Mael-zys / T2M-GPT

(CVPR 2023) Pytorch implementation of “T2M-GPT: Generating Human Motion from Textual Descriptions with Discrete Representations”

https://mael-zys.github.io/T2M-GPT/

Mael-zys/T2M-GPT Issues

About text_mot_match
Updated 13 days ago
Can not reproduce the reported results.
Closed 3 months ago3
Unable to reproduce the results using the pretrained weights
Closed a year ago6
A long time during evaluation
Updated 2 months ago
Motion in-betweening
Closed 2 months ago1
VQVAE -- translation initialization of training sequences
Closed 2 months ago2
Why is the length of the generated action sequences always different?
Closed 2 months ago3
One question about VQ-VAE
Closed 3 months ago1
Can you provide the pretrained model for HumanML3D?
Closed 3 months ago2
Dataset difference may exist between your work and official HumanML3D
Updated 4 months ago1
The issue on the evaluation of VQVAE
Closed 5 months ago2
ValueError: Imaginary component
Updated 5 months ago1
2D motion prediction and keypoint format
Closed 6 months ago1
Error in running demo code
Closed 8 months ago1
Use Hugging Face
Closed 8 months ago
How to calculate the trans between each frame?
Closed 8 months ago1
Constraining starting and ending pose
Closed 8 months ago1
where is training code for models used in evaluations?
Closed 8 months ago1
render smpl gif not the same as joints
Closed a year ago4
KIT VQ-VAE Training details
Closed 8 months ago1
How many motion indexes do you have?
Closed 9 months ago5
small question about the vq-vae paper
Updated 9 months ago
Sequences duration
Closed a year ago5
Maximum length of the code index sequence
Closed a year ago4
Render Multiple Frames in Single Image as Presented in the Paper
Updated 9 months ago
Generate SMPL Mesh for rotation data in colab
Updated 9 months ago
A Collab result inconsistent with the paper
Closed a year ago2
About the model's height and body shape
Updated 9 months ago
Difference between your work and TM2T.
Closed 10 months ago1
T2M-GPT generated the Fatal Error joints points
Updated 10 months ago3
question about vq-vae reconstruction loss
Updated 10 months ago
Number of motions computation
Closed 10 months ago1
Can this really run with just 32GB vram?
Updated a year ago
Controlling seed
Closed a year ago2
export onnx model
Updated a year ago1
[Feature] How to use `QuantizeEMAReset` with DDP elegantly and efficiently?
Updated a year ago1
Why the first stage dataset is only 64 frames?
Closed a year ago2
Environment file
Closed a year ago2
question about motion representations
Closed a year ago
Generate SMPL mesh
Updated a year ago
Getting issue while evaluating model
Closed a year ago3
Can you provide the pretrained model for KIT?
Closed a year ago1
How were the meta data for the mean and standard deviation generated ?
Closed a year ago1
The result of real motion
Closed a year ago1
training from scratch
Closed a year ago2
There is a mistake in `dataset/prepare/download_smpl.sh`
Closed a year ago4
95% confidence interval
Closed a year ago5
Question about Table 1comparing various state-of-the-art methods on the t2m dataset in the paper
Closed a year ago4
[T2M-GPT] A loss question about VQVAE
Closed a year ago2
continuous text-to motion generation
Closed a year ago3