diffusers

diffusers copied to clipboard

Loading PT Files for VAE with the AutoEncoder is still broken!!!

11

### Describe the bug Just doesn't load pt files anymore. Really frustrating as it's been broken for a long time now. I keep posting about it, so now I'll just...

JemiloII

Error creating pre-trained FluxPipeline with diffusers "Cannot instantiate this tokenizer from a slow version"

14

### Describe the bug I'm receiving an error trying to use the `diffusers` module to run the `flux` model. > ValueError: Cannot instantiate this tokenizer from a slow version. If...

pcgeek86

"previous_timestep()" in DDPM scheduling not compatible with "trailing" option. DDIM bugged too

6

### Describe the bug I'm currently on a quest to remove all noise from my very noise-sensitive diffusion application, so checking all the equations right now and I think I...

RochMollero

create_dynamic_module infinite recursion when setting HF_HOME using relative path

5

### Describe the bug create_dynamic_module infinite recursion when setting HF_HOME using relative path ### Reproduction ```py import os os.environ["HF_HOME"] = ".cache/huggingface" from diffusers.utils.dynamic_modules_utils import DIFFUSERS_DYNAMIC_MODULE_NAME, create_dynamic_module from diffusers.utils.constants import HF_HOME...

luocfprime

[refactor] sincos and rope embeddings in torch instead of numpy

1

# What does this PR do? Internal discussion: https://huggingface.slack.com/archives/C065E480NN9/p1727418894443269 Code ```python import numpy as np import torch from diffusers.models.embeddings import get_1d_sincos_pos_embed_from_grid, aryan_get_1d_sincos_pos_embed_from_grid from diffusers.models.embeddings import get_2d_sincos_pos_embed_from_grid, aryan_get_2d_sincos_pos_embed_from_grid from diffusers.models.embeddings import...

a-r-r-o-w

FLUX.1-dev FP8 Example Code Cleanup

1

I was looking at the [FLUX.1-dev FP8 example code](https://huggingface.co/docs/diffusers/main/en/api/pipelines/flux#single-file-loading-for-the-fluxtransformer2dmodel) in the documentation and noticed some unnecessary imports and variable declarations, which make it appear longer and cluttered. Here is the...

HashingTag

This PR raises asserts when conditioning is not provided to AdaLayerNorm. Specifically: 1) One has to provide either timestep or temb argument to `AdaLayerNorm`. 2) Modulation embedding has to be...

Godofnothing

[Q] Possibly unused `self.final_alpha_cumprod`

6

Hello team, quick question to make sure I understand the behavior of the `step` function in LCM Scheduler. https://github.com/huggingface/diffusers/blob/a7361dccdc581147620bbd74a6d295cd92daf616/src/diffusers/schedulers/scheduling_lcm.py#L534-L543 Here, it seems that the condition `prev_timestep >= 0` is always...

fdtomasi

Flux Controlnet Train Example, will run out of memory on validation step

14

### Describe the bug On default settings provided in flux train example readme, with 10 validation images training will error out with out of memory error during validation. on A100...

Night1099

[Bug fix] "previous_timestep()" in DDPM scheduling compatible with "trailing" and "linspace" options

5

# What does this PR do? Fixes the bug of `previous_timestep()` not giving the right timestep for numbers not a factor of 1000 during "trailing" and "linspace" options. The previous_timestep...

AnandK27