Smart Spectrum AI today announced that it willvs."Qingying"homologousofVideo Generation Model ——CogVideoX Open Source.
The CogVideoX open source model is described as containing several models of different sizes and dimensions.Currently open-sourcing CogVideoX-2B.It requires 18GB of video memory for inference at FP-16 precision and 40GB for fine-tuning, which means thatReasoning with a single 4090 graphics cardbut (not)Fine-tuning with a single A6000 graphics card.
CogVideoX-2B has a cue word limit of 226 tokens.Video length is 6 secondsThe frame rate is 8 frames per second and the video resolution is 720*480.
Officials said,Models with higher performance and higher number of parameters are on the way!Please stay tuned and look forward to it.
Attached related links:
-
Code Repository:https://github.com/THUDM/CogVideo
-
Model Download:https://huggingface.co/THUDM/CogVideoX-2b
-
Technical report:https://github.com/THUDM/CogVideo/blob/main/resources/CogVideoX.pdf