Setting Up the Development Environment on InternLM Studio

Begin by launching a develompent instance and configuring the runtime environment. Once inside, replicate the base environment for running InternLM by creating a custom conda environment named InternLM2_Huixiangdou:

studio-conda -o internlm-base -t InternLM2_Huixiangdou

After cloning the environment, verify its presence locally. Then activate the newly created Python virtual environment using Conda:

conda activate InternLM2_Huixiangdou

Next, prepare the required model files for the assistant. Create a dedicated directory and link the necessary models:

# Create models directory
cd /root && mkdir models

# Link BCE embedding model
ln -s /root/share/new_models/maidalun1020/bce-embedding-base_v1 /root/models/bce-embedding-base_v1

# Link BCE reranker model
ln -s /root/share/new_models/maidalun1020/bce-reranker-base_v1 /root/models/bce-reranker-base_v1

# Select one large language model based on your task (example below)
ln -s /root/share/new_models/Shanghai_AI_Laboratory/internlm2-chat-7b /root/models/internlm2-chat-7b

Install all dependencies needed for the assistant to function properly:

pip install protobuf==4.25.3 accelerate==0.28.0 aiohttp==3.9.3 auto-gptq==0.7.1 bcembedding==0.1.3 beautifulsoup4==4.8.2 einops==0.7.0 faiss-gpu==1.7.2 langchain==0.1.14 loguru==0.7.2 lxml_html_clean==0.1.0 openai==1.16.1 openpyxl==3.1.2 pandas==2.2.1 pydantic==2.6.4 pymupdf==1.24.1 python-docx==1.1.0 pytoml==0.1.21 readability-lxml==0.8.1 redis==5.0.3 requests==2.31.0 scikit-learn==1.4.1.post1 sentence_transformers==2.2.2 textract==1.6.5 tiktoken==0.6.0 transformers==4.39.3 transformers_stream_generator==0.0.5 unstructured==0.11.2

Note: Since InternLM Studio does not allow permenent system modifications, it is recommended to skip installing .doc/.docx parsing dependencies unless you plan to process Word documents in later stages. To enable document parsing, uncomment the following line:

# apt update && apt -y install python-dev python libxml2-dev libxslt1-dev antiword unrtf poppler-utils pstotext tesseract-ocr flac ffmpeg lame libmad0 libsox-fmt-mp3 sox libjpeg-dev swig libpulse-dev

Now, fetch the official repository for the assistant:

cd /root
git clone https://github.com/internlm/huixiangdou && cd huixiengdou
git checkout 447c6f7e68a1657fce1c4f7c740ea1700bde0440

Update the configuration file to point to the local model paths. Modify three entries in /root/huixiangdou/config.ini:

Set the embedding model path:

sed -i '6s#.*#embedding_model_path = "/root/models/bce-embedding-base_v1"#' /root/huixiangdou/config.ini

Set the reranking model path:

sed -i '7s#.*#reranker_model_path = "/root/models/bce-reranker-base_v1"#' /root/huixiangdou/config.ini

Set the local LLM path:

sed -i '29s#.*#local_llm_path = "/root/models/internlm2-chat-7b"#' /root/huixiangdou/config.ini

Prepare the knowledge base by cloning relevant source material:

cd /root/huixiangdou && mkdir repodir

git clone https://github.com/internlm/huixiangdou --depth=1 repodir/huixiangdou

Perform a basic test run:

Modify the default query list in the main script to include sample questions:

sed -i '74s/.*/    queries = ["What is Huixiangdou?", "How to deploy Huixiangdou in WeChat", "What's the weather today?"]/' /root/huixiangdou/huixiangdou/main.py

Finally, launch the assistant:

cd /root/huixiangdou
python3 -m huixiangdou.main --standalone

This completes the setup for handling queries such as "How to deploy Huixiangdou in WeChat?".

Integrating with WeChat via Web Interface

Use the web-based version of Huixiangdou for deployment.

Download the precompiled APK from the GitHub Releases page following the provided instructions.

1. Enter the webhook URL provided by the WeChat integration service and confirm.
2. Tap the "Tap Me" button to toggle the "Huixiangdou LLM RAG Response" feature on or off.
3. Keep the application running. Navigate to a WeChat group chat and send a message to experience real-time responses.