@Bao Zhuhan, @Wu Jiabei
I have successfully reproduced the Quilt1m project by following the original Repo and correcting some code errors. The demo image-text pair data generated is located at:
-
trans.py: Converts .webm, .mkv, or .webp files to .mp4 format using moviepy.
-
autorename.py: Recursively traverses directories to rename video files and invokes trans.py on non-mp4 files.
-
Inference_extraction.py: Analyzes text files and extracts causal or inferential sentences using OpenAI's API.