
|
| 976 |
+
"log_odds_ratio": -0.1596686691045761,
|
| 977 |
+
"logps/chosen": -0.39143842458724976,
|
| 978 |
+
"logps/rejected": -1.6413482427597046,
|
| 979 |
+
"loss": 10.6228,
|
| 980 |
+
"nll_loss": 0.6559435725212097,
|
| 981 |
+
"rewards/accuracies": 0.981249988079071,
|
| 982 |
+
"rewards/chosen": -0.019571922719478607,
|
| 983 |
+
"rewards/margins": 0.06249549239873886,
|
| 984 |
+
"rewards/rejected": -0.08206741511821747,
|
| 985 |
+
"step": 305
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 3.9627870150435474,
|
| 989 |
+
"grad_norm": 23.625,
|
| 990 |
+
"learning_rate": 6.294126437336734e-09,
|
| 991 |
+
"log_odds_chosen": 2.3062288761138916,
|
| 992 |
+
"log_odds_ratio": -0.17084279656410217,
|
| 993 |
+
"logps/chosen": -0.3852413594722748,
|
| 994 |
+
"logps/rejected": -1.5706732273101807,
|
| 995 |
+
"loss": 10.5082,
|
| 996 |
+
"nll_loss": 0.6482213139533997,
|
| 997 |
+
"rewards/accuracies": 0.96875,
|
| 998 |
+
"rewards/chosen": -0.019262069836258888,
|
| 999 |
+
"rewards/margins": 0.059271592646837234,
|
| 1000 |
+
"rewards/rejected": -0.07853366434574127,
|
| 1001 |
+
"step": 310
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 3.9881235154394297,
|
| 1005 |
+
"step": 312,
|
| 1006 |
+
"total_flos": 0.0,
|
| 1007 |
+
"train_loss": 25.76104594499637,
|
| 1008 |
+
"train_runtime": 3640.7426,
|
| 1009 |
+
"train_samples_per_second": 5.548,
|
| 1010 |
+
"train_steps_per_second": 0.086
|
| 1011 |
+
}
|
| 1012 |
+
],
|
| 1013 |
+
"logging_steps": 5,
|
| 1014 |
+
"max_steps": 312,
|
| 1015 |
+
"num_input_tokens_seen": 0,
|
| 1016 |
+
"num_train_epochs": 4,
|
| 1017 |
+
"save_steps": 100,
|
| 1018 |
+
"stateful_callbacks": {
|
| 1019 |
+
"TrainerControl": {
|
| 1020 |
+
"args": {
|
| 1021 |
+
"should_epoch_stop": false,
|
| 1022 |
+
"should_evaluate": false,
|
| 1023 |
+
"should_log": false,
|
| 1024 |
+
"should_save": true,
|
| 1025 |
+
"should_training_stop": true
|
| 1026 |
+
},
|
| 1027 |
+
"attributes": {}
|
| 1028 |
+
}
|
| 1029 |
+
},
|
| 1030 |
+
"total_flos": 0.0,
|
| 1031 |
+
"train_batch_size": 1,
|
| 1032 |
+
"trial_name": null,
|
| 1033 |
+
"trial_params": null
|
| 1034 |
+
}
|