lucidrains/muse-maskgit-pytorch
Implementation of Muse: Text-to-Image Generation via Masked Generative Transformers, in Pytorch
PythonMIT
Issues
- 4
Gumbel Sample
#42 opened by pranoyr - 0
About the VQGAN
#41 opened by duchenzhuang - 1
Why scale used in Attention is 8 (while dim_head is 64)? If dim or dim_head are changed, should scale be changed automatically?
#40 opened by lqniunjunlper - 6
Maskgit transformer is hard to train
#18 opened by zhuqiangLu - 0
A question about the use of parallel training
#39 opened by skyking-317 - 2
T5 bug?
#38 opened by lqniunjunlper - 1
What is minimum VRAM for running demo?
#16 opened by treasuraid - 5
- 0
- 9
Would you like flash attention?
#35 opened by Birch-san - 1
Can you release the official trained model?
#34 opened by JoonHyeokJ - 3
- 18
Trained weights
#20 opened by ArielReplicate - 0
How to train the maskgit transformers (base model)?
#33 opened by easonoob - 0
assert return_loss ^ return_discr_loss, 'you should either return autoencoder loss or discriminator loss, but not both
#28 opened by Qiyuan-Ge - 2
ValueError: You can only pass one `DistributedDataParallelKwargs` in `kwargs_handler`.
#31 opened by Zxlan - 1
beartype error: unhashable type: 'list'
#30 opened by forresti - 0
About Patch Size
#29 opened by tom99763 - 1
- 0
vae enc_dec configuration questions
#24 opened by treasuraid - 2
Add support for logging and visualizing the loss and learning rate with tensorboard.
#23 opened by ZeroCool940711 - 12
[From Hugging Face] Can we help in any way?
#14 opened by patrickvonplaten - 0
AssertionError while trying to train with base_maskgit_trainer() after finished training with vae_trainer()
#22 opened by easonoob - 0
- 4
- 1
T5-small model is too large
#13 opened by bekhzod-olimov - 1
How to resolve this error ?
#12 opened by PHOENIXFURY007 - 3
- 6
Usage questions
#9 opened by SB2020-eye - 2
Negative prompting from the Muse paper?
#7 opened by LinkAnJarad - 4
How to make your example work
#8 opened by ethancohen123 - 2
- 5
This looks extremely similar to Paella (not sure which one is the better approach)
#1 opened by Mut1nyJD - 1
mask schedule
#4 opened by lucidrains - 2
Request for basic documentation
#3 opened by trholding - 1
How to run this?
#2 opened by zhuofengli