Search results

Moving to CUDA after loading is extremely slow

enhancementNew feature or requestNew feature or request

performanceAnything related to performance improvements, profiling and benchmarkingAnything related to performance improvements, profiling and benchmarking

Status: Open.

#12599 In huggingface/diffusers;

· francois-rozet opened on Nov 5, 2025

[feature] implement TeaCache

contributions-welcome

performanceAnything related to performance improvements, profiling and benchmarkingAnything related to performance improvements, profiling and benchmarking

Status: Open.

#12589 In huggingface/diffusers;

· sayakpaul opened on Nov 4, 2025

What kernels should we integrate in Diffusers?

performanceAnything related to performance improvements, profiling and benchmarkingAnything related to performance improvements, profiling and benchmarking

Status: Open.

#12375 In huggingface/diffusers;

· sayakpaul opened on Sep 23, 2025

inference with lora suddenly significantly slower

bugSomething isn't workingSomething isn't working

performanceAnything related to performance improvements, profiling and benchmarkingAnything related to performance improvements, profiling and benchmarking

Status: Open.

#12131 In huggingface/diffusers;

· a-r-r-o-w opened on Aug 12, 2025

[tests] help us test <code>torch.compile()</code> for impactful models

Good second issue

performanceAnything related to performance improvements, profiling and benchmarkingAnything related to performance improvements, profiling and benchmarking

torch.compile

Status: Open.

#11430 In huggingface/diffusers;

· sayakpaul opened on Apr 28, 2025

Can we allow making everything on gpu/cuda for scheduler?

performanceAnything related to performance improvements, profiling and benchmarkingAnything related to performance improvements, profiling and benchmarking

scheduler

staleIssues that haven't received updatesIssues that haven't received updates

Status: Open.

#9485 In huggingface/diffusers;

· xiang9156 opened on Sep 20, 2024

Suggestion for speeding up <code>index_for_timestep</code> by removing sequential <code>nonzero()</code> calls in samplers

contributions-welcome

help wantedExtra attention is neededExtra attention is needed

performanceAnything related to performance improvements, profiling and benchmarkingAnything related to performance improvements, profiling and benchmarking

wip

Status: Open.

#9417 In huggingface/diffusers;

· ethanweber opened on Sep 11, 2024