This is the pytorch implementation of our paper:
Generative Recommendation: Towards Next-generation Recommender Paradigm
- Anaconda 3
- python 3.8.15
- pytorch 1.7.0
- numpy 1.23.5
The codes for AI Editor are in './code/AI_Editor/' folder, including three tasks 1) thumbnail selection & thumbnail generation, 2) micro-video clipping, and 3) micro-video content editing.
The personalized selected thumbnails can be obtained by running
cd code/AI_Editor
python select_thumbnail.py
The personalized generated thumbnails can be obtained by running
cd code/AI_Editor/Thumbnail_gen
python scripts/generate_thumbnail.py
The personalized clipped micro-videos can be obtained by running
cd code/AI_Editor
python clip_microVid.py
This task requires training, which can be achieved by running
cd code/AI_Editor/Content_editing/scripts
sh train.sh
At inference stage, the personalized edited micro-videos can be obtained by running
cd code/AI_Editor/Content_editing/scripts
sh edit.sh
The codes for AI Creator are in './code/AI_Creator/' folder, including the task of micro-video content creation.
This task also requires training, which can be achieved by running
cd code/AI_Creator/Content_creation/scripts
sh train.sh
At inference stage, the personalized created micro-videos can be obtained by running
cd code/AI_Creator/Content_creation/scripts
sh gen.sh
Please feel free to send a request email for the data. The use of the data should be for research purpose only.