From 6c19f0a6bc5b8b2f7501ee1ec381bacb99f4fe84 Mon Sep 17 00:00:00 2001 From: Dhruv Nair Date: Tue, 2 Jan 2024 12:38:43 +0000 Subject: [PATCH] update --- src/diffusers/loaders/single_file_utils.py | 3 +++ .../test_stable_diffusion_xl_img2img.py | 22 +++++++++---------- 2 files changed, 14 insertions(+), 11 deletions(-) diff --git a/src/diffusers/loaders/single_file_utils.py b/src/diffusers/loaders/single_file_utils.py index 4b7f4cafac..b8c2dcba34 100644 --- a/src/diffusers/loaders/single_file_utils.py +++ b/src/diffusers/loaders/single_file_utils.py @@ -61,6 +61,7 @@ CONFIG_URLS = { "v1": "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml", "v2": "https://raw.githubusercontent.com/Stability-AI/stablediffusion/main/configs/stable-diffusion/v2-inference-v.yaml", "xl": "https://raw.githubusercontent.com/Stability-AI/generative-models/main/configs/inference/sd_xl_base.yaml", + "xl_refiner": "https://raw.githubusercontent.com/Stability-AI/generative-models/main/configs/inference/sd_xl_refiner.yaml", "upscale": "https://raw.githubusercontent.com/Stability-AI/stablediffusion/main/configs/stable-diffusion/x4-upscaling.yaml", } @@ -1355,6 +1356,8 @@ def create_text_encoders_and_tokenizers( else: return { + "text_encoder": None, + "tokenizer": None, "tokenizer_2": tokenizer_2, "text_encoder_2": text_encoder_2, } diff --git a/tests/pipelines/stable_diffusion_xl/test_stable_diffusion_xl_img2img.py b/tests/pipelines/stable_diffusion_xl/test_stable_diffusion_xl_img2img.py index 8931180808..33986b077a 100644 --- a/tests/pipelines/stable_diffusion_xl/test_stable_diffusion_xl_img2img.py +++ b/tests/pipelines/stable_diffusion_xl/test_stable_diffusion_xl_img2img.py @@ -784,24 +784,24 @@ class StableDiffusionXLImg2ImgIntegrationTests(unittest.TestCase): "/stable_diffusion_img2img/sketch-mountains-input.png" ) - pipe = StableDiffusionXLImg2ImgPipeline.from_single_file(ckpt_path) - pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config) - pipe.enable_model_cpu_offload() + pipe_single_file = StableDiffusionXLImg2ImgPipeline.from_single_file(ckpt_path, torch_dtype=torch.float16) + pipe_single_file.scheduler = DDIMScheduler.from_config(pipe_single_file.scheduler.config) + pipe_single_file.enable_model_cpu_offload() - generator = torch.Generator(device="cpu").manual_seed(0) - image_ckpt = pipe( - "mountains", image=init_image, num_inference_steps=2, generator=generator, output_type="np" - ).images[0] - - pipe = StableDiffusionXLImg2ImgPipeline.from_pretrained("stabilityai/stable-diffusion-xl-refiner-1.0") + pipe = StableDiffusionXLImg2ImgPipeline.from_pretrained("stabilityai/stable-diffusion-xl-refiner-1.0", torch_dtype=torch.float16) pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config) pipe.enable_model_cpu_offload() generator = torch.Generator(device="cpu").manual_seed(0) image = pipe( - "mountains", image=init_image, num_inference_steps=2, generator=generator, output_type="np" + prompt="mountains", image=init_image, num_inference_steps=2, generator=generator, output_type="np" ).images[0] - max_diff = numpy_cosine_similarity_distance(image.flatten(), image_ckpt.flatten()) + generator = torch.Generator(device="cpu").manual_seed(0) + image_single_file = pipe_single_file( + prompt="mountains", image=init_image, num_inference_steps=2, generator=generator, output_type="np" + ).images[0] + + max_diff = numpy_cosine_similarity_distance(image.flatten(), image_single_file.flatten()) assert max_diff < 1e-3