Hugging Face's logo

InstantX
/
FLUX.1-dev-Controlnet-Union

Diffusers
Safetensors
Text-to-Image
ControlNet
Diffusers
Stable Diffusion
Model card Files
xet
 Community
37

Great model! I did a Gradio space based on it and found a couple of issues

#15
by jiuface - opened
Discussion
jiuface

  1. When the resolution is greater than 768, there is a high chance of encountering an Out Of Memory (OOM) error:

    torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 576.00 MiB (GPU 0; 39.25 GiB total capacity; 38.15 GiB already allocated; 199.88 MiB free; 38.78 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

  2. The training image set seems small? In many cases, the results are even less impressive compared to SDXL.

  3. The token limit restricts the creative potential of Flux:

    Token indices sequence length is longer than the specified maximum sequence length for this model (270 > 77). Running this sequence through the model will result in indexing errors. The following part of your input was truncated because CLIP can only handle sequences up to 77 tokens:

Finally, here's the link to my space. I'm eagerly looking forward to future updates:
https://huggingface.co/spaces/jiuface/FLUX.1-dev-Controlnet-Union

Sign up or log in to comment