update

update

src/diffusers/commands/env.py

@@ -31,6 +31,7 @@ from ..utils import (
     is_transformers_available,
     is_xformers_available,
 )
+from ..utils.testing_utils import get_python_version
 from . import BaseDiffusersCLICommand
 
 
@@ -105,6 +106,11 @@ class EnvironmentCommand(BaseDiffusersCLICommand):
 
             xformers_version = xformers.__version__
 
+        if get_python_version() >= (3, 10):
+            platform_info = f"{platform.freedesktop_os_release().get('PRETTY_NAME', None)} - {platform.platform()}"
+        else:
+            platform_info = platform.platform()
+
         is_notebook_str = "Yes" if is_notebook() else "No"
 
         is_google_colab_str = "Yes" if is_google_colab() else "No"
@@ -152,7 +158,7 @@ class EnvironmentCommand(BaseDiffusersCLICommand):
 
         info = {
             "🤗 Diffusers version": version,
-            "Platform": f"{platform.freedesktop_os_release().get('PRETTY_NAME', None)} - {platform.platform()}",
+            "Platform": platform_info,
             "Running on a notebook?": is_notebook_str,
             "Running on Google Colab?": is_google_colab_str,
             "Python version": platform.python_version(),

src/diffusers/pipelines/stable_video_diffusion/pipeline_stable_video_diffusion.py

@@ -178,7 +178,7 @@ class StableVideoDiffusionPipeline(DiffusionPipeline):
             feature_extractor=feature_extractor,
         )
         self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
-        self.video_processor = VideoProcessor(do_resize=True, vae_scale_factor=self.vae_scale_factor)
+        self.video_processor = VideoProcessor(do_resize=False, vae_scale_factor=self.vae_scale_factor)
 
     def _encode_image(
         self,

tests/models/unets/test_models_unet_stable_cascade.py

@@ -1,191 +0,0 @@
-# coding=utf-8
-# Copyright 2024 HuggingFace Inc.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-import gc
-import unittest
-
-import torch
-
-from diffusers import StableCascadeUNet
-from diffusers.utils import logging
-from diffusers.utils.testing_utils import (
-    enable_full_determinism,
-    numpy_cosine_similarity_distance,
-    require_torch_gpu,
-    slow,
-)
-from diffusers.utils.torch_utils import randn_tensor
-
-
-logger = logging.get_logger(__name__)
-
-enable_full_determinism()
-
-
-@slow
-class StableCascadeUNetModelSlowTests(unittest.TestCase):
-    def tearDown(self) -> None:
-        super().tearDown()
-        gc.collect()
-        torch.cuda.empty_cache()
-
-    def test_stable_cascade_unet_prior_single_file_components(self):
-        single_file_url = "https://huggingface.co/stabilityai/stable-cascade/blob/main/stage_c_bf16.safetensors"
-        single_file_unet = StableCascadeUNet.from_single_file(single_file_url)
-
-        single_file_unet_config = single_file_unet.config
-        del single_file_unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        unet = StableCascadeUNet.from_pretrained("stabilityai/stable-cascade-prior", subfolder="prior", variant="bf16")
-        unet_config = unet.config
-        del unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        PARAMS_TO_IGNORE = ["torch_dtype", "_name_or_path", "_use_default_values", "_diffusers_version"]
-        for param_name, param_value in single_file_unet_config.items():
-            if param_name in PARAMS_TO_IGNORE:
-                continue
-
-            assert unet_config[param_name] == param_value
-
-    def test_stable_cascade_unet_decoder_single_file_components(self):
-        single_file_url = "https://huggingface.co/stabilityai/stable-cascade/blob/main/stage_b_bf16.safetensors"
-        single_file_unet = StableCascadeUNet.from_single_file(single_file_url)
-
-        single_file_unet_config = single_file_unet.config
-        del single_file_unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        unet = StableCascadeUNet.from_pretrained("stabilityai/stable-cascade", subfolder="decoder", variant="bf16")
-        unet_config = unet.config
-        del unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        PARAMS_TO_IGNORE = ["torch_dtype", "_name_or_path", "_use_default_values", "_diffusers_version"]
-        for param_name, param_value in single_file_unet_config.items():
-            if param_name in PARAMS_TO_IGNORE:
-                continue
-
-            assert unet_config[param_name] == param_value
-
-    def test_stable_cascade_unet_config_loading(self):
-        config = StableCascadeUNet.load_config(
-            pretrained_model_name_or_path="diffusers/stable-cascade-configs", subfolder="prior"
-        )
-        single_file_url = "https://huggingface.co/stabilityai/stable-cascade/blob/main/stage_c_bf16.safetensors"
-
-        single_file_unet = StableCascadeUNet.from_single_file(single_file_url, config=config)
-        single_file_unet_config = single_file_unet.config
-        del single_file_unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        PARAMS_TO_IGNORE = ["torch_dtype", "_name_or_path", "_use_default_values", "_diffusers_version"]
-        for param_name, param_value in config.items():
-            if param_name in PARAMS_TO_IGNORE:
-                continue
-
-            assert single_file_unet_config[param_name] == param_value
-
-    @require_torch_gpu
-    def test_stable_cascade_unet_single_file_prior_forward_pass(self):
-        dtype = torch.bfloat16
-        generator = torch.Generator("cpu")
-
-        model_inputs = {
-            "sample": randn_tensor((1, 16, 24, 24), generator=generator.manual_seed(0)).to("cuda", dtype),
-            "timestep_ratio": torch.tensor([1]).to("cuda", dtype),
-            "clip_text_pooled": randn_tensor((1, 1, 1280), generator=generator.manual_seed(0)).to("cuda", dtype),
-            "clip_text": randn_tensor((1, 77, 1280), generator=generator.manual_seed(0)).to("cuda", dtype),
-            "clip_img": randn_tensor((1, 1, 768), generator=generator.manual_seed(0)).to("cuda", dtype),
-            "pixels": randn_tensor((1, 3, 8, 8), generator=generator.manual_seed(0)).to("cuda", dtype),
-        }
-
-        unet = StableCascadeUNet.from_pretrained(
-            "stabilityai/stable-cascade-prior",
-            subfolder="prior",
-            revision="refs/pr/2",
-            variant="bf16",
-            torch_dtype=dtype,
-        )
-        unet.to("cuda")
-        with torch.no_grad():
-            prior_output = unet(**model_inputs).sample.float().cpu().numpy()
-
-        # Remove UNet from GPU memory before loading the single file UNet model
-        del unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        single_file_url = "https://huggingface.co/stabilityai/stable-cascade/blob/main/stage_c_bf16.safetensors"
-        single_file_unet = StableCascadeUNet.from_single_file(single_file_url, torch_dtype=dtype)
-        single_file_unet.to("cuda")
-        with torch.no_grad():
-            prior_single_file_output = single_file_unet(**model_inputs).sample.float().cpu().numpy()
-
-        # Remove UNet from GPU memory before loading the single file UNet model
-        del single_file_unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        max_diff = numpy_cosine_similarity_distance(prior_output.flatten(), prior_single_file_output.flatten())
-        assert max_diff < 8e-3
-
-    @require_torch_gpu
-    def test_stable_cascade_unet_single_file_decoder_forward_pass(self):
-        dtype = torch.float32
-        generator = torch.Generator("cpu")
-
-        model_inputs = {
-            "sample": randn_tensor((1, 4, 256, 256), generator=generator.manual_seed(0)).to("cuda", dtype),
-            "timestep_ratio": torch.tensor([1]).to("cuda", dtype),
-            "clip_text": randn_tensor((1, 77, 1280), generator=generator.manual_seed(0)).to("cuda", dtype),
-            "clip_text_pooled": randn_tensor((1, 1, 1280), generator=generator.manual_seed(0)).to("cuda", dtype),
-            "pixels": randn_tensor((1, 3, 8, 8), generator=generator.manual_seed(0)).to("cuda", dtype),
-        }
-
-        unet = StableCascadeUNet.from_pretrained(
-            "stabilityai/stable-cascade",
-            subfolder="decoder",
-            revision="refs/pr/44",
-            torch_dtype=dtype,
-        )
-        unet.to("cuda")
-        with torch.no_grad():
-            prior_output = unet(**model_inputs).sample.float().cpu().numpy()
-
-        # Remove UNet from GPU memory before loading the single file UNet model
-        del unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        single_file_url = "https://huggingface.co/stabilityai/stable-cascade/blob/main/stage_b.safetensors"
-        single_file_unet = StableCascadeUNet.from_single_file(single_file_url, torch_dtype=dtype)
-        single_file_unet.to("cuda")
-        with torch.no_grad():
-            prior_single_file_output = single_file_unet(**model_inputs).sample.float().cpu().numpy()
-
-        # Remove UNet from GPU memory before loading the single file UNet model
-        del single_file_unet
-        gc.collect()
-        torch.cuda.empty_cache()
-
-        max_diff = numpy_cosine_similarity_distance(prior_output.flatten(), prior_single_file_output.flatten())
-        assert max_diff < 1e-4