[docs] Add safetensors flag (#4245)

* add safetensors flag

* apply review

docs/source/en/using-diffusers/inpaint.md

@@ -29,6 +29,7 @@ from diffusers import StableDiffusionInpaintPipeline
 pipeline = StableDiffusionInpaintPipeline.from_pretrained(
     "runwayml/stable-diffusion-inpainting",
     torch_dtype=torch.float16,
+    use_safetensors=True,
 )
 pipeline = pipeline.to("cuda")
 ```

docs/source/en/using-diffusers/loading.md

@@ -39,7 +39,7 @@ The [`DiffusionPipeline`] class is the simplest and most generic way to load any
 from diffusers import DiffusionPipeline
 
 repo_id = "runwayml/stable-diffusion-v1-5"
-pipe = DiffusionPipeline.from_pretrained(repo_id)
+pipe = DiffusionPipeline.from_pretrained(repo_id, use_safetensors=True)
 ```
 
 You can also load a checkpoint with it's specific pipeline class. The example above loaded a Stable Diffusion model; to get the same result, use the [`StableDiffusionPipeline`] class:
@@ -48,7 +48,7 @@ You can also load a checkpoint with it's specific pipeline class. The example ab
 from diffusers import StableDiffusionPipeline
 
 repo_id = "runwayml/stable-diffusion-v1-5"
-pipe = StableDiffusionPipeline.from_pretrained(repo_id)
+pipe = StableDiffusionPipeline.from_pretrained(repo_id, use_safetensors=True)
 ```
 
 A checkpoint (such as [`CompVis/stable-diffusion-v1-4`](https://huggingface.co/CompVis/stable-diffusion-v1-4) or [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5)) may also be used for more than one task, like text-to-image or image-to-image. To differentiate what task you want to use the checkpoint for, you have to load it directly with it's corresponding task-specific pipeline class:
@@ -75,7 +75,7 @@ Then pass the local path to [`~DiffusionPipeline.from_pretrained`]:
 from diffusers import DiffusionPipeline
 
 repo_id = "./stable-diffusion-v1-5"
-stable_diffusion = DiffusionPipeline.from_pretrained(repo_id)
+stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, use_safetensors=True)
 ```
 
 The [`~DiffusionPipeline.from_pretrained`] method won't download any files from the Hub when it detects a local path, but this also means it won't download and cache the latest changes to a checkpoint.
@@ -94,7 +94,7 @@ To find out which schedulers are compatible for customization, you can use the `
 from diffusers import DiffusionPipeline
 
 repo_id = "runwayml/stable-diffusion-v1-5"
-stable_diffusion = DiffusionPipeline.from_pretrained(repo_id)
+stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, use_safetensors=True)
 stable_diffusion.scheduler.compatibles
 ```
 
@@ -109,7 +109,7 @@ repo_id = "runwayml/stable-diffusion-v1-5"
 
 scheduler = EulerDiscreteScheduler.from_pretrained(repo_id, subfolder="scheduler")
 
-stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, scheduler=scheduler)
+stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, scheduler=scheduler, use_safetensors=True)
 ```
 
 ### Safety checker
@@ -120,7 +120,7 @@ Diffusion models like Stable Diffusion can generate harmful content, which is wh
 from diffusers import DiffusionPipeline
 
 repo_id = "runwayml/stable-diffusion-v1-5"
-stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, safety_checker=None)
+stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, safety_checker=None, use_safetensors=True)
 ```
 
 ### Reuse components across pipelines
@@ -131,7 +131,7 @@ You can also reuse the same components in multiple pipelines to avoid loading th
 from diffusers import StableDiffusionPipeline, StableDiffusionImg2ImgPipeline
 
 model_id = "runwayml/stable-diffusion-v1-5"
-stable_diffusion_txt2img = StableDiffusionPipeline.from_pretrained(model_id)
+stable_diffusion_txt2img = StableDiffusionPipeline.from_pretrained(model_id, use_safetensors=True)
 
 components = stable_diffusion_txt2img.components
 ```
@@ -148,7 +148,7 @@ You can also pass the components individually to the pipeline if you want more f
 from diffusers import StableDiffusionPipeline, StableDiffusionImg2ImgPipeline
 
 model_id = "runwayml/stable-diffusion-v1-5"
-stable_diffusion_txt2img = StableDiffusionPipeline.from_pretrained(model_id)
+stable_diffusion_txt2img = StableDiffusionPipeline.from_pretrained(model_id, use_safetensors=True)
 stable_diffusion_img2img = StableDiffusionImg2ImgPipeline(
     vae=stable_diffusion_txt2img.vae,
     text_encoder=stable_diffusion_txt2img.text_encoder,
@@ -194,10 +194,12 @@ import torch
 
 # load fp16 variant
 stable_diffusion = DiffusionPipeline.from_pretrained(
-    "runwayml/stable-diffusion-v1-5", variant="fp16", torch_dtype=torch.float16
+    "runwayml/stable-diffusion-v1-5", variant="fp16", torch_dtype=torch.float16, use_safetensors=True
 )
 # load non_ema variant
-stable_diffusion = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", variant="non_ema")
+stable_diffusion = DiffusionPipeline.from_pretrained(
+    "runwayml/stable-diffusion-v1-5", variant="non_ema", use_safetensors=True
+)
 ```
 
 To save a checkpoint stored in a different floating point type or as a non-EMA variant, use the [`DiffusionPipeline.save_pretrained`] method and specify the `variant` argument. You should try and save a variant to the same folder as the original checkpoint, so you can load both from the same folder:
@@ -215,10 +217,12 @@ If you don't save the variant to an existing folder, you must specify the `varia
 
 ```python
 # 👎 this won't work
-stable_diffusion = DiffusionPipeline.from_pretrained("./stable-diffusion-v1-5", torch_dtype=torch.float16)
+stable_diffusion = DiffusionPipeline.from_pretrained(
+    "./stable-diffusion-v1-5", torch_dtype=torch.float16, use_safetensors=True
+)
 # 👍 this works
 stable_diffusion = DiffusionPipeline.from_pretrained(
-    "./stable-diffusion-v1-5", variant="fp16", torch_dtype=torch.float16
+    "./stable-diffusion-v1-5", variant="fp16", torch_dtype=torch.float16, use_safetensors=True
 )
 ```
 
@@ -233,7 +237,7 @@ load model variants, e.g.:
 ```python
 from diffusers import DiffusionPipeline
 
-pipe = DiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", revision="fp16")
+pipe = DiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", revision="fp16", use_safetensors=True)
 ```
 
 However, this behavior is now deprecated since the "revision" argument should (just as it's done in GitHub) better be used to load model checkpoints from a specific commit or branch in development.
@@ -259,7 +263,7 @@ Models can be loaded from a subfolder with the `subfolder` argument. For example
 from diffusers import UNet2DConditionModel
 
 repo_id = "runwayml/stable-diffusion-v1-5"
-model = UNet2DConditionModel.from_pretrained(repo_id, subfolder="unet")
+model = UNet2DConditionModel.from_pretrained(repo_id, subfolder="unet", use_safetensors=True)
 ```
 
 Or directly from a repository's [directory](https://huggingface.co/google/ddpm-cifar10-32/tree/main):
@@ -268,7 +272,7 @@ Or directly from a repository's [directory](https://huggingface.co/google/ddpm-c
 from diffusers import UNet2DModel
 
 repo_id = "google/ddpm-cifar10-32"
-model = UNet2DModel.from_pretrained(repo_id)
+model = UNet2DModel.from_pretrained(repo_id, use_safetensors=True)
 ```
 
 You can also load and save model variants by specifying the `variant` argument in [`ModelMixin.from_pretrained`] and [`ModelMixin.save_pretrained`]:
@@ -276,7 +280,9 @@ You can also load and save model variants by specifying the `variant` argument i
 ```python
 from diffusers import UNet2DConditionModel
 
-model = UNet2DConditionModel.from_pretrained("runwayml/stable-diffusion-v1-5", subfolder="unet", variant="non-ema")
+model = UNet2DConditionModel.from_pretrained(
+    "runwayml/stable-diffusion-v1-5", subfolder="unet", variant="non-ema", use_safetensors=True
+)
 model.save_pretrained("./local-unet", variant="non-ema")
 ```
 
@@ -310,7 +316,7 @@ euler = EulerDiscreteScheduler.from_pretrained(repo_id, subfolder="scheduler")
 dpm = DPMSolverMultistepScheduler.from_pretrained(repo_id, subfolder="scheduler")
 
 # replace `dpm` with any of `ddpm`, `ddim`, `pndm`, `lms`, `euler_anc`, `euler`
-pipeline = StableDiffusionPipeline.from_pretrained(repo_id, scheduler=dpm)
+pipeline = StableDiffusionPipeline.from_pretrained(repo_id, scheduler=dpm, use_safetensors=True)
 ```
 
 ## DiffusionPipeline explained
@@ -326,7 +332,7 @@ The pipelines underlying folder structure corresponds directly with their class
 from diffusers import DiffusionPipeline
 
 repo_id = "runwayml/stable-diffusion-v1-5"
-pipeline = DiffusionPipeline.from_pretrained(repo_id)
+pipeline = DiffusionPipeline.from_pretrained(repo_id, use_safetensors=True)
 print(pipeline)
 ```
 

docs/source/en/using-diffusers/other-formats.md

@@ -111,7 +111,9 @@ If you prefer to run inference with code, click on the **Use in Diffusers** butt
 ```py
 from diffusers import DiffusionPipeline
 
-pipeline = DiffusionPipeline.from_pretrained("sayakpaul/textual-inversion-cat-kerascv_sd_diffusers_pipeline")
+pipeline = DiffusionPipeline.from_pretrained(
+    "sayakpaul/textual-inversion-cat-kerascv_sd_diffusers_pipeline", use_safetensors=True
+)
 ```
 
 Then you can generate an image like:
@@ -119,7 +121,9 @@ Then you can generate an image like:
 ```py
 from diffusers import DiffusionPipeline
 
-pipeline = DiffusionPipeline.from_pretrained("sayakpaul/textual-inversion-cat-kerascv_sd_diffusers_pipeline")
+pipeline = DiffusionPipeline.from_pretrained(
+    "sayakpaul/textual-inversion-cat-kerascv_sd_diffusers_pipeline", use_safetensors=True
+)
 pipeline.to("cuda")
 
 placeholder_token = "<my-funny-cat-token>"

docs/source/en/using-diffusers/reproducibility.md

@@ -40,7 +40,7 @@ import numpy as np
 model_id = "google/ddpm-cifar10-32"
 
 # load model and scheduler
-ddim = DDIMPipeline.from_pretrained(model_id)
+ddim = DDIMPipeline.from_pretrained(model_id, use_safetensors=True)
 
 # run pipeline for just two steps and return numpy tensor
 image = ddim(num_inference_steps=2, output_type="np").images
@@ -65,7 +65,7 @@ import numpy as np
 model_id = "google/ddpm-cifar10-32"
 
 # load model and scheduler
-ddim = DDIMPipeline.from_pretrained(model_id)
+ddim = DDIMPipeline.from_pretrained(model_id, use_safetensors=True)
 
 # create a generator for reproducibility
 generator = torch.Generator(device="cpu").manual_seed(0)
@@ -100,7 +100,7 @@ import numpy as np
 model_id = "google/ddpm-cifar10-32"
 
 # load model and scheduler
-ddim = DDIMPipeline.from_pretrained(model_id)
+ddim = DDIMPipeline.from_pretrained(model_id, use_safetensors=True)
 ddim.to("cuda")
 
 # create a generator for reproducibility
@@ -125,7 +125,7 @@ import numpy as np
 model_id = "google/ddpm-cifar10-32"
 
 # load model and scheduler
-ddim = DDIMPipeline.from_pretrained(model_id)
+ddim = DDIMPipeline.from_pretrained(model_id, use_safetensors=True)
 ddim.to("cuda")
 
 # create a generator for reproducibility; notice you don't place it on the GPU!
@@ -174,7 +174,7 @@ from diffusers import DDIMScheduler, StableDiffusionPipeline
 import numpy as np
 
 model_id = "runwayml/stable-diffusion-v1-5"
-pipe = StableDiffusionPipeline.from_pretrained(model_id).to("cuda")
+pipe = StableDiffusionPipeline.from_pretrained(model_id, use_safetensors=True).to("cuda")
 pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
 g = torch.Generator(device="cuda")
 

docs/source/en/using-diffusers/reusing_seeds.md

@@ -27,7 +27,9 @@ Instantiate a pipeline with [`DiffusionPipeline.from_pretrained`] and place it o
 ```python
 >>> from diffusers import DiffusionPipeline
 
->>> pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
+>>> pipe = DiffusionPipeline.from_pretrained(
+...     "runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16, use_safetensors=True
+... )
 >>> pipe = pipe.to("cuda")
 ```
 

docs/source/en/using-diffusers/schedulers.md

@@ -39,7 +39,9 @@ import torch
 
 login()
 
-pipeline = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
+pipeline = DiffusionPipeline.from_pretrained(
+    "runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16, use_safetensors=True
+)
 ```
 
 Next, we move it to GPU:

docs/source/en/using-diffusers/textual_inversion_inference.md

@@ -49,7 +49,9 @@ repo_id_embeds = "sd-concepts-library/cat-toy"
 Now you can load a pipeline, and pass the pre-learned concept to it:
 
 ```py
-pipeline = StableDiffusionPipeline.from_pretrained(pretrained_model_name_or_path, torch_dtype=torch.float16).to("cuda")
+pipeline = StableDiffusionPipeline.from_pretrained(
+    pretrained_model_name_or_path, torch_dtype=torch.float16, use_safetensors=True
+).to("cuda")
 
 pipeline.load_textual_inversion(repo_id_embeds)
 ```

docs/source/en/using-diffusers/unconditional_image_generation.md

@@ -32,7 +32,7 @@ In this guide, you'll use [`DiffusionPipeline`] for unconditional image generati
 ```python
 >>> from diffusers import DiffusionPipeline
 
->>> generator = DiffusionPipeline.from_pretrained("anton-l/ddpm-butterflies-128")
+>>> generator = DiffusionPipeline.from_pretrained("anton-l/ddpm-butterflies-128", use_safetensors=True)
 ```
 
 The [`DiffusionPipeline`] downloads and caches all modeling, tokenization, and scheduling components. 

docs/source/en/using-diffusers/using_safetensors.md

@@ -40,7 +40,9 @@ You can use the model with the new `.safetensors` weights by specifying the refe
 ```py
 from diffusers import DiffusionPipeline
 
-pipeline = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1", revision="refs/pr/22")
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-2-1", revision="refs/pr/22", use_safetensors=True
+)
 ```
 
 ## Why use safetensors?
@@ -55,7 +57,7 @@ There are several reasons for using safetensors:
 	```py
  from diffusers import StableDiffusionPipeline
 
- pipeline = StableDiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1")
+ pipeline = StableDiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1", use_safetensors=True)
  "Loaded in safetensors 0:00:02.033658"
  "Loaded in PyTorch 0:00:02.663379"
 	```

docs/source/en/using-diffusers/write_own_pipeline.md

@@ -25,7 +25,7 @@ A pipeline is a quick and easy way to run a model for inference, requiring no mo
 ```py
 >>> from diffusers import DDPMPipeline
 
->>> ddpm = DDPMPipeline.from_pretrained("google/ddpm-cat-256").to("cuda")
+>>> ddpm = DDPMPipeline.from_pretrained("google/ddpm-cat-256", use_safetensors=True).to("cuda")
 >>> image = ddpm(num_inference_steps=25).images[0]
 >>> image
 ```
@@ -46,7 +46,7 @@ To recreate the pipeline with the model and scheduler separately, let's write ou
 >>> from diffusers import DDPMScheduler, UNet2DModel
 
 >>> scheduler = DDPMScheduler.from_pretrained("google/ddpm-cat-256")
->>> model = UNet2DModel.from_pretrained("google/ddpm-cat-256").to("cuda")
+>>> model = UNet2DModel.from_pretrained("google/ddpm-cat-256", use_safetensors=True).to("cuda")
 ```
 
 2. Set the number of timesteps to run the denoising process for:
@@ -124,10 +124,14 @@ Now that you know what you need for the Stable Diffusion pipeline, load all thes
 >>> from transformers import CLIPTextModel, CLIPTokenizer
 >>> from diffusers import AutoencoderKL, UNet2DConditionModel, PNDMScheduler
 
->>> vae = AutoencoderKL.from_pretrained("CompVis/stable-diffusion-v1-4", subfolder="vae")
+>>> vae = AutoencoderKL.from_pretrained("CompVis/stable-diffusion-v1-4", subfolder="vae", use_safetensors=True)
 >>> tokenizer = CLIPTokenizer.from_pretrained("CompVis/stable-diffusion-v1-4", subfolder="tokenizer")
->>> text_encoder = CLIPTextModel.from_pretrained("CompVis/stable-diffusion-v1-4", subfolder="text_encoder")
->>> unet = UNet2DConditionModel.from_pretrained("CompVis/stable-diffusion-v1-4", subfolder="unet")
+>>> text_encoder = CLIPTextModel.from_pretrained(
+...     "CompVis/stable-diffusion-v1-4", subfolder="text_encoder", use_safetensors=True
+... )
+>>> unet = UNet2DConditionModel.from_pretrained(
+...     "CompVis/stable-diffusion-v1-4", subfolder="unet", use_safetensors=True
+... )
 ```
 
 Instead of the default [`PNDMScheduler`], exchange it for the [`UniPCMultistepScheduler`] to see how easy it is to plug a different scheduler in: