About code

[JYC1012]

Hello, I am very interested in your work. Do you have any plans to open source training code?

[exitudio]

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 command:

conda activate momask
python train_t2m_transformer.py \
    --name training_name \
    --gpu_id 0 \
    --dataset_name t2m \
    --milestones 50000 80000 \
    --batch_size 512 \
    --max_epoch 2000 \
    --vq_name 2024-05-13-20-41-19_100_8192x32_reset5 \
    --trans official

I didn't test it so there may be some issues. Note that Stage 1 (VQ-VAE) and the Refinement stage (Residual) can be used from the MoMask pre-trained model.