README.md

metadata

license: apache-2.0

Optimum Habana is the interface between the Hugging Face Transformers and Diffusers libraries and Habana's Gaudi processor (HPU). It provides a set of tools enabling easy and fast model loading, training and inference on single- and multi-HPU settings for different downstream tasks. Learn more about how to take advantage of the power of Habana HPUs to train and deploy Transformers and Diffusers models at hf.co/hardware/habana.

Stable Diffusion HPU configuration

This model only contains the GaudiConfig file for running Stable Diffusion v2 (e.g. stabilityai/stable-diffusion-2-1) on Habana's Gaudi processors (HPU).

This model contains no model weights, only a GaudiConfig.

This enables to specify:

• use_torch_autocast: whether to use Torch Autocast for mixed precision
  • hmp_bf16_ops: list of operators that should run in bf16
  • hmp_fp32_ops: list of operators that should run in fp32

Usage

The GaudiStableDiffusionPipeline (GaudiDDIMScheduler) is instantiated the same way as the StableDiffusionPipeline (DDIMScheduler) in the 🤗 Diffusers library. The only difference is that there are a few new training arguments specific to HPUs.

Here is an example with one prompt:

from optimum.habana import GaudiConfig
from optimum.habana.diffusers import GaudiDDIMScheduler, GaudiStableDiffusionPipeline


model_name = "stabilityai/stable-diffusion-2-1"

scheduler = GaudiDDIMScheduler.from_pretrained(model_name, subfolder="scheduler")

pipeline = GaudiStableDiffusionPipeline.from_pretrained(
    model_name,
    height=768,
    width=768,
    scheduler=scheduler,
    use_habana=True,
    use_hpu_graphs=True,
    gaudi_config="Habana/stable-diffusion-2",
)

outputs = pipeline(
    ["An image of a squirrel in Picasso style"],
    num_images_per_prompt=6,
    batch_size=2,
)

Check out the documentation and this example for more advanced usage.