How to run

[xxTVOJAMAMAxx]

Can someone explain how to install and run this with a bit more detail?

[linyq17]

Thanks for your interest in our work!

Here is the simplest way to install and run Kiwi-Edit.

1) If you just want to try it (no setup)

Use the online demo: https://huggingface.co/spaces/linyq/KiwiEdit

2) Run locally (easiest path: Diffusers inference)

Requirements: Linux + NVIDIA GPU (CUDA 12.8), Python 3.10, Conda.

git clone https://github.com/showlab/Kiwi-Edit.git
cd Kiwi-Edit

conda create -n diffusers python=3.10 -y
conda activate diffusers

pip install torch==2.7.0 torchvision==0.22.0 torchaudio==2.7.0 --index-url https://download.pytorch.org/whl/cu128
pip install diffusers einops accelerate transformers==4.57.0 opencv-python av

Run the demo command:

python diffusers_demo.py \
  --video_path ./demo_data/video/source/0005e4ad9f49814db1d3f2296b911abf.mp4 \
  --prompt "Remove the monkey." \
  --save_path ./output.mp4 \
  --model_path linyq/kiwi-edit-5b-instruct-only-diffusers

This will generate output.mp4.

3) About --ckpt_path in demo.py

--ckpt_path should point to a local LoRA checkpoint file (for example a .safetensors file) after you download one of the model checkpoints from the README table.
If helpful, I can add an exact “download + run” example for one specific checkpoint in the README.