From a8ef54954678404251fed5ac40d15ce8c4a57226 Mon Sep 17 00:00:00 2001
From: Juni May <juni_may@outlook.com>
Date: Mon, 18 Dec 2023 15:29:06 +0800
Subject: [PATCH] Add readme and requirements for qwen example

---
 qwen/README.md        | 29 +++++++++++++++++++++++++++++
 qwen/requirements.txt |  4 ++++
 2 files changed, 33 insertions(+)
 create mode 100644 qwen/README.md
 create mode 100644 qwen/requirements.txt

diff --git a/qwen/README.md b/qwen/README.md
new file mode 100644
index 00000000..db1d61e9
--- /dev/null
+++ b/qwen/README.md
@@ -0,0 +1,29 @@
+# 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).
diff --git a/qwen/requirements.txt b/qwen/requirements.txt
new file mode 100644
index 00000000..8a318500
--- /dev/null
+++ b/qwen/requirements.txt
@@ -0,0 +1,4 @@
+mlx
+numpy
+transformers>=4.35
+torch
\ No newline at end of file