You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thank you for your outstanding work! Without a doubt, your recently published VILA v1.5 series pushes the boundaries of multimodal large language models. It is arguably the most powerful and user-friendly MLLM available today.
Many of us who are interested in the VILA v1.5 series are curious about the development of its video intelligence capabilities. A detailed technical report on this aspect would be incredibly beneficial. Could you let us know if the VILA development team is planning to release such a report?
Thank you!
The text was updated successfully, but these errors were encountered:
@Lyken17 Thank you for your amazing work! May I ask where to view the technical report or blog? Or, specifically, what LLMs are vila-1.5 3b, 7b and 13b using? I noticed there seems to be some changes? Thank you for your attention!
Hi,
Thank you for your outstanding work! Without a doubt, your recently published VILA v1.5 series pushes the boundaries of multimodal large language models. It is arguably the most powerful and user-friendly MLLM available today.
Many of us who are interested in the VILA v1.5 series are curious about the development of its video intelligence capabilities. A detailed technical report on this aspect would be incredibly beneficial. Could you let us know if the VILA development team is planning to release such a report?
Thank you!
The text was updated successfully, but these errors were encountered: