wip: Initial implementation of safetensor support for IP Adapter

invokeai/app/invocations/ip_adapter.py

@@ -4,18 +4,19 @@ from typing import List, Union
 from pydantic import BaseModel, Field, field_validator, model_validator
 from typing_extensions import Self
 
-from invokeai.app.invocations.baseinvocation import (
+from invokeai.app.invocations.baseinvocation import BaseInvocation, BaseInvocationOutput, invocation, invocation_output
-    BaseInvocation,
-    BaseInvocationOutput,
-    invocation,
-    invocation_output,
-)
 from invokeai.app.invocations.fields import FieldDescriptions, Input, InputField, OutputField, UIType
 from invokeai.app.invocations.model import ModelIdentifierField
 from invokeai.app.invocations.primitives import ImageField
 from invokeai.app.invocations.util import validate_begin_end_step, validate_weights
 from invokeai.app.services.shared.invocation_context import InvocationContext
-from invokeai.backend.model_manager.config import AnyModelConfig, BaseModelType, IPAdapterConfig, ModelType
+from invokeai.backend.model_manager.config import (
+    AnyModelConfig,
+    BaseModelType,
+    IPAdapterCheckpointConfig,
+    IPAdapterDiffusersConfig,
+    ModelType,
+)
 
 
 class IPAdapterField(BaseModel):
@@ -86,8 +87,12 @@ class IPAdapterInvocation(BaseInvocation):
     def invoke(self, context: InvocationContext) -> IPAdapterOutput:
         # Lookup the CLIP Vision encoder that is intended to be used with the IP-Adapter model.
         ip_adapter_info = context.models.get_config(self.ip_adapter_model.key)
-        assert isinstance(ip_adapter_info, IPAdapterConfig)
+        assert isinstance(ip_adapter_info, (IPAdapterDiffusersConfig, IPAdapterCheckpointConfig))
-        image_encoder_model_id = ip_adapter_info.image_encoder_model_id
+        image_encoder_model_id = (
+            ip_adapter_info.image_encoder_model_id
+            if isinstance(ip_adapter_info, IPAdapterDiffusersConfig)
+            else "InvokeAI/ip_adapter_sd_image_encoder"
+        )
         image_encoder_model_name = image_encoder_model_id.split("/")[-1].strip()
         image_encoder_model = self._get_image_encoder(context, image_encoder_model_name)
         return IPAdapterOutput(

invokeai/backend/ip_adapter/ip_adapter.py

@@ -1,10 +1,11 @@
 # copied from https://github.com/tencent-ailab/IP-Adapter (Apache License 2.0)
 #   and modified as needed
 
-from typing import Optional, Union
+from typing import List, Optional, TypedDict, Union
 
 import torch
 from PIL import Image
+from safetensors import safe_open
 from transformers import CLIPImageProcessor, CLIPVisionModelWithProjection
 
 from invokeai.backend.ip_adapter.ip_attention_weights import IPAttentionWeights
@@ -13,10 +14,17 @@ from ..raw_model import RawModel
 from .resampler import Resampler
 
 
+class IPAdapterStateDict(TypedDict):
+    ip_adapter: dict[str, torch.Tensor]
+    image_proj: dict[str, torch.Tensor]
+
+
 class ImageProjModel(torch.nn.Module):
     """Image Projection Model"""
 
-    def __init__(self, cross_attention_dim=1024, clip_embeddings_dim=1024, clip_extra_context_tokens=4):
+    def __init__(
+        self, cross_attention_dim: int = 1024, clip_embeddings_dim: int = 1024, clip_extra_context_tokens: int = 4
+    ):
         super().__init__()
 
         self.cross_attention_dim = cross_attention_dim
@@ -25,7 +33,7 @@ class ImageProjModel(torch.nn.Module):
         self.norm = torch.nn.LayerNorm(cross_attention_dim)
 
     @classmethod
-    def from_state_dict(cls, state_dict: dict[torch.Tensor], clip_extra_context_tokens=4):
+    def from_state_dict(cls, state_dict: dict[str, torch.Tensor], clip_extra_context_tokens: int = 4):
         """Initialize an ImageProjModel from a state_dict.
 
         The cross_attention_dim and clip_embeddings_dim are inferred from the shape of the tensors in the state_dict.
@@ -57,7 +65,7 @@ class ImageProjModel(torch.nn.Module):
 class MLPProjModel(torch.nn.Module):
     """SD model with image prompt"""
 
-    def __init__(self, cross_attention_dim=1024, clip_embeddings_dim=1024):
+    def __init__(self, cross_attention_dim: int = 1024, clip_embeddings_dim: int = 1024):
         super().__init__()
 
         self.proj = torch.nn.Sequential(
@@ -68,7 +76,7 @@ class MLPProjModel(torch.nn.Module):
         )
 
     @classmethod
-    def from_state_dict(cls, state_dict: dict[torch.Tensor]):
+    def from_state_dict(cls, state_dict: dict[str, torch.Tensor]):
         """Initialize an MLPProjModel from a state_dict.
 
         The cross_attention_dim and clip_embeddings_dim are inferred from the shape of the tensors in the state_dict.
@@ -97,7 +105,7 @@ class IPAdapter(RawModel):
 
     def __init__(
         self,
-        state_dict: dict[str, torch.Tensor],
+        state_dict: IPAdapterStateDict,
         device: torch.device,
         dtype: torch.dtype = torch.float16,
         num_tokens: int = 4,
@@ -129,13 +137,11 @@ class IPAdapter(RawModel):
 
         return calc_model_size_by_data(self._image_proj_model) + calc_model_size_by_data(self.attn_weights)
 
-    def _init_image_proj_model(self, state_dict):
+    def _init_image_proj_model(self, state_dict: dict[str, torch.Tensor]):
         return ImageProjModel.from_state_dict(state_dict, self._num_tokens).to(self.device, dtype=self.dtype)
 
     @torch.inference_mode()
-    def get_image_embeds(self, pil_image, image_encoder: CLIPVisionModelWithProjection):
+    def get_image_embeds(self, pil_image: List[Image.Image], image_encoder: CLIPVisionModelWithProjection):
-        if isinstance(pil_image, Image.Image):
-            pil_image = [pil_image]
         clip_image = self._clip_image_processor(images=pil_image, return_tensors="pt").pixel_values
         clip_image_embeds = image_encoder(clip_image.to(self.device, dtype=self.dtype)).image_embeds
         image_prompt_embeds = self._image_proj_model(clip_image_embeds)
@@ -146,7 +152,7 @@ class IPAdapter(RawModel):
 class IPAdapterPlus(IPAdapter):
     """IP-Adapter with fine-grained features"""
 
-    def _init_image_proj_model(self, state_dict):
+    def _init_image_proj_model(self, state_dict: dict[str, torch.Tensor]):
         return Resampler.from_state_dict(
             state_dict=state_dict,
             depth=4,
@@ -157,9 +163,7 @@ class IPAdapterPlus(IPAdapter):
         ).to(self.device, dtype=self.dtype)
 
     @torch.inference_mode()
-    def get_image_embeds(self, pil_image, image_encoder: CLIPVisionModelWithProjection):
+    def get_image_embeds(self, pil_image: List[Image.Image], image_encoder: CLIPVisionModelWithProjection):
-        if isinstance(pil_image, Image.Image):
-            pil_image = [pil_image]
         clip_image = self._clip_image_processor(images=pil_image, return_tensors="pt").pixel_values
         clip_image = clip_image.to(self.device, dtype=self.dtype)
         clip_image_embeds = image_encoder(clip_image, output_hidden_states=True).hidden_states[-2]
@@ -174,14 +178,14 @@ class IPAdapterPlus(IPAdapter):
 class IPAdapterFull(IPAdapterPlus):
     """IP-Adapter Plus with full features."""
 
-    def _init_image_proj_model(self, state_dict: dict[torch.Tensor]):
+    def _init_image_proj_model(self, state_dict: dict[str, torch.Tensor]):
         return MLPProjModel.from_state_dict(state_dict).to(self.device, dtype=self.dtype)
 
 
 class IPAdapterPlusXL(IPAdapterPlus):
     """IP-Adapter Plus for SDXL."""
 
-    def _init_image_proj_model(self, state_dict):
+    def _init_image_proj_model(self, state_dict: dict[str, torch.Tensor]):
         return Resampler.from_state_dict(
             state_dict=state_dict,
             depth=4,
@@ -195,7 +199,19 @@ class IPAdapterPlusXL(IPAdapterPlus):
 def build_ip_adapter(
     ip_adapter_ckpt_path: str, device: torch.device, dtype: torch.dtype = torch.float16
 ) -> Union[IPAdapter, IPAdapterPlus]:
-    state_dict = torch.load(ip_adapter_ckpt_path, map_location="cpu")
+    state_dict: IPAdapterStateDict = {"ip_adapter": {}, "image_proj": {}}
+
+    if ip_adapter_ckpt_path.endswith("safetensors"):
+        state_dict = {"ip_adapter": {}, "image_proj": {}}
+        model = safe_open(ip_adapter_ckpt_path, device=device.type, framework="pt")
+        for key in model.keys():
+            if key.startswith("image_proj."):
+                state_dict["image_proj"][key.replace("image_proj.", "")] = model.get_tensor(key)
+            if key.startswith("ip_adapter."):
+                state_dict["ip_adapter"][key.replace("ip_adapter.", "")] = model.get_tensor(key)
+    else:
+        ip_adapter_diffusers_checkpoint_path = ip_adapter_ckpt_path + "/ip_adapter.bin"
+        state_dict = torch.load(ip_adapter_diffusers_checkpoint_path, map_location="cpu")
 
     if "proj.weight" in state_dict["image_proj"]:  # IPAdapter (with ImageProjModel).
         return IPAdapter(state_dict, device=device, dtype=dtype)

invokeai/backend/ip_adapter/resampler.py

@@ -9,8 +9,8 @@ import torch.nn as nn
 
 
 # FFN
-def FeedForward(dim, mult=4):
+def FeedForward(dim: int, mult: int = 4):
-    inner_dim = int(dim * mult)
+    inner_dim = dim * mult
     return nn.Sequential(
         nn.LayerNorm(dim),
         nn.Linear(dim, inner_dim, bias=False),
@@ -19,8 +19,8 @@ def FeedForward(dim, mult=4):
     )
 
 
-def reshape_tensor(x, heads):
+def reshape_tensor(x: torch.Tensor, heads: int):
-    bs, length, width = x.shape
+    bs, length, _ = x.shape
     # (bs, length, width) --> (bs, length, n_heads, dim_per_head)
     x = x.view(bs, length, heads, -1)
     # (bs, length, n_heads, dim_per_head) --> (bs, n_heads, length, dim_per_head)
@@ -31,7 +31,7 @@ def reshape_tensor(x, heads):
 
 
 class PerceiverAttention(nn.Module):
-    def __init__(self, *, dim, dim_head=64, heads=8):
+    def __init__(self, *, dim: int, dim_head: int = 64, heads: int = 8):
         super().__init__()
         self.scale = dim_head**-0.5
         self.dim_head = dim_head
@@ -45,7 +45,7 @@ class PerceiverAttention(nn.Module):
         self.to_kv = nn.Linear(dim, inner_dim * 2, bias=False)
         self.to_out = nn.Linear(inner_dim, dim, bias=False)
 
-    def forward(self, x, latents):
+    def forward(self, x: torch.Tensor, latents: torch.Tensor):
         """
         Args:
             x (torch.Tensor): image features
@@ -80,14 +80,14 @@ class PerceiverAttention(nn.Module):
 class Resampler(nn.Module):
     def __init__(
         self,
-        dim=1024,
+        dim: int = 1024,
-        depth=8,
+        depth: int = 8,
-        dim_head=64,
+        dim_head: int = 64,
-        heads=16,
+        heads: int = 16,
-        num_queries=8,
+        num_queries: int = 8,
-        embedding_dim=768,
+        embedding_dim: int = 768,
-        output_dim=1024,
+        output_dim: int = 1024,
-        ff_mult=4,
+        ff_mult: int = 4,
     ):
         super().__init__()
 
@@ -110,7 +110,15 @@ class Resampler(nn.Module):
             )
 
     @classmethod
-    def from_state_dict(cls, state_dict: dict[torch.Tensor], depth=8, dim_head=64, heads=16, num_queries=8, ff_mult=4):
+    def from_state_dict(
+        cls,
+        state_dict: dict[str, torch.Tensor],
+        depth: int = 8,
+        dim_head: int = 64,
+        heads: int = 16,
+        num_queries: int = 8,
+        ff_mult: int = 4,
+    ):
         """A convenience function that initializes a Resampler from a state_dict.
 
         Some of the shape parameters are inferred from the state_dict (e.g. dim, embedding_dim, etc.). At the time of
@@ -145,7 +153,7 @@ class Resampler(nn.Module):
         model.load_state_dict(state_dict)
         return model
 
-    def forward(self, x):
+    def forward(self, x: torch.Tensor):
         latents = self.latents.repeat(x.size(0), 1, 1)
 
         x = self.proj_in(x)

invokeai/backend/model_manager/config.py

@@ -323,10 +323,13 @@ class MainDiffusersConfig(DiffusersConfigBase, MainConfigBase):
         return Tag(f"{ModelType.Main.value}.{ModelFormat.Diffusers.value}")
 
 
-class IPAdapterConfig(ModelConfigBase):
+class IPAdapterBaseConfig(ModelConfigBase):
-    """Model config for IP Adaptor format models."""
-
     type: Literal[ModelType.IPAdapter] = ModelType.IPAdapter
+
+
+class IPAdapterDiffusersConfig(IPAdapterBaseConfig):
+    """Model config for IP Adapter diffusers format models."""
+
     image_encoder_model_id: str
     format: Literal[ModelFormat.InvokeAI]
 
@@ -335,6 +338,16 @@ class IPAdapterConfig(ModelConfigBase):
         return Tag(f"{ModelType.IPAdapter.value}.{ModelFormat.InvokeAI.value}")
 
 
+class IPAdapterCheckpointConfig(IPAdapterBaseConfig):
+    """Model config for IP Adapter checkpoint format models."""
+
+    format: Literal[ModelFormat.Checkpoint]
+
+    @staticmethod
+    def get_tag() -> Tag:
+        return Tag(f"{ModelType.IPAdapter.value}.{ModelFormat.Checkpoint.value}")
+
+
 class CLIPVisionDiffusersConfig(DiffusersConfigBase):
     """Model config for CLIPVision."""
 
@@ -390,7 +403,8 @@ AnyModelConfig = Annotated[
         Annotated[LoRADiffusersConfig, LoRADiffusersConfig.get_tag()],
         Annotated[TextualInversionFileConfig, TextualInversionFileConfig.get_tag()],
         Annotated[TextualInversionFolderConfig, TextualInversionFolderConfig.get_tag()],
-        Annotated[IPAdapterConfig, IPAdapterConfig.get_tag()],
+        Annotated[IPAdapterDiffusersConfig, IPAdapterDiffusersConfig.get_tag()],
+        Annotated[IPAdapterCheckpointConfig, IPAdapterCheckpointConfig.get_tag()],
         Annotated[T2IAdapterConfig, T2IAdapterConfig.get_tag()],
         Annotated[CLIPVisionDiffusersConfig, CLIPVisionDiffusersConfig.get_tag()],
     ],

invokeai/backend/model_manager/load/model_loaders/ip_adapter.py

@@ -19,6 +19,7 @@ from invokeai.backend.model_manager.load import ModelLoader, ModelLoaderRegistry
 
 
 @ModelLoaderRegistry.register(base=BaseModelType.Any, type=ModelType.IPAdapter, format=ModelFormat.InvokeAI)
+@ModelLoaderRegistry.register(base=BaseModelType.Any, type=ModelType.IPAdapter, format=ModelFormat.Checkpoint)
 class IPAdapterInvokeAILoader(ModelLoader):
     """Class to load IP Adapter diffusers models."""
 
@@ -31,7 +32,7 @@ class IPAdapterInvokeAILoader(ModelLoader):
         if submodel_type is not None:
             raise ValueError("There are no submodels in an IP-Adapter model.")
         model = build_ip_adapter(
-            ip_adapter_ckpt_path=str(model_path / "ip_adapter.bin"),
+            ip_adapter_ckpt_path=str(model_path),
             device=torch.device("cpu"),
             dtype=self._torch_dtype,
         )

invokeai/backend/model_manager/probe.py

@@ -230,9 +230,10 @@ class ModelProbe(object):
                 return ModelType.LoRA
             elif any(key.startswith(v) for v in {"controlnet", "control_model", "input_blocks"}):
                 return ModelType.ControlNet
+            elif any(key.startswith(v) for v in {"image_proj.", "ip_adapter."}):
+                return ModelType.IPAdapter
             elif key in {"emb_params", "string_to_param"}:
                 return ModelType.TextualInversion
-
         else:
             # diffusers-ti
             if len(ckpt) < 10 and all(isinstance(v, torch.Tensor) for v in ckpt.values()):
@@ -527,8 +528,15 @@ class ControlNetCheckpointProbe(CheckpointProbeBase):
 
 
 class IPAdapterCheckpointProbe(CheckpointProbeBase):
+    """Class for probing IP Adapters"""
+
     def get_base_type(self) -> BaseModelType:
-        raise NotImplementedError()
+        checkpoint = self.checkpoint
+        for key in checkpoint.keys():
+            if not key.startswith(("image_proj.", "ip_adapter.")):
+                continue
+            return BaseModelType.StableDiffusionXL
+        raise InvalidModelConfigException(f"{self.model_path}: Unable to determine base type")
 
 
 class CLIPVisionCheckpointProbe(CheckpointProbeBase):
@@ -689,9 +697,7 @@ class ControlNetFolderProbe(FolderProbeBase):
             else (
                 BaseModelType.StableDiffusion2
                 if dimension == 1024
-                else BaseModelType.StableDiffusionXL
+                else BaseModelType.StableDiffusionXL if dimension == 2048 else None
-                if dimension == 2048
-                else None
             )
         )
         if not base_model:
@@ -768,7 +774,7 @@ class T2IAdapterFolderProbe(FolderProbeBase):
             )
 
 
-############## register probe classes ######
+# Register probe classes
 ModelProbe.register_probe("diffusers", ModelType.Main, PipelineFolderProbe)
 ModelProbe.register_probe("diffusers", ModelType.VAE, VaeFolderProbe)
 ModelProbe.register_probe("diffusers", ModelType.LoRA, LoRAFolderProbe)