EXIT

Hi Seyed, I used pretrain legnth estimator from [T2M](https://github.com/EricGuo5513/text-to-motion?tab=readme-ov-file#train-text2length-model) (the proposed model from HumanML3D dataset paper). Please see "Train text2length model:" section. Also, you can see from our [unclean version...

Thank you for your interest in our work. The tensor global_joint always has the shape of [batch, 196, 22, 3]. If you want to generate a shorter sequence, you can...

First, let me clearify that HumanML3D represents relative joint positions, so directly concatenating multiple motion clips can lead to inconsistent global positions. And here are the answers to your questions:...

I didn’t clean up the code for editing generation, but the code is already in the evaluation function. Here are some information: 1. Change [edit_task](https://github.com/exitudio/BAMM/blob/main/utils/eval_t2m.py#L937) to one of these: "inpainting,...

Thank you for reporting this issue. The bug is from the length prediction. I fixed it already.

Thank you for your interest in my papers. In both MMM and BAMM, the text token is prepended to the beginning of the input sequence. For BAMM, which is an...

I didn't clean the VQVQAE training code. Not sure if it has any errors. But the training for VQVAE is the same as MoMask. You can use their script.

I calculated the cross-entropy loss averaged per sample, instead of per token as in MoMask. I implemented the loss function [here](https://github.com/exitudio/BAMM/blob/main/models/mask_transformer/transformer.py#L362-L371) Then, I commented out the [loss](https://github.com/exitudio/BAMM/blob/main/models/mask_transformer/tools.py#L133) in that function.

Thank you for your interest. We plan to release the code next month.

Thank you for your interest in our work. I did not clean the code for training yet. But I already put the raw code in the repo. Here is the...