mirror of
https://github.com/ml-explore/mlx-examples.git
synced 2025-08-30 10:56:38 +08:00
30 lines
702 B
Markdown
30 lines
702 B
Markdown
|
# Qwen
|
||
|
|
|
||
|
|
Qwen (通义千问) is a language model proposed by Alibaba Cloud[^1]. The architecture of Qwen is similar to Llama except for the bias in the attention layers.
|
||
|
|
|
||
|
|
## Setup
|
||
|
|
|
||
|
|
Download (from huggingface) and conver the model. By default, the model is `Qwen/Qwen-1_8B`.
|
||
|
|
|
||
|
|
```sh
|
||
|
|
python convert.py
|
||
|
|
```
|
||
|
|
|
||
|
|
This will make the `weights.npz` file which MLX can read.
|
||
|
|
|
||
|
|
## Generate
|
||
|
|
|
||
|
|
To generate text with the default prompt (default tokenizer is `Qwen/Qwen-1_8B`):
|
||
|
|
|
||
|
|
```sh
|
||
|
|
python qwen.py
|
||
|
|
```
|
||
|
|
|
||
|
|
To see a list of options, run:
|
||
|
|
|
||
|
|
```sh
|
||
|
|
python qwen.py --help
|
||
|
|
```
|
||
|
|
|
||
|
|
[^1]: For more details on the model see the official repo of [Qwen](https://github.com/QwenLM/Qwen) and the [Hugging Face](https://huggingface.co/Qwen).
|