DeepKE-mcp-tools

This is the standard prediction part of the mcp service for the four natural language processing tasks in DeepKE. Therefore, it is necessary to have a DeepKE model trained for the corresponding task, ensuring that the corresponding predict.py can run.

Download the Code

bash
cd DeepKE
git clone https://github.com/Shotsuke/deepke-mcp-tools.git

Configure .env Environment Variables

• Set up the deepke / deepke-ee virtual environment.
• If the environment has already been set up during model training, this section can be skipped.

bash
cd DeepKE
conda create -n deepke python=3.8 -y
conda activate deepke

pip install pip==24.0.0 # Requires pip<=24.0
pip install -r requirements.txt
pip install -U transformers==4.36.2 # Some libraries specify transformers == 3.4.0, but it actually doesn't work

requirements.txt does not check torch and cuda versions, so you need to manually check them

conda list | grep "torch"

nvidia-smi

Choose the appropriate version to download based on the nvidia compatible cuda from (Start Locally | PyTorch), for example:

pip install torch==2.4.1
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

python setup.py install
python setup.py develop

After setting up deepke, configure deepke-ee, which is essentially the same, but with an additional pip install hydra-core==1.3.1 as the EE task uses a higher version of hydra.

Add the conda directory containing PY and the DeepKE directory to the .env file, for example:

DEEPKE_PATH="~/DeepKE"
CONDA_PY="/home/user_name/anaconda3/envs/deepke/bin/"
CONDA_EE_PY="/home/user_name/anaconda3/envs/deepke-ee/bin/"

• API_KEY

The project uses Alibaba's Qwen large model. You can change DASHSCOPE_API_KEY accordingly.

Configure the MCP Project UV Environment

bash

curl -LsSf https://astral.sh/uv/install.sh | sh # Install uv

pipx install uv # Choose one method to install uv

pip install uv

Now in: DeepKE/

cd deepke-mcp-tools
uv venv
source .venv/bin/activate
uv add "mcp[cli]" httpx openai pyyaml

Run

bash

Now in: DeepKE/deepke-mcp-tools/

python run.py