EXIT
EXIT
The model should learn to ignore [PAD] tokens (following CLIP). For more information, to get global (sentence) text embedding, CLIP simply applies linear layer to the local (word) embedding. https://github.com/openai/CLIP/blob/main/clip/model.py#L343-L356...
Hi. Thank you for your interest. Are you evaluating on our pretrain model? Does the problem occur every time?
I found a potential problem in this [line](https://github.com/exitudio/MMM/blob/849d4ce18aee85c504bc5d99b4b1759089a9db48/models/t2m_trans.py#L265). I already commit the change. Along with the updated pretrain model. Can you try to 1. Download the new pretrain model (Section...
This issue should come from the randomness from [this line](https://github.com/exitudio/MMM/blob/849d4ce18aee85c504bc5d99b4b1759089a9db48/models/t2m_trans.py#L262-L264). I just commented it out in the previous commit. Can you double check that it is updated?
We use pretrain length estimator from Text-to-motion (the model that proposed along with HumanML3D paper) [here](https://github.com/EricGuo5513/text-to-motion/blob/f8eecd27341b04464e363f0acd993cabef52b1ce/gen_motion_script.py#L134-L138) This can be plugged into our existing model. I will let you know after...
Hi, thank you for your interest. I attached the logs here [VQVAE_HML3D_run.log](https://github.com/user-attachments/files/15776376/VQVAE_HML3D_run.log) [VQVAE_KIT_run.log](https://github.com/user-attachments/files/15776377/VQVAE_KIT_run.log)
Hi, Thanks for pointing that out! Yes, the control shouldn’t go to the main (frozen) network. I actually found this after the submission and tried removing it, but the results...
1. I use .1*xent + .9*joint_loss for **"pelvis only"**. And .5*xent+.9*joint_loss for **"all joints"** 2. I would suggest evaluating only ControlNet to see how it perform. Since Logits Optimization has...
Thank you for your interest in our work. I added the training script. Please see the script [here](https://github.com/exitudio/MaskControl?tab=readme-ov-file#book-training)
Thank you for your interest. I added obstacle avoidance examples. You can use this script to generate. ``` conda activate ControlMM python -m generation.avoidance2 --path_name ./output/avoidance2 --iter_each 100 --iter_last 600...