Instructions to use ostris/Flex.2-preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use ostris/Flex.2-preview with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("ostris/Flex.2-preview", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps
- Draw Things
- DiffusionBee
Pipeline shape mismatch
#6
by vladmandic - opened
Looking at the custom diffusers pipeline:num_control_channels = 33
but latent_controls = torch.cat([inpaint_latents, inpaint_latents_mask, control_latents], dim=1)
and if there is no inpaint or control image provided, those are torch.zeros with 3+1+3=7 channels?
i've hacked it by changing channels in torch.zeros for inpaint and control latents from 3 to 16 so it becomes 16+1+16
Please could these changes be merged in pipeline.py
Oh, sorry I missed this, only tested with all controls active. Just pushed a fix. Thank you for bringing it to my attention
closing as fixed...
vladmandic changed discussion status to closed