CoinCheung

We know that for qkv attention, the result of `q @ k` should be divided by `sqrt(d)`, will this also be same for efficientVit? Does relu-based-linear-attention need layernorm or position...

Hi, I noticed that the `AutoencoderKL` has a method of `encode()`, which requires a input tensor of the image. For flux, should this input image tensor be normalized? If so,...

question about training recipe

1

Hi, Thanks for bring the work to public !! I have a question about experiments in Table 5. In the paper, it is claimed that the training method of repVIT...

### Description Here is the results of live_grep, I do not want it to show the line content in the "Results" part of the packer. Is there an option for...

NVJPEG error "2" : nvJPEG error (2): Wrong parameter was passed. For example, a null pointer as input data, or an image index not in the allowed range.

2

### Version 1.51.2 ### Describe the bug. Here is the image:  Here is code to reproduce: ```python import numpy as np from nvidia.dali import pipeline_def import nvidia.dali.fn as fn...

Larger model and large dataset trained with method in this codebase. Is there any such pretrained models ?

Current status is like this:  `float32_t`/`float64_t` is not supported. Would you please add this? BTW, I have submitted an issue in the repo of `vim`, and they let me...