o Gi0@sddlmZddlZddlmZddlmZddlmZm Z m Z ddl m Z dd l mZmZer=ddlmmZd Znd Ze eZd ZGd ddeZdS))CallableN) UNet2DModel)CMStochasticIterativeScheduler)is_torch_xla_availableloggingreplace_example_docstring) randn_tensor)DiffusionPipelineImagePipelineOutputTFa Examples: ```py >>> import torch >>> from diffusers import ConsistencyModelPipeline >>> device = "cuda" >>> # Load the cd_imagenet64_l2 checkpoint. >>> model_id_or_path = "openai/diffusers-cd_imagenet64_l2" >>> pipe = ConsistencyModelPipeline.from_pretrained(model_id_or_path, torch_dtype=torch.float16) >>> pipe.to(device) >>> # Onestep Sampling >>> image = pipe(num_inference_steps=1).images[0] >>> image.save("cd_imagenet64_l2_onestep_sample.png") >>> # Onestep sampling, class-conditional image generation >>> # ImageNet-64 class label 145 corresponds to king penguins >>> image = pipe(num_inference_steps=1, class_labels=145).images[0] >>> image.save("cd_imagenet64_l2_onestep_sample_penguin.png") >>> # Multistep sampling, class-conditional image generation >>> # Timesteps can be explicitly specified; the particular timesteps below are from the original GitHub repo: >>> # https://github.com/openai/consistency_models/blob/main/scripts/launch.sh#L77 >>> image = pipe(num_inference_steps=None, timesteps=[22, 0], class_labels=145).images[0] >>> image.save("cd_imagenet64_l2_multistep_sample_penguin.png") ``` cseZdZdZdZdededdffdd Zd dd Zd!d e j d e fd dZ d ddZ ddZe ee         d"dede j eeBeBdBdedeede jee jBdBde j dBd e dBdedeeee j gdfdBdefddZZS)#ConsistencyModelPipelineab Pipeline for unconditional or class-conditional image generation. This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods implemented for all pipelines (downloading, saving, running on a particular device, etc.). Args: unet ([`UNet2DModel`]): A `UNet2DModel` to denoise the encoded image latents. scheduler ([`SchedulerMixin`]): A scheduler to be used in combination with `unet` to denoise the encoded image latents. Currently only compatible with [`CMStochasticIterativeScheduler`]. unet schedulerreturnNcs"t|j||dd|_dS)N)rr)super__init__register_modulessafety_checker)selfrr __class__v/home/ubuntu/.local/lib/python3.10/site-packages/diffusers/pipelines/consistency_models/pipeline_consistency_models.pyrTs  z!ConsistencyModelPipeline.__init__c Cst||||f} t|trt||krtdt|d|d|dur+t| |||d}n|j||d}||jj}|S)Nz/You have passed a list of generators of length z+, but requested an effective batch size of z@. Make sure the batch size matches the length of the generators.) generatordevicedtype)rr) isinstancelistlen ValueErrorr torinit_noise_sigma) r batch_size num_channelsheightwidthrrrlatentsshaperrrprepare_latents^s   z(ConsistencyModelPipeline.prepare_latentspilsample output_typecCsj|dvr td|d|dddd}|dkr|S|ddd d}|d kr.|S||}|S) N)ptnpr*z output_type=zD is not supported. Make sure to choose one of ['pt', 'np', or 'pil']r g?rr-rr.)r clampcpupermutenumpy numpy_to_pil)rr+r,rrrpostprocess_imageps  z*ConsistencyModelPipeline.postprocess_imagecCs|jjjdurCt|trtj|tjd}n't|tr,|dks"Jdtj|gtjd}n|durr set_timestepsrDr progress_bar enumeratescale_model_inputstepupdate XLA_AVAILABLExm mark_stepr5maybe_free_model_hooksr )rr#r=rCrDrr'r,rIrJrFrErr+rQit scaled_sample model_outputimagerrr__call__sL 6     z!ConsistencyModelPipeline.__call__)N)r*) r/Nr/NNNr*TNr/)__name__ __module__ __qualname____doc__model_cpu_offload_seqrrrr)r9Tensorstrr5r>rHno_gradrEXAMPLE_DOC_STRINGr;r Generatorboolrr_ __classcell__rrrrr CsR     r )typingrr9modelsr schedulersrutilsrrrutils.torch_utilsr pipeline_utilsr r torch_xla.core.xla_modelcore xla_modelrWrV get_loggerr`r@rhr rrrrs