search

TonyLianLong / LLM-groundedVideoDiffusion

[ICLR 2024] LLM-grounded Video Diffusion Models (LVD): official implementation for the LVD paper
https://llm-grounded-video-diffusion.github.io/
104 stars 4 forks source link

confusion regarding models in Notebook #2

Open GARV-k opened 3 weeks ago

GARV-k commented 3 weeks ago

https://colab.research.google.com/drive/17He4bFAF8lXmT9Nfv-Sg29iKtPelDUNZ in the colab notebook you provided here, is the baseline model, the model proposed by you ?

TonyLianLong commented 3 weeks ago

If you set lvd_gligen_scheduled_sampling_beta to 0, it is equivalent to not using the GLIGEN adapters trained in IGLIGEN project. This way it falls back to the original baseline model (since the adapters are not turned on).

GARV-k commented 3 weeks ago

Yea but is that baseline model , the base model proposed by you. Or is the baseline model in the model some other model. I am really sorry for this confusion but this detail is important for my project.

TonyLianLong commented 2 weeks ago

The base model is modelscope for square videos and zeroscope for horizontal videos.