search

EricGuo5513 / text-to-motion

Official implementation for "Generating Diverse and Natural 3D Human Motions from Texts (CVPR2022)."
MIT License
422 stars 39 forks source link

Why don't use CLIP text encoder as the text encoder for text-motion evaluator? #43

Open luomingshuang opened 1 month ago

luomingshuang commented 1 month ago

Hi, @EricGuo5513 , about the text-motion evaluator, I want to know why you don't use the CLIP text encoder as the text encoder for text-motion evaluator. I think it is much easier if using the CLIP text encoder directly.

EricGuo5513 commented 1 month ago

Hi, CLIP text encoder could be a better choice now. However, it was not that common to use CLIP at that time.

On Thu, 30 May 2024 at 23:01, Mingshuang Luo @.***> wrote:

Hi, @EricGuo5513 https://github.com/EricGuo5513 , about the text-motion evaluator, I want to know why you don't use the CLIP text encoder as the text encoder for text-motion evaluator. I think it is much easier if using the CLIP text encoder directly.

— Reply to this email directly, view it on GitHub https://github.com/EricGuo5513/text-to-motion/issues/43, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKRYNB357IEZHIX7UV3HNTLZE7R2BAVCNFSM6AAAAABISACADWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGMZDMOBXGEZDCMQ . You are receiving this because you were mentioned.Message ID: @.***>

luomingshuang commented 1 month ago

Oh, thanks for your reply. I think if we change to use CLIP text encoder, maybe we will have a new comparison between different methods.