Skip to content

align_anything.md

Latest commit

 

History

History
61 lines (41 loc) · 1.59 KB

align_anything.md

File metadata and controls

61 lines (41 loc) · 1.59 KB

Align-Anything

:::{Note} Support: MiniCPM-o 2.6 :::

Environment Setup

# clone the repository
git clone git@github.com:PKU-Alignment/align-anything.git
cd align-anything

# create virtual env
conda create -n align-anything python==3.11
conda activate align-anything

On Nvidia GPU

  • [Optional] We recommend installing CUDA in the conda environment and set the environment variable.
# We tested on the H800 computing cluster, and this version of CUDA works well.
# You can adjust this version according to the actual situation of the computing cluster.

conda install nvidia/label/cuda-12.2.0::cuda
export CUDA_HOME=$CONDA_PREFIX

If your CUDA installed in a different location, such as /usr/local/cuda/bin/nvcc, you can set the environment variables as follows:

export CUDA_HOME="/usr/local/cuda"

Finally, install align-anything by:

pip3 install -e .

pip3 install vllm==0.7.2 # to run ppo on vllm engine

Training

You can find SFT & DPO training script in the ./scripts/minicpmo directory. These scripts would automatically download the model and dataset, and run the training or evaluation.

For example, scripts/minicpmo/minicpmo_dpo_vision.sh is the script for Text + Image -> Text modality, you can run it by:

cd scripts
bash minicpmo/minicpmo_dpo_vision.sh

Note: The scripts will automatically download the model and dataset from huggingface. If you are prohibited from the internet, please try to use the HF Mirror:

export HF_ENDPOINT=https://hf-mirror.com