This is an example project on how to use the auto-coder.

We will show you how to use the auto-coder to create a simple Chatbot based on Python + ReactJs + TypeScript + TailwindCSS.

Install Auto-Coder

conda create -n auto-coder python=3.10.11
conda activate auto-coder
pip install auto-coder
ray start --head

Start some LLMs:

byzerllm deploy --pretrained_model_type saas/qianwen \
--cpus_per_worker 0.001 \
--gpus_per_worker 0 \
--num_workers 1 \
--infer_params saas.api_key=xxxxxx  saas.model=qwen-max \
--model qianwen_chat


byzerllm deploy --pretrained_model_type saas/qianwen_vl \
--cpus_per_worker 0.001 \
--gpus_per_worker 0 \
--num_workers 1 \
--infer_params saas.api_key=xxxxx  saas.model=qwen-vl-max \
--model qianwen_vl_chat

Please replace the api_key with your own api_key.

How to reproduce this project

Try to get a copy of this project:

git clone https://github.com/allwefantasy/auto-coder.example.git auto-coder.example.copy
cd auot-coder.example.copy

Start the Chatbot:

Server:

python src/example/server/server.py

cd web
npm install
npm run start

Open the browser and visit http://localhost:3000

You will see a empty project, Now let's start to create a Chatbot.

Create a Chatbot

Run the following command:

auto-coder --file actions/001_add_llm.yml

Then the auto-coder will ask you to input the result(we enable the human_as_model mode in the yml file), Then you need to copy the content of actions/001_add_llm.output (which is generated by Calude Opus), type 'EOF' in a newline and click ENTER to finish the input.

Repeat the above steps for the following files:

1. actions/002_add_page.yml
2. actions/003_remove_fastapi_cors.yml

After all the above steps, you still need to modify the web/src/pages/Chat.tsx file manually:

const response = await fetch('/llm/chat', {

to

const response = await fetch('http://127.0.0.1:9001/llm/chat', {

For now, you can visit http://localhost:3000 to see the Chatbot.