JLNTV_IPD/CogVideo
4
0
Fork 0
mirror of https://github.com/THUDM/CogVideo.git synced 2026-05-13 11:58:56 +08:00
Code Issues Packages Projects Releases Wiki Activity
CogVideo/inference/gradio_composite_demo
History
Octopus 3fcb95bbeb feat: add MiniMax provider support for prompt enhancement 
- Add _get_llm_client() helper to auto-select LLM provider from env vars
- Support MINIMAX_API_KEY to use MiniMax (MiniMax-M2.7) as provider
- MINIMAX_API_KEY takes priority over OPENAI_API_KEY when both are set
- Default base URL: https://api.minimax.io/v1 (overridable via MINIMAX_BASE_URL)
- Update all three demo scripts: convert_demo.py, gradio_web_demo.py, gradio_composite_demo/app.py
- Add unit tests covering provider selection and fallback behaviour
2026-04-06 22:44:32 +08:00
..
example_images
llm-flux-cogvideox-i2v-tools
2024-09-16 23:01:37 +08:00
example_videos
horse
2024-09-28 16:36:40 +08:00
rife
format
2025-03-22 15:14:06 +08:00
app.py
feat: add MiniMax provider support for prompt enhancement
2026-04-06 22:44:32 +08:00
README.md
format
2025-03-22 15:14:06 +08:00
requirements.txt
format
2025-03-22 15:14:06 +08:00
rife_model.py
format
2025-03-22 15:14:06 +08:00
utils.py
format
2025-03-22 15:14:06 +08:00

README.md

title emoji colorFrom colorTo sdk sdk_version suggested_hardware suggested_storage app_port app_file models tags short_description disable_embedding
CogVideoX-5B 🎥 yellow blue gradio 4.42.0 a10g-large large 7860 app.py
THUDM/CogVideoX-5b
cogvideox
video-generation
thudm
Text-to-Video false

Gradio Composite Demo

This Gradio demo integrates the CogVideoX-5B model, allowing you to perform video inference directly in your browser. It supports features like UpScale, RIFE, and other functionalities.

Environment Setup

Set the following environment variables in your system:

  • OPENAI_API_KEY = your_api_key
  • OPENAI_BASE_URL= your_base_url
  • GRADIO_TEMP_DIR= gradio_tmp

Installation

pip install -r requirements.txt

Running the code

python app.py