|
--- |
|
license: cc-by-nc-4.0 |
|
base_model: mlabonne/NeuralBeagle14-7B |
|
tags: |
|
- merge |
|
- mergekit |
|
- lazymergekit |
|
- dpo |
|
- rlhf |
|
--- |
|
|
|
![](https://i.imgur.com/89ZAKcn.png) |
|
|
|
# NeuralBeagle14-7B |
|
|
|
**Update 01/16/24: NeuralBeagle14-7B is probably the best 7B model you can find. π** |
|
|
|
NeuralBeagle14-7B is a DPO fine-tune of [mlabonne/Beagle14-7B](https://huggingface.co/mlabonne/Beagle14-7B) using the [argilla/distilabel-intel-orca-dpo-pairs](https://huggingface.co/datasets/argilla/distilabel-intel-orca-dpo-pairs) preference dataset and my DPO notebook from [this article](https://towardsdatascience.com/fine-tune-a-mistral-7b-model-with-direct-preference-optimization-708042745aac). |
|
|
|
Thanks [Argilla](https://huggingface.co/argilla) for providing the dataset and the training recipe [here](https://huggingface.co/argilla/distilabeled-Marcoro14-7B-slerp). πͺ |
|
|
|
## π Applications |
|
|
|
This model uses a context window of 8k. It is compatible with different templates, like chatml and Llama's chat template. |
|
|
|
Compared to other 7B models, it displays good performance in instruction following and reasoning tasks. It can also be used for RP and storytelling. |
|
|
|
## π Evaluation |
|
|
|
The evaluation was performed using [LLM AutoEval](https://github.com/mlabonne/llm-autoeval) on Nous suite. It is the best 7B model to date. |
|
|
|
| Model | Average | AGIEval | GPT4All | TruthfulQA | Bigbench | |
|
|---|---:|---:|---:|---:|---:| |
|
| [**mlabonne/NeuralBeagle14-7B**](https://huggingface.co/mlabonne/NeuralBeagle14-7B) [π](https://gist.github.com/mlabonne/ad0c665bbe581c8420136c3b52b3c15c) | **60.25** | **46.06** | **76.77** | **70.32** | **47.86** | |
|
| [mlabonne/Beagle14-7B](https://huggingface.co/mlabonne/Beagle14-7B) [π](https://gist.github.com/mlabonne/f5a5bf8c0827bbec2f05b97cc62d642c) | 59.4 | 44.38 | 76.53 | 69.44 | 47.25 | |
|
| [mlabonne/NeuralDaredevil-7B](https://huggingface.co/mlabonne/NeuralDaredevil-7B) [π](https://gist.github.com/mlabonne/cbeb077d1df71cb81c78f742f19f4155) | 59.39 | 45.23 | 76.2 | 67.61 | 48.52 | |
|
| [argilla/distilabeled-Marcoro14-7B-slerp](https://huggingface.co/argilla/distilabeled-Marcoro14-7B-slerp) [π](https://gist.github.com/mlabonne/9082c4e59f4d3f3543c5eda3f4807040) | 58.93 | 45.38 | 76.48 | 65.68 | 48.18 | |
|
| [mlabonne/NeuralMarcoro14-7B](https://huggingface.co/mlabonne/NeuralMarcoro14-7B) [π](https://gist.github.com/mlabonne/b31572a4711c945a4827e7242cfc4b9d) | 58.4 | 44.59 | 76.17 | 65.94 | 46.9 | |
|
| [openchat/openchat-3.5-0106](https://huggingface.co/openchat/openchat-3.5-0106) [π](https://gist.github.com/mlabonne/1afab87b543b0717ec08722cf086dcc3) | 53.71 | 44.17 | 73.72 | 52.53 | 44.4 | |
|
| [teknium/OpenHermes-2.5-Mistral-7B](https://huggingface.co/teknium/OpenHermes-2.5-Mistral-7B) [π](https://gist.github.com/mlabonne/88b21dd9698ffed75d6163ebdc2f6cc8) | 52.42 | 42.75 | 72.99 | 52.99 | 40.94 | |
|
|
|
You can find the complete benchmark on [YALL - Yet Another LLM Leaderboard](https://huggingface.co/spaces/mlabonne/Yet_Another_LLM_Leaderboard). |
|
|
|
It's also on top of the Open LLM Leaderboard: |
|
|
|
![](https://i.imgur.com/62gUTFn.png) |
|
|
|
Compared to Beagle14, there's no improvement in this benchmark. This might be due to an unlucky run, but I think I might be overexploiting argilla/distilabel-intel-orca-dpo-pairs at this point. Another preference dataset could improve it even further. Note that the Beagle models perform better than Turdus, which is purposely contaminated on Winogrande (very high score). |
|
|
|
## π» Usage |
|
|
|
```python |
|
!pip install -qU transformers accelerate |
|
|
|
from transformers import AutoTokenizer |
|
import transformers |
|
import torch |
|
|
|
model = "mlabonne/NeuralBeagle14-7B" |
|
messages = [{"role": "user", "content": "What is a large language model?"}] |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model) |
|
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) |
|
pipeline = transformers.pipeline( |
|
"text-generation", |
|
model=model, |
|
torch_dtype=torch.float16, |
|
device_map="auto", |
|
) |
|
|
|
outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95) |
|
print(outputs[0]["generated_text"]) |
|
``` |
|
|
|
<p align="center"> |
|
<a href="https://github.com/argilla-io/distilabel"> |
|
<img src="https://raw.githubusercontent.com/argilla-io/distilabel/main/docs/assets/distilabel-badge-light.png" alt="Built with Distilabel" width="200" height="32"/> |
|
</a> |
|
</p> |