search

sail-sg / MDT

Masked Diffusion Transformer is the SOTA for image synthesis. (ICCV 2023)
Apache License 2.0
500 stars 35 forks source link

What is model MDT_S_2? #9

Closed KevinWang676 closed 6 months ago

KevinWang676 commented 1 year ago

Hi, I wonder what --model MDT_S_2 is in the run.sh file. Is it just a random model name? Thank you.

KevinWang676 commented 1 year ago

And when I tried to train the model on CIFAR-10 dataset, I got an error called TypeError: __init__() got an unexpected keyword argument 'approximate'. Could you help me fix this issue? Thank you! The error is shown below.

image

gasvn commented 1 year ago

Hi, I wonder what --model MDT_S_2 is in the run.sh file. Is it just a random model name? Thank you.

It's the S size model with 2 patch size. We follow the DiT to name the model.

gasvn commented 1 year ago

And when I tried to train the model on CIFAR-10 dataset, I got an error called TypeError: __init__() got an unexpected keyword argument 'approximate'. Could you help me fix this issue? Thank you! The error is shown below.

image

You can try to update your pytorch version.

KevinWang676 commented 1 year ago

Thanks for your reply. I tried to update the pytorch version by running pip install torch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1. And I got another error unfortunately, which is shown below. I wonder how to fix the error called huggingface_hub.utils._errors.LocalEntryNotFoundError. Thanks!

image

KevinWang676 commented 1 year ago

Hi, I tried to fix the problem by running 

set https_proxy=http://username:password@proxy.example.com:8080
set http_proxy=http://username:password@proxy.example.com:8080

and pip install pylxd==2.3.0. But I still got the same error.

gasvn commented 1 year ago

I didn't encounter this problem. Here is a possible solution from other repo. Maybe you can try it https://github.com/huggingface/transformers/issues/23404