Step 1. Create a conda environment and activate it.
conda create -n intern_clean python=3.9 -y
conda activate intern_clean
Step 2. Install PyTorch (We use PyTorch 2.0.1 / CUDA 11.7)
pip3 install torch torchvision torchaudio
# Please use the following command to install PyTorch so you can replicate our results:
# pip install torch==2.0.1+cu117 torchvision==0.15.2+cu117 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu117
Step 3. Install require packages
pip install transformers==4.33.2 timm==0.4.12 sentencepiece==0.1.99 gradio==4.13.0 markdown2==2.4.10 xlsxwriter==3.1.2 einops
pip install auto_gptq
Fine-turning requires deepspeed, peft (optional for LoRA fine-tuning)
# install deepspeed
pip install deepspeed
# install peft
pip install peft