Add support for multiple controlnet (#691)

* add sd multi controlnet support * add tests * add doc * make style * add cv2 dependencies * fix ci error introduced by BatchFeature in transformers * fix * fix changes introduced with transformers/#31980 * pin libneuronxla * Update docs/source/inference_tutorials/stable_diffusion.mdx * Update optimum/neuron/modeling_diffusion.py
huggingface · Sep 18, 2024 · f8b5259 · f8b5259
1 parent d55d3ad
commit f8b5259
Show file tree

Hide file tree

Showing 16 changed files with 196 additions and 89 deletions.
diff --git a/.github/workflows/test_inf2_inference.yml b/.github/workflows/test_inf2_inference.yml
@@ -34,6 +34,9 @@ jobs:
           sudo apt-get update -y
           sudo apt-get install aws-neuronx-tools=2.17.1.0 aws-neuronx-runtime-lib=2.20.22.0-1b3ca6425 aws-neuronx-collectives=2.20.22.0-c101c322e  -y
           export PATH=/opt/aws/neuron/bin:$PATH
+      - name: Install cv2 dependencies
+        run: |
+          sudo apt-get install ffmpeg libsm6 libxext6  -y
       - name: Checkout
         uses: actions/checkout@v2
       - name: Install python dependencies

diff --git a/docs/source/inference_tutorials/stable_diffusion.mdx b/docs/source/inference_tutorials/stable_diffusion.mdx
@@ -635,6 +635,68 @@ compare.save("compare.png")
 />
 
 
+### MultiControlNet
+
+With Optimum Neuron, you can also compose multiple ControlNet conditionings from different image inputs:
+
+* Compile multiple ControlNet for SD1.5
+
+```bash
+optimum-cli export neuron --inline-weights-neff --model jyoung105/stable-diffusion-v1-5 --task stable-diffusion --auto_cast matmul --auto_cast_type bf16 --batch_size 1 --num_images_per_prompt 1 --controlnet_ids lllyasviel/control_v11p_sd15_openpose lllyasviel/control_v11f1p_sd15_depth --height 512 --width 512 sd15-512x512-bf16-openpose-depth
+```
+
+* Run SD1.5 with OpenPose and Depth conditionings:
+
+```python
+import numpy as np
+import torch
+from PIL import Image
+
+from controlnet_aux import OpenposeDetector
+from transformers import pipeline
+from diffusers import UniPCMultistepScheduler
+from diffusers.utils import load_image
+from optimum.neuron import NeuronStableDiffusionControlNetPipeline
+
+
+# OpenPose+Depth ControlNet
+model_id = "sd15-512x512-bf16-openpose-depth"
+
+# Load ControlNet images
+
+# 1. openpose
+image = load_image("https://huggingface.co/lllyasviel/control_v11p_sd15_openpose/resolve/main/images/input.png")
+processor = OpenposeDetector.from_pretrained('lllyasviel/ControlNet')
+openpose_image = processor(image)
+
+# 2. depth
+image = load_image("https://huggingface.co/lllyasviel/control_v11p_sd15_depth/resolve/main/images/input.png")
+depth_estimator = pipeline('depth-estimation')
+image = depth_estimator(image)['depth']
+image = np.array(image)
+image = image[:, :, None]
+image = np.concatenate([image, image, image], axis=2)
+depth_image = Image.fromarray(image)
+
+images = [openpose_image.resize((512, 512)), depth_image.resize((512, 512))]
+
+# 3. inference
+pipe = NeuronStableDiffusionControlNetPipeline.from_pretrained(model_id)
+pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
+prompt = "a giant in a fantasy landscape, best quality"
+negative_prompt = "monochrome, lowres, bad anatomy, worst quality, low quality"
+
+image = pipe(prompt=prompt, image=images).images[0]
+image.save('out.png')
+```
+
+<img
+  src="https://huggingface.co/datasets/Jingya/document_images/resolve/main/optimum/neuron/multicontrolnet.png"
+  width="768"
+  height="256"
+  alt="stable diffusion 1.5 generated image with OpenPose and Depth controlnet."
+/>
+
 
 ## ControlNet with Stable Diffusion XL
 

diff --git a/optimum/exporters/neuron/__main__.py b/optimum/exporters/neuron/__main__.py
@@ -472,7 +472,7 @@ def load_models_and_neuron_configs(
     revision: str,
     force_download: bool,
     local_files_only: bool,
-    use_auth_token: Optional[Union[bool, str]],
+    token: Optional[Union[bool, str]],
     submodels: Optional[Dict[str, Union[Path, str]]],
     lora_model_ids: Optional[Union[str, List[str]]],
     lora_weight_names: Optional[Union[str, List[str]]],
@@ -494,7 +494,7 @@ def load_models_and_neuron_configs(
         "subfolder": subfolder,
         "revision": revision,
         "cache_dir": cache_dir,
-        "use_auth_token": use_auth_token,
+        "token": token,
         "local_files_only": local_files_only,
         "force_download": force_download,
         "trust_remote_code": trust_remote_code,
@@ -544,7 +544,7 @@ def main_export(
     revision: str = "main",
     force_download: bool = False,
     local_files_only: bool = False,
-    use_auth_token: Optional[Union[bool, str]] = None,
+    token: Optional[Union[bool, str]] = None,
     do_validation: bool = True,
     submodels: Optional[Dict[str, Union[Path, str]]] = None,
     output_attentions: bool = False,
@@ -575,7 +575,7 @@ def main_export(
         revision=revision,
         force_download=force_download,
         local_files_only=local_files_only,
-        use_auth_token=use_auth_token,
+        token=token,
         submodels=submodels,
         output_attentions=output_attentions,
         output_hidden_states=output_hidden_states,

diff --git a/optimum/neuron/generation/utils.py b/optimum/neuron/generation/utils.py
@@ -719,7 +719,7 @@ def generate(
                 batch_size=batch_size,
                 model_input_name=model_input_name,
                 model_kwargs=model_kwargs,
-                decoder_start_token_id=generation_config.decoder_start_token_id,
+                decoder_start_token_id=generation_config._decoder_start_token_tensor,
                 device=inputs_tensor.device,
             )
         else:

diff --git a/optimum/neuron/modeling.py b/optimum/neuron/modeling.py
@@ -684,6 +684,13 @@ class NeuronModelForImageClassification(NeuronTracedModel):
 
     auto_model_class = AutoModelForImageClassification
 
+    @property
+    def dtype(self) -> Optional["torch.dtype"]:
+        """
+        Torch dtype of the inputs to avoid error in transformers on casting a BatchFeature to type None.
+        """
+        return getattr(self.config.neuron, "input_dtype", torch.float32)
+
     @add_start_docstrings_to_model_forward(
         NEURON_IMAGE_INPUTS_DOCSTRING.format("batch_size, num_channels, height, width")
         + IMAGE_CLASSIFICATION_EXAMPLE.format(
@@ -763,6 +770,13 @@ class NeuronModelForSemanticSegmentation(NeuronTracedModel):
 
     auto_model_class = AutoModelForSemanticSegmentation
 
+    @property
+    def dtype(self) -> Optional["torch.dtype"]:
+        """
+        Torch dtype of the inputs to avoid error in transformers on casting a BatchFeature to type None.
+        """
+        return getattr(self.config.neuron, "input_dtype", torch.float32)
+
     @add_start_docstrings_to_model_forward(
         NEURON_IMAGE_INPUTS_DOCSTRING.format("batch_size, num_channels, height, width")
         + SEMANTIC_SEGMENTATION_EXAMPLE.format(
@@ -843,6 +857,13 @@ class NeuronModelForObjectDetection(NeuronTracedModel):
 
     auto_model_class = AutoModelForObjectDetection
 
+    @property
+    def dtype(self) -> Optional["torch.dtype"]:
+        """
+        Torch dtype of the inputs to avoid error in transformers on casting a BatchFeature to type None.
+        """
+        return getattr(self.config.neuron, "input_dtype", torch.float32)
+
     @add_start_docstrings_to_model_forward(
         NEURON_IMAGE_INPUTS_DOCSTRING.format("batch_size, num_channels, height, width")
         + OBJECT_DETECTION_EXAMPLE.format(

diff --git a/optimum/neuron/modeling_decoder.py b/optimum/neuron/modeling_decoder.py
@@ -25,8 +25,7 @@
 from tempfile import TemporaryDirectory
 from typing import TYPE_CHECKING, Optional, Tuple, Union
 
-from huggingface_hub import HfApi, get_token, snapshot_download
-from huggingface_hub.utils import is_google_colab
+from huggingface_hub import HfApi, snapshot_download
 from transformers import AutoConfig, AutoModel, GenerationConfig
 
 from ..exporters.neuron.model_configs import *  # noqa: F403
@@ -225,7 +224,7 @@ def __init__(
     def _create_checkpoint(
         cls,
         model_id: str,
-        use_auth_token: Optional[Union[bool, str]] = None,
+        token: Optional[Union[bool, str]] = None,
         revision: Optional[str] = None,
         force_download: bool = False,
         cache_dir: Optional[str] = None,
@@ -243,7 +242,7 @@ def _create_checkpoint(
             revision=revision,
             framework="pt",
             cache_dir=cache_dir,
-            use_auth_token=use_auth_token,
+            token=token,
             local_files_only=local_files_only,
             force_download=force_download,
             trust_remote_code=trust_remote_code,
@@ -269,7 +268,7 @@ def get_export_config(
         cls,
         model_id: str,
         config: "PretrainedConfig",
-        use_auth_token: Optional[str] = None,
+        token: Optional[Union[bool, str]] = None,
         revision: Optional[str] = None,
         task: Optional[str] = None,
         batch_size: Optional[int] = None,
@@ -286,7 +285,7 @@ def get_export_config(
         else:
             checkpoint_id = model_id
             # Get the exact checkpoint revision (SHA1)
-            api = HfApi(token=use_auth_token)
+            api = HfApi(token=token)
             model_info = api.repo_info(model_id, revision=revision)
             checkpoint_revision = model_info.sha
 
@@ -337,7 +336,7 @@ def _export(
         cls,
         model_id: str,
         config: "PretrainedConfig",
-        use_auth_token: Optional[str] = None,
+        token: Optional[Union[bool, str]] = None,
         revision: Optional[str] = None,
         task: Optional[str] = None,
         batch_size: Optional[int] = None,
@@ -353,7 +352,7 @@ def _export(
         new_config = cls.get_export_config(
             model_id,
             config,
-            use_auth_token=use_auth_token,
+            token=token,
             revision=revision,
             task=task,
             batch_size=batch_size,
@@ -396,7 +395,7 @@ def _from_pretrained(
         cls,
         model_id: Union[str, Path],
         config: "PretrainedConfig",
-        use_auth_token: Optional[str] = None,
+        token: Optional[Union[bool, str]] = None,
         revision: Optional[str] = None,
         **kwargs,
     ) -> "NeuronDecoderModel":
@@ -411,7 +410,7 @@ def _from_pretrained(
 
         model_path = model_id
         if not os.path.isdir(model_id):
-            model_path = snapshot_download(model_id, token=use_auth_token, revision=revision)
+            model_path = snapshot_download(model_id, token=token, revision=revision)
 
         checkpoint_dir, compiled_dir = cls._get_neuron_dirs(model_path)
         if not os.path.isdir(checkpoint_dir):
@@ -425,7 +424,7 @@ def _from_pretrained(
                 checkpoint_id,
                 task=task,
                 revision=checkpoint_revision,
-                use_auth_token=use_auth_token,
+                token=token,
                 **kwargs,
             )
         assert os.path.isdir(compiled_dir)
@@ -467,24 +466,13 @@ def push_to_hub(
         repository_id: str,
         private: Optional[bool] = None,
         revision: Optional[str] = None,
-        use_auth_token: Union[bool, str] = True,
+        token: Union[bool, str] = True,
         endpoint: Optional[str] = None,
     ) -> str:
-        if isinstance(use_auth_token, str):
-            huggingface_token = use_auth_token
-        elif use_auth_token:
-            huggingface_token = get_token()
-        else:
-            raise ValueError("You need to provide `use_auth_token` to be able to push to the hub")
         api = HfApi(endpoint=endpoint)
 
-        user = api.whoami(huggingface_token)
-        if is_google_colab():
-            # Only in Google Colab to avoid the warning message
-            self.git_config_username_and_email(git_email=user["email"], git_user=user["fullname"])
-
         api.create_repo(
-            token=huggingface_token,
+            token=token,
             repo_id=repository_id,
             exist_ok=True,
             private=private,
@@ -498,7 +486,7 @@ def push_to_hub(
         api.upload_folder(
             repo_id=repository_id,
             folder_path=save_directory,
-            token=huggingface_token,
+            token=token,
             revision=revision,
             ignore_patterns=ignore_patterns,
         )
diff --git a/optimum/neuron/modeling_diffusion.py b/optimum/neuron/modeling_diffusion.py
@@ -548,7 +548,7 @@ def _from_pretrained(
         cls,
         model_id: Union[str, Path],
         config: Dict[str, Any],
-        use_auth_token: Optional[Union[bool, str]] = None,
+        token: Optional[Union[bool, str]] = None,
         revision: Optional[str] = None,
         force_download: bool = False,
         cache_dir: Optional[str] = None,
@@ -592,7 +592,7 @@ def _from_pretrained(
                 model_id,
                 cache_dir=cache_dir,
                 local_files_only=local_files_only,
-                use_auth_token=use_auth_token,
+                token=token,
                 revision=revision,
                 force_download=force_download,
                 allow_patterns=allow_patterns,
@@ -720,7 +720,7 @@ def _export(
         model_id: Union[str, Path],
         config: Dict[str, Any],
         unet_id: Optional[Union[str, Path]] = None,
-        use_auth_token: Optional[Union[bool, str]] = None,
+        token: Optional[Union[bool, str]] = None,
         revision: str = "main",
         force_download: bool = True,
         cache_dir: Optional[str] = None,
@@ -758,9 +758,9 @@ def _export(
                 configuration files of compatible classes.
             unet_id (`Optional[Union[str, Path]]`, defaults to `None`):
                 A string or a path point to the U-NET model to replace the one in the original pipeline.
-            use_auth_token (`Optional[Union[bool, str]]`, defaults to `None`):
+            token (`Optional[Union[bool, str]]`, defaults to `None`):
                 The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated
-                when running `transformers-cli login` (stored in `~/.huggingface`).
+                when running `huggingface-cli login` (stored in `huggingface_hub.constants.HF_TOKEN_PATH`).
             revision (`str`, defaults to `"main"`):
                 The specific model version to use (can be a branch name, tag name or commit id).
             force_download (`bool`, defaults to `True`):
@@ -837,7 +837,7 @@ def _export(
             framework="pt",
             library_name=cls.library_name,
             cache_dir=cache_dir,
-            use_auth_token=use_auth_token,
+            token=token,
             local_files_only=local_files_only,
             force_download=force_download,
             trust_remote_code=trust_remote_code,
@@ -863,7 +863,7 @@ def _export(
                 revision=revision,
                 force_download=force_download,
                 local_files_only=local_files_only,
-                use_auth_token=use_auth_token,
+                token=token,
                 submodels=submodels,
                 output_hidden_states=output_hidden_states,
                 lora_model_ids=lora_model_ids,
@@ -938,7 +938,7 @@ def _export(
                 revision=revision,
                 force_download=force_download,
                 local_files_only=local_files_only,
-                use_auth_token=use_auth_token,
+                token=token,
                 do_validation=False,
                 submodels={"unet": unet_id},
                 output_hidden_states=output_hidden_states,
@@ -1189,9 +1189,14 @@ def forward(
         encoder_hidden_states: torch.Tensor,
         controlnet_cond: torch.Tensor,
         conditioning_scale: float = 1.0,
+        guess_mode: bool = False,
         return_dict: bool = True,
     ) -> Union["ControlNetOutput", Tuple[Tuple[torch.Tensor, ...], torch.Tensor]]:
-        for i, (image, scale, controlnet) in enumerate(zip(controlnet_cond, conditioning_scale, self.model)):
+        if guess_mode:
+            logger.info(
+                "Guess mode is not yet supported. File us an issue on: https://github.com/huggingface/optimum-neuron/issues."
+            )
+        for i, (image, scale, controlnet) in enumerate(zip(controlnet_cond, conditioning_scale, self.nets)):
             inputs = (sample, timestep, encoder_hidden_states, image, scale)
             down_samples, mid_sample = controlnet(*inputs)