zhangyiss/mlx-examples
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx-examples.git synced 2025-08-31 11:54:37 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

nits in README

Browse Source
This commit is contained in:
Awni Hannun 2023-12-28 21:40:18 -08:00
parent 9ff0a96ab0
commit 7d923b41f1
1 changed files with 9 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
llms/deepseek-coder/README.md
View File

@ -1,7 +1,7 @@
# Deepseek Coder
Deepseek Coder is a family of code generating language models based on the
LLama architecture.[^1] The models were trained from scratch on a corpus of 2T
Llama architecture.[^1] The models were trained from scratch on a corpus of 2T
tokens, with a composition of 87% code and 13% natural language containing both
English and Chinese.
@ -16,7 +16,7 @@ pip install -r requirements.txt
Next, download and convert the model.
```sh
python convert.py --hf-path <path_to_huggingface_model> --mlx-path <path_to_save_converted_model>
python convert.py --hf-path <path_to_huggingface_model>
```
To generate a 4-bit quantized model, use `-q`. For a full list of options run:
@ -26,20 +26,19 @@ python convert.py --help
```
The converter downloads the model from Hugging Face. The default model is
`deepseek-ai/deepseek-coder-6.7b-instruct`. Check out the Hugging Face
page[^1] to see a list of available models.
`deepseek-ai/deepseek-coder-6.7b-instruct`. Check out the [Hugging Face
page]((https://huggingface.co/deepseek-ai) to see a list of available models.
By default, the conversion script will save the converted `weights.npz`,
`tokenizer`, and `config.json` in the path provided by `--mlx-path`.
tokenizer, and `config.json` in the `mlx_model` directory.
### Run
Once you've converted the weights to MLX format, you can interact with the
Deepseek coder model:
Once you've converted the weights, you can interact with the Deepseek coder
model:
```
python deepseek-coder.py --model-path <path_to_save_converted_model> --prompt "write a quick sort algorithm in python."
python deepseek_coder.py --prompt "write a quick sort algorithm in python."
```
[^1] For more information see the [Hugging Face page](https://huggingface.co/deepseek-ai).
[^1]: For more information [blog post](https://deepseekcoder.github.io/) by DeepSeek AI