WIP - not working

We need to reset the staging area if we are discarding the last image.

.github/workflows/build-installer.yml

@@ -41,5 +41,5 @@ jobs:
       - name: upload installer artifact
         uses: actions/upload-artifact@v4
         with:
-          name: ${{ steps.create_installer.outputs.INSTALLER_FILENAME }}
+          name: installer
           path: ${{ steps.create_installer.outputs.INSTALLER_PATH }}

docs/RELEASE.md

@@ -61,11 +61,33 @@ This sets up both python and frontend dependencies and builds the python package
 
 #### Sanity Check & Smoke Test
 
-At this point, the release workflow pauses as the remaining publish jobs require approval.
+At this point, the release workflow pauses as the remaining publish jobs require approval. Time to test the installer.
 
-A maintainer should go to the **Summary** tab of the workflow, download the installer and test it. Ensure the app loads and generates.
+Because the installer pulls from PyPI, and we haven't published to PyPI yet, you will need to install from the wheel:
 
-> The same wheel file is bundled in the installer and in the `dist` artifact, which is uploaded to PyPI. You should end up with the exactly the same installation of the `invokeai` package from any of these methods.
+- Download and unzip `dist.zip` and the installer from the **Summary** tab of the workflow
+- Run the installer script using the `--wheel` CLI arg, pointing at the wheel:
+
+  ```sh
+  ./install.sh --wheel ../InvokeAI-4.0.0rc6-py3-none-any.whl
+  ```
+
+- Install to a temporary directory so you get the new user experience
+- Download a model and generate
+
+> The same wheel file is bundled in the installer and in the `dist` artifact, which is uploaded to PyPI. You should end up with the exactly the same installation as if the installer got the wheel from PyPI.
+
+##### Something isn't right
+
+If testing reveals any issues, no worries. Cancel the workflow, which will cancel the pending publish jobs (you didn't approve them prematurely, right?).
+
+Now you can start from the top:
+
+- Fix the issues and PR the fixes per usual
+- Get the PR approved and merged per usual
+- Switch to `main` and pull in the fixes
+- Run `make tag-release` to move the tag to `HEAD` (which has the fixes) and kick off the release workflow again
+- Re-do the sanity check
 
 #### PyPI Publish Jobs
 
@@ -81,6 +103,12 @@ Both jobs require a maintainer to approve them from the workflow's **Summary** t
 
 > **If the version already exists on PyPI, the publish jobs will fail.** PyPI only allows a given version to be published once - you cannot change it. If version published on PyPI has a problem, you'll need to "fail forward" by bumping the app version and publishing a followup release.
 
+##### Failing PyPI Publish
+
+Check the [python infrastructure status page] for incidents.
+
+If there are no incidents, contact @hipsterusername or @lstein, who have owner access to GH and PyPI, to see if access has expired or something like that.
+
 #### `publish-testpypi` Job
 
 Publishes the distribution on the [Test PyPI] index, using the `testpypi` GitHub environment.
@@ -110,11 +138,13 @@ Publishes the distribution on the production PyPI index, using the `pypi` GitHub
 Once the release is published to PyPI, it's time to publish the GitHub release.
 
 1. [Draft a new release] on GitHub, choosing the tag that triggered the release.
-2. Write the release notes, describing important changes. The **Generate release notes** button automatically inserts the changelog and new contributors, and you can copy/paste the intro from previous releases.
-3. Upload the zip file created in **`build`** job into the Assets section of the release notes. You can also upload the zip into the body of the release notes, since it can be hard for users to find the Assets section.
-4. Check the **Set as a pre-release** and **Create a discussion for this release** checkboxes at the bottom of the release page.
-5. Publish the pre-release.
-6. Announce the pre-release in Discord.
+1. Write the release notes, describing important changes. The **Generate release notes** button automatically inserts the changelog and new contributors, and you can copy/paste the intro from previous releases.
+1. Use `scripts/get_external_contributions.py` to get a list of external contributions to shout out in the release notes.
+1. Upload the zip file created in **`build`** job into the Assets section of the release notes.
+1. Check **Set as a pre-release** if it's a pre-release.
+1. Check **Create a discussion for this release**.
+1. Publish the release.
+1. Announce the release in Discord.
 
 > **TODO** Workflows can create a GitHub release from a template and upload release assets. One popular action to handle this is [ncipollo/release-action]. A future enhancement to the release process could set this up.
 
@@ -140,3 +170,4 @@ This functionality is available as a fallback in case something goes wonky. Typi
 [trusted publishers]: https://docs.pypi.org/trusted-publishers/
 [samuelcolvin/check-python-version]: https://github.com/samuelcolvin/check-python-version
 [manually]: #manual-release
+[python infrastructure status page]: https://status.python.org/

docs/help/FAQ.md

@@ -18,6 +18,22 @@ Note that any releases marked as _pre-release_ are in a beta state. You may expe
 
 The Model Manager tab in the UI provides a few ways to install models, including using your already-downloaded models. You'll see a popup directing you there on first startup. For more information, see the [model install docs].
 
+## Missing models after updating to v4
+
+If you find some models are missing after updating to v4, it's likely they weren't correctly registered before the update and didn't get picked up in the migration.
+
+You can use the `Scan Folder` tab in the Model Manager UI to fix this. The models will either be in the old, now-unused `autoimport` folder, or your `models` folder.
+
+- Find and copy your install's old `autoimport` folder path, install the main install folder.
+- Go to the Model Manager and click `Scan Folder`.
+- Paste the path and scan.
+- IMPORTANT: Uncheck `Inplace install`.
+- Click `Install All` to install all found models, or just install the models you want.
+
+Next, find and copy your install's `models` folder path (this could be your custom models folder path, or the `models` folder inside the main install folder).
+
+Follow the same steps to scan and import the missing models.
+
 ## Slow generation
 
 - Check the [system requirements] to ensure that your system is capable of generating images.

docs/installation/010_INSTALL_AUTOMATED.md

@@ -44,7 +44,7 @@ The installation process is simple, with a few prompts:
 
 - Select the version to install. Unless you have a specific reason to install a specific version, select the default (the latest version).
 - Select location for the install. Be sure you have enough space in this folder for the base application, as described in the [installation requirements].
-- Select a GPU device. If you are unsure, you can let the installer figure it out.
+- Select a GPU device.
 
 !!! info "Slow Installation"
 

docs/installation/020_INSTALL_MANUAL.md

@@ -6,11 +6,7 @@
 
 ## Introduction
 
-!!! tip "Conda"
-
-    As of InvokeAI v2.3.0 installation using the `conda` package manager is no longer being supported. It will likely still work, but we are not testing this installation method.
-
-InvokeAI is distributed as a python package on PyPI, installable with `pip`. There are a few things that are handled by the installer that you'll need to manage manually, described in this guide.
+InvokeAI is distributed as a python package on PyPI, installable with `pip`. There are a few things that are handled by the installer and launcher that you'll need to manage manually, described in this guide.
 
 ### Requirements
 
@@ -40,11 +36,11 @@ Before you start, go through the [installation requirements].
 
 1. Enter the root (invokeai) directory and create a virtual Python environment within it named `.venv`.
 
-    !!! info "Virtual Environment Location"
+    !!! warning "Virtual Environment Location"
 
         While you may create the virtual environment anywhere in the file system, we recommend that you create it within the root directory as shown here. This allows the application to automatically detect its data directories.
 
-        If you choose a different location for the venv, then you must set the `INVOKEAI_ROOT` environment variable or pass the directory using the `--root` CLI arg.
+        If you choose a different location for the venv, then you _must_ set the `INVOKEAI_ROOT` environment variable or specify the root directory using the `--root` CLI arg.
 
     ```terminal
     cd $INVOKEAI_ROOT
@@ -81,31 +77,23 @@ Before you start, go through the [installation requirements].
     python3 -m pip install --upgrade pip
     ```
 
-1. Install the InvokeAI Package. The `--extra-index-url` option is used to select the correct `torch` backend:
+1. Install the InvokeAI Package. The base command is `pip install InvokeAI --use-pep517`, but you may need to change this depending on your system and the desired features.
 
-    === "CUDA (NVidia)"
+    - You may need to provide an [extra index URL]. Select your platform configuration using [this tool on the PyTorch website]. Copy the `--extra-index-url` string from this and append it to your install command.
 
-         ```bash
-         pip install "InvokeAI[xformers]" --use-pep517 --extra-index-url https://download.pytorch.org/whl/cu121
-         ```
+        !!! example "Install with an extra index URL"
 
-    === "ROCm (AMD)"
+            ```bash
+            pip install InvokeAI --use-pep517 --extra-index-url https://download.pytorch.org/whl/cu121
+            ```
 
-         ```bash
-         pip install InvokeAI --use-pep517 --extra-index-url https://download.pytorch.org/whl/rocm5.6
-         ```
+    - If you have a CUDA GPU and want to install with `xformers`, you need to add an option to the package name. Note that `xformers` is not necessary. PyTorch includes an implementation of the SDP attention algorithm with the same performance.
 
-    === "CPU (Intel Macs & non-GPU systems)"
+        !!! example "Install with `xformers`"
 
-         ```bash
-         pip install InvokeAI --use-pep517 --extra-index-url https://download.pytorch.org/whl/cpu
-         ```
-
-    === "MPS (Apple Silicon)"
-
-         ```bash
-         pip install InvokeAI --use-pep517
-         ```
+            ```bash
+            pip install "InvokeAI[xformers]" --use-pep517
+            ```
 
 1. Deactivate and reactivate your runtime directory so that the invokeai-specific commands become available in the environment:
 
@@ -126,37 +114,6 @@ Before you start, go through the [installation requirements].
 
     Run `invokeai-web` to start the UI. You must activate the virtual environment before running the app.
 
-    If the virtual environment you selected is NOT inside `INVOKEAI_ROOT`, then you must specify the path to the root directory by adding
-    `--root_dir \path\to\invokeai`.
+    !!! warning
 
-    !!! tip
-
-        You can permanently set the location of the runtime directory
-        by setting the environment variable `INVOKEAI_ROOT` to the
-        path of the directory. As mentioned previously, this is
-        recommended if your virtual environment is located outside of
-        your runtime directory.
-
-## Unsupported Conda Install
-
-Congratulations, you found the "secret" Conda installation instructions. If you really **really** want to use Conda with InvokeAI, you can do so using this unsupported recipe:
-
-```sh
-mkdir ~/invokeai
-conda create -n invokeai python=3.11
-conda activate invokeai
-# Adjust this as described above for the appropriate torch backend
-pip install InvokeAI[xformers] --use-pep517 --extra-index-url https://download.pytorch.org/whl/cu121
-invokeai-web --root ~/invokeai
-```
-
-The `pip install` command shown in this recipe is for Linux/Windows
-systems with an NVIDIA GPU. See step (6) above for the command to use
-with other platforms/GPU combinations. If you don't wish to pass the
-`--root` argument to `invokeai` with each launch, you may set the
-environment variable `INVOKEAI_ROOT` to point to the installation directory.
-
-Note that if you run into problems with the Conda installation, the InvokeAI
-staff will **not** be able to help you out. Caveat Emptor!
-
-[installation requirements]: INSTALL_REQUIREMENTS.md
+        If the virtual environment is _not_ inside the root directory, then you _must_ specify the path to the root directory with `--root_dir \path\to\invokeai` or the `INVOKEAI_ROOT` environment variable.

docs/installation/INSTALL_DEVELOPMENT.md

@@ -32,5 +32,5 @@ As described in the [frontend dev toolchain] docs, you can run the UI using a de
 [Fork and clone]: https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/fork-a-repo
 [InvokeAI repo]: https://github.com/invoke-ai/InvokeAI
 [frontend dev toolchain]: ../contributing/frontend/OVERVIEW.md
-[manual installation]: installation/020_INSTALL_MANUAL.md
+[manual installation]: ./020_INSTALL_MANUAL.md
 [editable install]: https://pip.pypa.io/en/latest/cli/pip_install/#cmdoption-e

installer/lib/installer.py

@@ -3,6 +3,7 @@
 InvokeAI installer script
 """
 
+import locale
 import os
 import platform
 import re
@@ -316,7 +317,9 @@ def upgrade_pip(venv_path: Path) -> str | None:
     python = str(venv_path.expanduser().resolve() / python)
 
     try:
-        result = subprocess.check_output([python, "-m", "pip", "install", "--upgrade", "pip"]).decode()
+        result = subprocess.check_output([python, "-m", "pip", "install", "--upgrade", "pip"]).decode(
+            encoding=locale.getpreferredencoding()
+        )
     except subprocess.CalledProcessError as e:
         print(e)
         result = None
@@ -404,22 +407,29 @@ def get_torch_source() -> Tuple[str | None, str | None]:
     # device can be one of: "cuda", "rocm", "cpu", "cuda_and_dml, autodetect"
     device = select_gpu()
 
+    # The correct extra index URLs for torch are inconsistent, see https://pytorch.org/get-started/locally/#start-locally
+
     url = None
-    optional_modules = "[onnx]"
+    optional_modules: str | None = None
     if OS == "Linux":
         if device.value == "rocm":
             url = "https://download.pytorch.org/whl/rocm5.6"
         elif device.value == "cpu":
             url = "https://download.pytorch.org/whl/cpu"
-
+        elif device.value == "cuda":
+            # CUDA uses the default PyPi index
+            optional_modules = "[xformers,onnx-cuda]"
     elif OS == "Windows":
         if device.value == "cuda":
             url = "https://download.pytorch.org/whl/cu121"
             optional_modules = "[xformers,onnx-cuda]"
-        if device.value == "cuda_and_dml":
-            url = "https://download.pytorch.org/whl/cu121"
-            optional_modules = "[xformers,onnx-directml]"
+        elif device.value == "cpu":
+            # CPU  uses the default PyPi index, no optional modules
+            pass
+    elif OS == "Darwin":
+        # macOS uses the default PyPi index, no optional modules
+        pass
 
-    # in all other cases, Torch wheels should be coming from PyPi as of Torch 1.13
+    # Fall back to defaults
 
     return (url, optional_modules)

installer/lib/messages.py

@@ -207,10 +207,8 @@ def dest_path(dest: Optional[str | Path] = None) -> Path | None:
 
 class GpuType(Enum):
     CUDA = "cuda"
-    CUDA_AND_DML = "cuda_and_dml"
     ROCM = "rocm"
     CPU = "cpu"
-    AUTODETECT = "autodetect"
 
 
 def select_gpu() -> GpuType:
@@ -226,10 +224,6 @@ def select_gpu() -> GpuType:
         "an [gold1 b]NVIDIA[/] GPU (using CUDA™)",
         GpuType.CUDA,
     )
-    nvidia_with_dml = (
-        "an [gold1 b]NVIDIA[/] GPU (using CUDA™, and DirectML™ for ONNX) -- ALPHA",
-        GpuType.CUDA_AND_DML,
-    )
     amd = (
         "an [gold1 b]AMD[/] GPU (using ROCm™)",
         GpuType.ROCM,
@@ -238,27 +232,19 @@ def select_gpu() -> GpuType:
         "Do not install any GPU support, use CPU for generation (slow)",
         GpuType.CPU,
     )
-    autodetect = (
-        "I'm not sure what to choose",
-        GpuType.AUTODETECT,
-    )
 
     options = []
     if OS == "Windows":
-        options = [nvidia, nvidia_with_dml, cpu]
+        options = [nvidia, cpu]
     if OS == "Linux":
         options = [nvidia, amd, cpu]
     elif OS == "Darwin":
         options = [cpu]
-        # future CoreML?
 
     if len(options) == 1:
         print(f'Your platform [gold1]{OS}-{ARCH}[/] only supports the "{options[0][1]}" driver. Proceeding with that.')
         return options[0][1]
 
-    # "I don't know" is always added the last option
-    options.append(autodetect)  # type: ignore
-
     options = {str(i): opt for i, opt in enumerate(options, 1)}
 
     console.rule(":space_invader: GPU (Graphics Card) selection :space_invader:")
@@ -292,11 +278,6 @@ def select_gpu() -> GpuType:
         ),
     )
 
-    if options[choice][1] is GpuType.AUTODETECT:
-        console.print(
-            "No problem. We will install CUDA support first :crossed_fingers: If Invoke does not detect a GPU, please re-run the installer and select one of the other GPU types."
-        )
-
     return options[choice][1]
 
 

invokeai/app/api/routers/model_manager.py

@@ -219,28 +219,13 @@ async def scan_for_models(
         non_core_model_paths = [p for p in found_model_paths if not p.is_relative_to(core_models_path)]
 
         installed_models = ApiDependencies.invoker.services.model_manager.store.search_by_attr()
-        resolved_installed_model_paths: list[str] = []
-        installed_model_sources: list[str] = []
-
-        # This call lists all installed models.
-        for model in installed_models:
-            path = pathlib.Path(model.path)
-            # If the model has a source, we need to add it to the list of installed sources.
-            if model.source:
-                installed_model_sources.append(model.source)
-            # If the path is not absolute, that means it is in the app models directory, and we need to join it with
-            # the models path before resolving.
-            if not path.is_absolute():
-                resolved_installed_model_paths.append(str(pathlib.Path(models_path, path).resolve()))
-                continue
-            resolved_installed_model_paths.append(str(path.resolve()))
 
         scan_results: list[FoundModel] = []
 
-        # Check if the model is installed by comparing the resolved paths, appending to the scan result.
+        # Check if the model is installed by comparing paths, appending to the scan result.
         for p in non_core_model_paths:
             path = str(p)
-            is_installed = path in resolved_installed_model_paths or path in installed_model_sources
+            is_installed = any(str(models_path / m.path) == path for m in installed_models)
             found_model = FoundModel(path=path, is_installed=is_installed)
             scan_results.append(found_model)
     except Exception as e:
@@ -614,8 +599,8 @@ async def convert_model(
     The return value is the model configuration for the converted model.
     """
     model_manager = ApiDependencies.invoker.services.model_manager
+    loader = model_manager.load
     logger = ApiDependencies.invoker.services.logger
-    loader = ApiDependencies.invoker.services.model_manager.load
     store = ApiDependencies.invoker.services.model_manager.store
     installer = ApiDependencies.invoker.services.model_manager.install
 
@@ -630,7 +615,13 @@ async def convert_model(
         raise HTTPException(400, f"The model with key {key} is not a main checkpoint model.")
 
     # loading the model will convert it into a cached diffusers file
-    model_manager.load.load_model(model_config, submodel_type=SubModelType.Scheduler)
+    try:
+        cc_size = loader.convert_cache.max_size
+        if cc_size == 0:  # temporary set the convert cache to a positive number so that cached model is written
+            loader._convert_cache.max_size = 1.0
+        loader.load_model(model_config, submodel_type=SubModelType.Scheduler)
+    finally:
+        loader._convert_cache.max_size = cc_size
 
     # Get the path of the converted model from the loader
     cache_path = loader.convert_cache.cache_path(key)

invokeai/app/invocations/compel.py

@@ -9,7 +9,8 @@ from invokeai.app.invocations.fields import FieldDescriptions, Input, InputField
 from invokeai.app.invocations.primitives import ConditioningOutput
 from invokeai.app.services.shared.invocation_context import InvocationContext
 from invokeai.app.util.ti_utils import generate_ti_list
-from invokeai.backend.lora import LoRAModelRaw
+from invokeai.backend.lora.lora_model import LoRAModelRaw
+from invokeai.backend.lora.lora_model_patcher import LoraModelPatcher
 from invokeai.backend.model_patcher import ModelPatcher
 from invokeai.backend.stable_diffusion.diffusion.conditioning_data import (
     BasicConditioningInfo,
@@ -80,7 +81,7 @@ class CompelInvocation(BaseInvocation):
             ),
             text_encoder_info as text_encoder,
             # Apply the LoRA after text_encoder has been moved to its target device for faster patching.
-            ModelPatcher.apply_lora_text_encoder(text_encoder, _lora_loader()),
+            LoraModelPatcher.apply_lora_text_encoder(text_encoder, _lora_loader()),
             # Apply CLIP Skip after LoRA to prevent LoRA application from failing on skipped layers.
             ModelPatcher.apply_clip_skip(text_encoder_model, self.clip.skipped_layers),
         ):
@@ -181,7 +182,7 @@ class SDXLPromptInvocationBase:
             ),
             text_encoder_info as text_encoder,
             # Apply the LoRA after text_encoder has been moved to its target device for faster patching.
-            ModelPatcher.apply_lora(text_encoder, _lora_loader(), lora_prefix),
+            LoraModelPatcher.apply_lora(text_encoder, _lora_loader(), lora_prefix),
             # Apply CLIP Skip after LoRA to prevent LoRA application from failing on skipped layers.
             ModelPatcher.apply_clip_skip(text_encoder_model, clip_field.skipped_layers),
         ):

invokeai/app/invocations/custom_nodes/init.py

@@ -3,6 +3,7 @@ Invoke-managed custom node loader. See README.md for more information.
 """
 
 import sys
+import traceback
 from importlib.util import module_from_spec, spec_from_file_location
 from pathlib import Path
 
@@ -41,11 +42,15 @@ for d in Path(__file__).parent.iterdir():
 
     logger.info(f"Loading node pack {module_name}")
 
-    module = module_from_spec(spec)
-    sys.modules[spec.name] = module
-    spec.loader.exec_module(module)
+    try:
+        module = module_from_spec(spec)
+        sys.modules[spec.name] = module
+        spec.loader.exec_module(module)
 
-    loaded_count += 1
+        loaded_count += 1
+    except Exception:
+        full_error = traceback.format_exc()
+        logger.error(f"Failed to load node pack {module_name}:\n{full_error}")
 
     del init, module_name
 

invokeai/app/invocations/ip_adapter.py

@@ -1,21 +1,22 @@
 from builtins import float
-from typing import List, Union
+from typing import List, Literal, Union
 
 from pydantic import BaseModel, Field, field_validator, model_validator
 from typing_extensions import Self
 
-from invokeai.app.invocations.baseinvocation import (
-    BaseInvocation,
-    BaseInvocationOutput,
-    invocation,
-    invocation_output,
-)
+from invokeai.app.invocations.baseinvocation import 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,
+    IPAdapterInvokeAIConfig,
+    ModelType,
+)
 
 
 class IPAdapterField(BaseModel):
@@ -48,12 +49,15 @@ class IPAdapterOutput(BaseInvocationOutput):
     ip_adapter: IPAdapterField = OutputField(description=FieldDescriptions.ip_adapter, title="IP-Adapter")
 
 
+CLIP_VISION_MODEL_MAP = {"ViT-H": "ip_adapter_sd_image_encoder", "ViT-G": "ip_adapter_sdxl_image_encoder"}
+
+
 @invocation("ip_adapter", title="IP-Adapter", tags=["ip_adapter", "control"], category="ip_adapter", version="1.2.2")
 class IPAdapterInvocation(BaseInvocation):
     """Collects IP-Adapter info to pass to other nodes."""
 
     # Inputs
-    image: Union[ImageField, List[ImageField]] = InputField(description="The IP-Adapter image prompt(s).")
+    image: Union[ImageField, List[ImageField]] = InputField(description="The IP-Adapter image prompt(s).", ui_order=1)
     ip_adapter_model: ModelIdentifierField = InputField(
         description="The IP-Adapter model.",
         title="IP-Adapter Model",
@@ -61,7 +65,11 @@ class IPAdapterInvocation(BaseInvocation):
         ui_order=-1,
         ui_type=UIType.IPAdapterModel,
     )
-
+    clip_vision_model: Literal["auto", "ViT-H", "ViT-G"] = InputField(
+        description="CLIP Vision model to use. Overrides model settings. Mandatory for checkpoint models.",
+        default="auto",
+        ui_order=2,
+    )
     weight: Union[float, List[float]] = InputField(
         default=1, description="The weight given to the IP-Adapter", title="Weight"
     )
@@ -86,10 +94,21 @@ 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)
-        image_encoder_model_id = ip_adapter_info.image_encoder_model_id
-        image_encoder_model_name = image_encoder_model_id.split("/")[-1].strip()
+        assert isinstance(ip_adapter_info, (IPAdapterInvokeAIConfig, IPAdapterCheckpointConfig))
+
+        if self.clip_vision_model == "auto":
+            if isinstance(ip_adapter_info, IPAdapterInvokeAIConfig):
+                image_encoder_model_id = ip_adapter_info.image_encoder_model_id
+                image_encoder_model_name = image_encoder_model_id.split("/")[-1].strip()
+            else:
+                raise RuntimeError(
+                    "You need to set the appropriate CLIP Vision model for checkpoint IP Adapter models."
+                )
+        else:
+            image_encoder_model_name = CLIP_VISION_MODEL_MAP[self.clip_vision_model]
+
         image_encoder_model = self._get_image_encoder(context, image_encoder_model_name)
+
         return IPAdapterOutput(
             ip_adapter=IPAdapterField(
                 image=self.image,
@@ -102,19 +121,25 @@ class IPAdapterInvocation(BaseInvocation):
         )
 
     def _get_image_encoder(self, context: InvocationContext, image_encoder_model_name: str) -> AnyModelConfig:
-        found = False
-        while not found:
+        image_encoder_models = context.models.search_by_attrs(
+            name=image_encoder_model_name, base=BaseModelType.Any, type=ModelType.CLIPVision
+        )
+
+        if not len(image_encoder_models) > 0:
+            context.logger.warning(
+                f"The image encoder required by this IP Adapter ({image_encoder_model_name}) is not installed. \
+                    Downloading and installing now. This may take a while."
+            )
+
+            installer = context._services.model_manager.install
+            job = installer.heuristic_import(f"InvokeAI/{image_encoder_model_name}")
+            installer.wait_for_job(job, timeout=600)  # Wait for up to 10 minutes
             image_encoder_models = context.models.search_by_attrs(
                 name=image_encoder_model_name, base=BaseModelType.Any, type=ModelType.CLIPVision
             )
-            found = len(image_encoder_models) > 0
-            if not found:
-                context.logger.warning(
-                    f"The image encoder required by this IP Adapter ({image_encoder_model_name}) is not installed."
-                )
-                context.logger.warning("Downloading and installing now. This may take a while.")
-                installer = context._services.model_manager.install
-                job = installer.heuristic_import(f"InvokeAI/{image_encoder_model_name}")
-                installer.wait_for_job(job, timeout=600)  # wait up to 10 minutes - then raise a TimeoutException
-        assert len(image_encoder_models) == 1
+
+            if len(image_encoder_models) == 0:
+                context.logger.error("Error while fetching CLIP Vision Image Encoder")
+                assert len(image_encoder_models) == 1
+
         return image_encoder_models[0]

invokeai/app/invocations/latent.py

@@ -43,16 +43,13 @@ from invokeai.app.invocations.fields import (
     WithMetadata,
 )
 from invokeai.app.invocations.ip_adapter import IPAdapterField
-from invokeai.app.invocations.primitives import (
-    DenoiseMaskOutput,
-    ImageOutput,
-    LatentsOutput,
-)
+from invokeai.app.invocations.primitives import DenoiseMaskOutput, ImageOutput, LatentsOutput
 from invokeai.app.invocations.t2i_adapter import T2IAdapterField
 from invokeai.app.services.shared.invocation_context import InvocationContext
 from invokeai.app.util.controlnet_utils import prepare_control_image
 from invokeai.backend.ip_adapter.ip_adapter import IPAdapter, IPAdapterPlus
-from invokeai.backend.lora import LoRAModelRaw
+from invokeai.backend.lora.lora_model import LoRAModelRaw
+from invokeai.backend.lora.lora_model_patcher import LoraModelPatcher
 from invokeai.backend.model_manager import BaseModelType, LoadedModel
 from invokeai.backend.model_patcher import ModelPatcher
 from invokeai.backend.stable_diffusion import PipelineIntermediateState, set_seamless
@@ -68,12 +65,7 @@ from ...backend.stable_diffusion.diffusers_pipeline import (
 )
 from ...backend.stable_diffusion.schedulers import SCHEDULER_MAP
 from ...backend.util.devices import choose_precision, choose_torch_device
-from .baseinvocation import (
-    BaseInvocation,
-    BaseInvocationOutput,
-    invocation,
-    invocation_output,
-)
+from .baseinvocation import BaseInvocation, BaseInvocationOutput, invocation, invocation_output
 from .controlnet_image_processors import ControlField
 from .model import ModelIdentifierField, UNetField, VAEField
 
@@ -739,7 +731,7 @@ class DenoiseLatentsInvocation(BaseInvocation):
                 set_seamless(unet_info.model, self.unet.seamless_axes),  # FIXME
                 unet_info as unet,
                 # Apply the LoRA after unet has been moved to its target device for faster patching.
-                ModelPatcher.apply_lora_unet(unet, _lora_loader()),
+                LoraModelPatcher.apply_lora_unet(unet, _lora_loader()),
             ):
                 assert isinstance(unet, UNet2DConditionModel)
                 latents = latents.to(device=unet.device, dtype=unet.dtype)

invokeai/app/invocations/metadata.py

@@ -2,16 +2,8 @@ from typing import Any, Literal, Optional, Union
 
 from pydantic import BaseModel, ConfigDict, Field
 
-from invokeai.app.invocations.baseinvocation import (
-    BaseInvocation,
-    BaseInvocationOutput,
-    invocation,
-    invocation_output,
-)
-from invokeai.app.invocations.controlnet_image_processors import (
-    CONTROLNET_MODE_VALUES,
-    CONTROLNET_RESIZE_VALUES,
-)
+from invokeai.app.invocations.baseinvocation import BaseInvocation, BaseInvocationOutput, invocation, invocation_output
+from invokeai.app.invocations.controlnet_image_processors import CONTROLNET_MODE_VALUES, CONTROLNET_RESIZE_VALUES
 from invokeai.app.invocations.fields import (
     FieldDescriptions,
     ImageField,
@@ -43,6 +35,7 @@ class IPAdapterMetadataField(BaseModel):
 
     image: ImageField = Field(description="The IP-Adapter image prompt.")
     ip_adapter_model: ModelIdentifierField = Field(description="The IP-Adapter model.")
+    clip_vision_model: Literal["ViT-H", "ViT-G"] = Field(description="The CLIP Vision model")
     weight: Union[float, list[float]] = Field(description="The weight given to the IP-Adapter")
     begin_step_percent: float = Field(description="When the IP-Adapter is first applied (% of total steps)")
     end_step_percent: float = Field(description="When the IP-Adapter is last applied (% of total steps)")

invokeai/app/services/config/config_default.py

@@ -3,6 +3,7 @@
 
 from __future__ import annotations
 
+import locale
 import os
 import re
 import shutil
@@ -317,11 +318,10 @@ class InvokeAIAppConfig(BaseSettings):
     @staticmethod
     def find_root() -> Path:
         """Choose the runtime root directory when not specified on command line or init file."""
-        venv = Path(os.environ.get("VIRTUAL_ENV") or ".")
         if os.environ.get("INVOKEAI_ROOT"):
             root = Path(os.environ["INVOKEAI_ROOT"])
-        elif any((venv.parent / x).exists() for x in [INIT_FILE, LEGACY_INIT_FILE]):
-            root = (venv.parent).resolve()
+        elif venv := os.environ.get("VIRTUAL_ENV", None):
+            root = Path(venv).parent.resolve()
         else:
             root = Path("~/invokeai").expanduser().resolve()
         return root
@@ -373,13 +373,16 @@ def migrate_v3_config_dict(config_dict: dict[str, Any]) -> InvokeAIAppConfig:
             if k == "conf_path":
                 parsed_config_dict["legacy_models_yaml_path"] = v
             if k == "legacy_conf_dir":
-                # The old default for this was "configs/stable-diffusion". If if the incoming config has that as the value, we won't set it.
-                # Else if the path ends in "stable-diffusion", we assume the parent is the new correct path.
-                # Else we do not attempt to migrate this setting
-                if v != "configs/stable-diffusion":
-                    parsed_config_dict["legacy_conf_dir"] = v
+                # The old default for this was "configs/stable-diffusion" ("configs\stable-diffusion" on Windows).
+                if v == "configs/stable-diffusion" or v == "configs\\stable-diffusion":
+                    # If if the incoming config has the default value, skip
+                    continue
                 elif Path(v).name == "stable-diffusion":
+                    # Else if the path ends in "stable-diffusion", we assume the parent is the new correct path.
                     parsed_config_dict["legacy_conf_dir"] = str(Path(v).parent)
+                else:
+                    # Else we do not attempt to migrate this setting
+                    parsed_config_dict["legacy_conf_dir"] = v
             elif k in InvokeAIAppConfig.model_fields:
                 # skip unknown fields
                 parsed_config_dict[k] = v
@@ -399,7 +402,7 @@ def load_and_migrate_config(config_path: Path) -> InvokeAIAppConfig:
         An instance of `InvokeAIAppConfig` with the loaded and migrated settings.
     """
     assert config_path.suffix == ".yaml"
-    with open(config_path) as file:
+    with open(config_path, "rt", encoding=locale.getpreferredencoding()) as file:
         loaded_config_dict = yaml.safe_load(file)
 
     assert isinstance(loaded_config_dict, dict)

invokeai/app/services/model_install/model_install_default.py

@@ -1,5 +1,6 @@
 """Model installation class."""
 
+import locale
 import os
 import re
 import signal
@@ -323,7 +324,8 @@ class ModelInstallService(ModelInstallServiceBase):
             legacy_models_yaml_path = Path(self._app_config.root_path, legacy_models_yaml_path)
 
         if legacy_models_yaml_path.exists():
-            legacy_models_yaml = yaml.safe_load(legacy_models_yaml_path.read_text())
+            with open(legacy_models_yaml_path, "rt", encoding=locale.getpreferredencoding()) as file:
+                legacy_models_yaml = yaml.safe_load(file)
 
             yaml_metadata = legacy_models_yaml.pop("__metadata__")
             yaml_version = yaml_metadata.get("version")
@@ -348,8 +350,13 @@ class ModelInstallService(ModelInstallServiceBase):
                     config: dict[str, Any] = {}
                     config["name"] = model_name
                     config["description"] = stanza.get("description")
-                    config["config_path"] = stanza.get("config")
-
+                    legacy_config_path = stanza.get("config")
+                    if legacy_config_path:
+                        # In v3, these paths were relative to the root. Migrate them to be relative to the legacy_conf_dir.
+                        legacy_config_path: Path = self._app_config.root_path / legacy_config_path
+                        if legacy_config_path.is_relative_to(self._app_config.legacy_conf_path):
+                            legacy_config_path = legacy_config_path.relative_to(self._app_config.legacy_conf_path)
+                        config["config_path"] = str(legacy_config_path)
                     try:
                         id = self.register_path(model_path=model_path, config=config)
                         self._logger.info(f"Migrated {model_name} with id {id}")
@@ -368,11 +375,13 @@ class ModelInstallService(ModelInstallServiceBase):
     def delete(self, key: str) -> None:  # noqa D102
         """Unregister the model. Delete its files only if they are within our models directory."""
         model = self.record_store.get_model(key)
-        models_dir = self.app_config.models_path
-        model_path = models_dir / Path(model.path)  # handle legacy relative model paths
-        if model_path.is_relative_to(models_dir):
+        model_path = self.app_config.models_path / model.path
+
+        if model_path.is_relative_to(self.app_config.models_path):
+            # If the models is in the Invoke-managed models dir, we delete it
             self.unconditionally_delete(key)
         else:
+            # Else we only unregister it, leaving the file in place
             self.unregister(key)
 
     def unconditionally_delete(self, key: str) -> None:  # noqa D102
@@ -500,9 +509,9 @@ class ModelInstallService(ModelInstallServiceBase):
     def _scan_for_missing_models(self) -> list[AnyModelConfig]:
         """Scan the models directory for missing models and return a list of them."""
         missing_models: list[AnyModelConfig] = []
-        for x in self.record_store.all_models():
-            if not Path(x.path).resolve().exists():
-                missing_models.append(x)
+        for model_config in self.record_store.all_models():
+            if not (self.app_config.models_path / model_config.path).resolve().exists():
+                missing_models.append(model_config)
         return missing_models
 
     def _register_orphaned_models(self) -> None:
@@ -512,7 +521,9 @@ class ModelInstallService(ModelInstallServiceBase):
         only situations in which we may have orphaned models in the models directory.
         """
 
-        installed_model_paths = {Path(x.path).resolve() for x in self.record_store.all_models()}
+        installed_model_paths = {
+            (self._app_config.models_path / x.path).resolve() for x in self.record_store.all_models()
+        }
 
         # The bool returned by this callback determines if the model is added to the list of models found by the search
         def on_model_found(model_path: Path) -> bool:
@@ -548,20 +559,21 @@ class ModelInstallService(ModelInstallServiceBase):
         May raise an UnknownModelException.
         """
         model = self.record_store.get_model(key)
-        old_path = Path(model.path).resolve()
-        models_dir = self.app_config.models_path.resolve()
+        models_dir = self.app_config.models_path
+        old_path = self.app_config.models_path / model.path
 
         if not old_path.is_relative_to(models_dir):
+            # The model is not in the models directory - we don't need to move it.
             return model
 
-        new_path = (models_dir / model.base.value / model.type.value / model.name).with_suffix(old_path.suffix)
+        new_path = models_dir / model.base.value / model.type.value / old_path.name
 
         if old_path == new_path or new_path.exists() and old_path == new_path.resolve():
             return model
 
         self._logger.info(f"Moving {model.name} to {new_path}.")
         new_path = self._move_model(old_path, new_path)
-        model.path = new_path.as_posix()
+        model.path = new_path.relative_to(models_dir).as_posix()
         self.record_store.update_model(key, ModelRecordChanges(path=model.path))
         return model
 
@@ -600,12 +612,19 @@ class ModelInstallService(ModelInstallServiceBase):
 
         model_path = model_path.resolve()
 
+        # Models in the Invoke-managed models dir should use relative paths.
+        if model_path.is_relative_to(self.app_config.models_path):
+            model_path = model_path.relative_to(self.app_config.models_path)
+
         info.path = model_path.as_posix()
 
-        # Checkpoints have a config file needed for conversion - resolve this to an absolute path
         if isinstance(info, CheckpointConfigBase):
-            legacy_conf = (self.app_config.legacy_conf_path / info.config_path).resolve()
-            info.config_path = legacy_conf.as_posix()
+            # Checkpoints have a config file needed for conversion. Same handling as the model weights - if it's in the
+            # invoke-managed legacy config dir, we use a relative path.
+            legacy_config_path = self.app_config.legacy_conf_path / info.config_path
+            if legacy_config_path.is_relative_to(self.app_config.legacy_conf_path):
+                legacy_config_path = legacy_config_path.relative_to(self.app_config.legacy_conf_path)
+            info.config_path = legacy_config_path.as_posix()
         self.record_store.add_model(info)
         return info.key
 

invokeai/app/services/model_load/model_load_base.py

@@ -5,7 +5,8 @@ from abc import ABC, abstractmethod
 from typing import Optional
 
 from invokeai.app.services.shared.invocation_context import InvocationContextData
-from invokeai.backend.model_manager import AnyModel, AnyModelConfig, SubModelType
+from invokeai.backend.model_manager import AnyModelConfig, SubModelType
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.model_manager.load import LoadedModel
 from invokeai.backend.model_manager.load.convert_cache import ModelConvertCacheBase
 from invokeai.backend.model_manager.load.model_cache.model_cache_base import ModelCacheBase

invokeai/app/services/model_load/model_load_default.py

@@ -6,7 +6,8 @@ from typing import Optional, Type
 from invokeai.app.services.config import InvokeAIAppConfig
 from invokeai.app.services.invoker import Invoker
 from invokeai.app.services.shared.invocation_context import InvocationContextData
-from invokeai.backend.model_manager import AnyModel, AnyModelConfig, SubModelType
+from invokeai.backend.model_manager import AnyModelConfig, SubModelType
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.model_manager.load import (
     LoadedModel,
     ModelLoaderRegistry,

invokeai/app/services/model_manager/__init__.py

@@ -1,6 +1,6 @@
 """Initialization file for model manager service."""
 
-from invokeai.backend.model_manager import AnyModel, AnyModelConfig, BaseModelType, ModelType, SubModelType
+from invokeai.backend.model_manager import AnyModelConfig, BaseModelType, ModelType, SubModelType
 from invokeai.backend.model_manager.load import LoadedModel
 
 from .model_manager_default import ModelManagerService, ModelManagerServiceBase
@@ -8,7 +8,6 @@ from .model_manager_default import ModelManagerService, ModelManagerServiceBase
 __all__ = [
     "ModelManagerServiceBase",
     "ModelManagerService",
-    "AnyModel",
     "AnyModelConfig",
     "BaseModelType",
     "ModelType",

invokeai/app/services/session_processor/session_processor_default.py

@@ -70,8 +70,18 @@ class DefaultSessionProcessor(SessionProcessorBase):
     async def _on_queue_event(self, event: FastAPIEvent) -> None:
         event_name = event[1]["event"]
 
-        if event_name == "session_canceled" or event_name == "queue_cleared":
-            # These both mean we should cancel the current session.
+        if (
+            event_name == "session_canceled"
+            and self._queue_item
+            and self._queue_item.item_id == event[1]["data"]["queue_item_id"]
+        ):
+            self._cancel_event.set()
+            self._poll_now()
+        elif (
+            event_name == "queue_cleared"
+            and self._queue_item
+            and self._queue_item.queue_id == event[1]["data"]["queue_id"]
+        ):
             self._cancel_event.set()
             self._poll_now()
         elif event_name == "batch_enqueued":
@@ -111,141 +121,146 @@ class DefaultSessionProcessor(SessionProcessorBase):
                 poll_now_event.clear()
                 # Middle processor try block; any unhandled exception is a non-fatal processor error
                 try:
+                    # If we are paused, wait for resume event
+                    resume_event.wait()
+
                     # Get the next session to process
                     self._queue_item = self._invoker.services.session_queue.dequeue()
-                    if self._queue_item is not None and resume_event.is_set():
-                        self._invoker.services.logger.debug(f"Executing queue item {self._queue_item.item_id}")
-                        cancel_event.clear()
 
-                        # If profiling is enabled, start the profiler
-                        if self._profiler is not None:
-                            self._profiler.start(profile_id=self._queue_item.session_id)
+                    if self._queue_item is None:
+                        # The queue was empty, wait for next polling interval or event to try again
+                        self._invoker.services.logger.debug("Waiting for next polling interval or event")
+                        poll_now_event.wait(self._polling_interval)
+                        continue
 
-                        # Prepare invocations and take the first
-                        self._invocation = self._queue_item.session.next()
+                    self._invoker.services.logger.debug(f"Executing queue item {self._queue_item.item_id}")
+                    cancel_event.clear()
 
-                        # Loop over invocations until the session is complete or canceled
-                        while self._invocation is not None and not cancel_event.is_set():
-                            # get the source node id to provide to clients (the prepared node id is not as useful)
-                            source_invocation_id = self._queue_item.session.prepared_source_mapping[self._invocation.id]
+                    # If profiling is enabled, start the profiler
+                    if self._profiler is not None:
+                        self._profiler.start(profile_id=self._queue_item.session_id)
 
-                            # Send starting event
-                            self._invoker.services.events.emit_invocation_started(
-                                queue_batch_id=self._queue_item.batch_id,
-                                queue_item_id=self._queue_item.item_id,
-                                queue_id=self._queue_item.queue_id,
-                                graph_execution_state_id=self._queue_item.session_id,
-                                node=self._invocation.model_dump(),
-                                source_node_id=source_invocation_id,
-                            )
+                    # Prepare invocations and take the first
+                    self._invocation = self._queue_item.session.next()
 
-                            # Innermost processor try block; any unhandled exception is an invocation error & will fail the graph
-                            try:
-                                with self._invoker.services.performance_statistics.collect_stats(
-                                    self._invocation, self._queue_item.session.id
-                                ):
-                                    # Build invocation context (the node-facing API)
-                                    data = InvocationContextData(
-                                        invocation=self._invocation,
-                                        source_invocation_id=source_invocation_id,
-                                        queue_item=self._queue_item,
-                                    )
-                                    context = build_invocation_context(
-                                        data=data,
-                                        services=self._invoker.services,
-                                        cancel_event=self._cancel_event,
-                                    )
+                    # Loop over invocations until the session is complete or canceled
+                    while self._invocation is not None and not cancel_event.is_set():
+                        # get the source node id to provide to clients (the prepared node id is not as useful)
+                        source_invocation_id = self._queue_item.session.prepared_source_mapping[self._invocation.id]
 
-                                    # Invoke the node
-                                    outputs = self._invocation.invoke_internal(
-                                        context=context, services=self._invoker.services
-                                    )
+                        # Send starting event
+                        self._invoker.services.events.emit_invocation_started(
+                            queue_batch_id=self._queue_item.batch_id,
+                            queue_item_id=self._queue_item.item_id,
+                            queue_id=self._queue_item.queue_id,
+                            graph_execution_state_id=self._queue_item.session_id,
+                            node=self._invocation.model_dump(),
+                            source_node_id=source_invocation_id,
+                        )
 
-                                    # Save outputs and history
-                                    self._queue_item.session.complete(self._invocation.id, outputs)
-
-                                    # Send complete event
-                                    self._invoker.services.events.emit_invocation_complete(
-                                        queue_batch_id=self._queue_item.batch_id,
-                                        queue_item_id=self._queue_item.item_id,
-                                        queue_id=self._queue_item.queue_id,
-                                        graph_execution_state_id=self._queue_item.session.id,
-                                        node=self._invocation.model_dump(),
-                                        source_node_id=source_invocation_id,
-                                        result=outputs.model_dump(),
-                                    )
-
-                            except KeyboardInterrupt:
-                                # TODO(MM2): Create an event for this
-                                pass
-
-                            except CanceledException:
-                                # When the user cancels the graph, we first set the cancel event. The event is checked
-                                # between invocations, in this loop. Some invocations are long-running, and we need to
-                                # be able to cancel them mid-execution.
-                                #
-                                # For example, denoising is a long-running invocation with many steps. A step callback
-                                # is executed after each step. This step callback checks if the canceled event is set,
-                                # then raises a CanceledException to stop execution immediately.
-                                #
-                                # When we get a CanceledException, we don't need to do anything - just pass and let the
-                                # loop go to its next iteration, and the cancel event will be handled correctly.
-                                pass
-
-                            except Exception as e:
-                                error = traceback.format_exc()
-
-                                # Save error
-                                self._queue_item.session.set_node_error(self._invocation.id, error)
-                                self._invoker.services.logger.error(
-                                    f"Error while invoking session {self._queue_item.session_id}, invocation {self._invocation.id} ({self._invocation.get_type()}):\n{e}"
+                        # Innermost processor try block; any unhandled exception is an invocation error & will fail the graph
+                        try:
+                            with self._invoker.services.performance_statistics.collect_stats(
+                                self._invocation, self._queue_item.session.id
+                            ):
+                                # Build invocation context (the node-facing API)
+                                data = InvocationContextData(
+                                    invocation=self._invocation,
+                                    source_invocation_id=source_invocation_id,
+                                    queue_item=self._queue_item,
+                                )
+                                context = build_invocation_context(
+                                    data=data,
+                                    services=self._invoker.services,
+                                    cancel_event=self._cancel_event,
                                 )
-                                self._invoker.services.logger.error(error)
 
-                                # Send error event
-                                self._invoker.services.events.emit_invocation_error(
-                                    queue_batch_id=self._queue_item.session_id,
+                                # Invoke the node
+                                outputs = self._invocation.invoke_internal(
+                                    context=context, services=self._invoker.services
+                                )
+
+                                # Save outputs and history
+                                self._queue_item.session.complete(self._invocation.id, outputs)
+
+                                # Send complete event
+                                self._invoker.services.events.emit_invocation_complete(
+                                    queue_batch_id=self._queue_item.batch_id,
                                     queue_item_id=self._queue_item.item_id,
                                     queue_id=self._queue_item.queue_id,
                                     graph_execution_state_id=self._queue_item.session.id,
                                     node=self._invocation.model_dump(),
                                     source_node_id=source_invocation_id,
-                                    error_type=e.__class__.__name__,
-                                    error=error,
+                                    result=outputs.model_dump(),
                                 )
-                                pass
 
-                            # The session is complete if the all invocations are complete or there was an error
-                            if self._queue_item.session.is_complete() or cancel_event.is_set():
-                                # Send complete event
-                                self._invoker.services.events.emit_graph_execution_complete(
-                                    queue_batch_id=self._queue_item.batch_id,
-                                    queue_item_id=self._queue_item.item_id,
-                                    queue_id=self._queue_item.queue_id,
-                                    graph_execution_state_id=self._queue_item.session.id,
+                        except KeyboardInterrupt:
+                            # TODO(MM2): Create an event for this
+                            pass
+
+                        except CanceledException:
+                            # When the user cancels the graph, we first set the cancel event. The event is checked
+                            # between invocations, in this loop. Some invocations are long-running, and we need to
+                            # be able to cancel them mid-execution.
+                            #
+                            # For example, denoising is a long-running invocation with many steps. A step callback
+                            # is executed after each step. This step callback checks if the canceled event is set,
+                            # then raises a CanceledException to stop execution immediately.
+                            #
+                            # When we get a CanceledException, we don't need to do anything - just pass and let the
+                            # loop go to its next iteration, and the cancel event will be handled correctly.
+                            pass
+
+                        except Exception as e:
+                            error = traceback.format_exc()
+
+                            # Save error
+                            self._queue_item.session.set_node_error(self._invocation.id, error)
+                            self._invoker.services.logger.error(
+                                f"Error while invoking session {self._queue_item.session_id}, invocation {self._invocation.id} ({self._invocation.get_type()}):\n{e}"
+                            )
+                            self._invoker.services.logger.error(error)
+
+                            # Send error event
+                            self._invoker.services.events.emit_invocation_error(
+                                queue_batch_id=self._queue_item.session_id,
+                                queue_item_id=self._queue_item.item_id,
+                                queue_id=self._queue_item.queue_id,
+                                graph_execution_state_id=self._queue_item.session.id,
+                                node=self._invocation.model_dump(),
+                                source_node_id=source_invocation_id,
+                                error_type=e.__class__.__name__,
+                                error=error,
+                            )
+                            pass
+
+                        # The session is complete if the all invocations are complete or there was an error
+                        if self._queue_item.session.is_complete() or cancel_event.is_set():
+                            # Send complete event
+                            self._invoker.services.events.emit_graph_execution_complete(
+                                queue_batch_id=self._queue_item.batch_id,
+                                queue_item_id=self._queue_item.item_id,
+                                queue_id=self._queue_item.queue_id,
+                                graph_execution_state_id=self._queue_item.session.id,
+                            )
+                            # If we are profiling, stop the profiler and dump the profile & stats
+                            if self._profiler:
+                                profile_path = self._profiler.stop()
+                                stats_path = profile_path.with_suffix(".json")
+                                self._invoker.services.performance_statistics.dump_stats(
+                                    graph_execution_state_id=self._queue_item.session.id, output_path=stats_path
                                 )
-                                # If we are profiling, stop the profiler and dump the profile & stats
-                                if self._profiler:
-                                    profile_path = self._profiler.stop()
-                                    stats_path = profile_path.with_suffix(".json")
-                                    self._invoker.services.performance_statistics.dump_stats(
-                                        graph_execution_state_id=self._queue_item.session.id, output_path=stats_path
-                                    )
-                                # We'll get a GESStatsNotFoundError if we try to log stats for an untracked graph, but in the processor
-                                # we don't care about that - suppress the error.
-                                with suppress(GESStatsNotFoundError):
-                                    self._invoker.services.performance_statistics.log_stats(self._queue_item.session.id)
-                                    self._invoker.services.performance_statistics.reset_stats()
+                            # We'll get a GESStatsNotFoundError if we try to log stats for an untracked graph, but in the processor
+                            # we don't care about that - suppress the error.
+                            with suppress(GESStatsNotFoundError):
+                                self._invoker.services.performance_statistics.log_stats(self._queue_item.session.id)
+                                self._invoker.services.performance_statistics.reset_stats()
 
-                                # Set the invocation to None to prepare for the next session
-                                self._invocation = None
-                            else:
-                                # Prepare the next invocation
-                                self._invocation = self._queue_item.session.next()
-
-                        # The session is complete, immediately poll for next session
-                        self._queue_item = None
-                        poll_now_event.set()
+                            # Set the invocation to None to prepare for the next session
+                            self._invocation = None
+                        else:
+                            # Prepare the next invocation
+                            self._invocation = self._queue_item.session.next()
                     else:
                         # The queue was empty, wait for next polling interval or event to try again
                         self._invoker.services.logger.debug("Waiting for next polling interval or event")

invokeai/app/services/shared/sqlite/sqlite_util.py

@@ -10,6 +10,8 @@ from invokeai.app.services.shared.sqlite_migrator.migrations.migration_4 import
 from invokeai.app.services.shared.sqlite_migrator.migrations.migration_5 import build_migration_5
 from invokeai.app.services.shared.sqlite_migrator.migrations.migration_6 import build_migration_6
 from invokeai.app.services.shared.sqlite_migrator.migrations.migration_7 import build_migration_7
+from invokeai.app.services.shared.sqlite_migrator.migrations.migration_8 import build_migration_8
+from invokeai.app.services.shared.sqlite_migrator.migrations.migration_9 import build_migration_9
 from invokeai.app.services.shared.sqlite_migrator.sqlite_migrator_impl import SqliteMigrator
 
 
@@ -37,6 +39,8 @@ def init_db(config: InvokeAIAppConfig, logger: Logger, image_files: ImageFileSto
     migrator.register_migration(build_migration_5())
     migrator.register_migration(build_migration_6())
     migrator.register_migration(build_migration_7())
+    migrator.register_migration(build_migration_8(app_config=config))
+    migrator.register_migration(build_migration_9())
     migrator.run_migrations()
 
     return db

invokeai/app/services/shared/sqlite_migrator/migrations/migration_7.py

@@ -11,7 +11,7 @@ class Migration7Callback:
     def _drop_old_models_tables(self, cursor: sqlite3.Cursor) -> None:
         """Drops the old model_records, model_metadata, model_tags and tags tables."""
 
-        tables = ["model_records", "model_metadata", "model_tags", "tags"]
+        tables = ["model_config", "model_metadata", "model_tags", "tags"]
 
         for table in tables:
             cursor.execute(f"DROP TABLE IF EXISTS {table};")

invokeai/app/services/shared/sqlite_migrator/migrations/migration_8.py

@@ -0,0 +1,91 @@
+import sqlite3
+from pathlib import Path
+
+from invokeai.app.services.config.config_default import InvokeAIAppConfig
+from invokeai.app.services.shared.sqlite_migrator.sqlite_migrator_common import Migration
+
+
+class Migration8Callback:
+    def __init__(self, app_config: InvokeAIAppConfig) -> None:
+        self._app_config = app_config
+
+    def __call__(self, cursor: sqlite3.Cursor) -> None:
+        self._drop_model_config_table(cursor)
+        self._migrate_abs_models_to_rel(cursor)
+
+    def _drop_model_config_table(self, cursor: sqlite3.Cursor) -> None:
+        """Drops the old model_config table. This was missed in a previous migration."""
+
+        cursor.execute("DROP TABLE IF EXISTS model_config;")
+
+    def _migrate_abs_models_to_rel(self, cursor: sqlite3.Cursor) -> None:
+        """Check all model paths & legacy config paths to determine if they are inside Invoke-managed directories. If
+        they are, update the paths to be relative to the managed directories.
+
+        This migration is a no-op for normal users (their paths will already be relative), but is necessary for users
+        who have been testing the RCs with their live databases. The paths were made absolute in the initial RC, but this
+        change was reverted. To smooth over the revert for our tests, we can migrate the paths back to relative.
+        """
+
+        models_path = self._app_config.models_path
+        legacy_conf_path = self._app_config.legacy_conf_path
+        legacy_conf_dir = self._app_config.legacy_conf_dir
+
+        stmt = """---sql
+        SELECT
+            id,
+            path,
+            json_extract(config, '$.config_path') as config_path
+        FROM models;
+        """
+
+        all_models = cursor.execute(stmt).fetchall()
+
+        for model_id, model_path, model_config_path in all_models:
+            # If the model path is inside the models directory, update it to be relative to the models directory.
+            if Path(model_path).is_relative_to(models_path):
+                new_path = Path(model_path).relative_to(models_path)
+                cursor.execute(
+                    """--sql
+                    UPDATE models
+                    SET config = json_set(config, '$.path', ?)
+                    WHERE id = ?;
+                    """,
+                    (str(new_path), model_id),
+                )
+            # If the model has a legacy config path and it is inside the legacy conf directory, update it to be
+            # relative to the legacy conf directory. This also fixes up cases in which the config path was
+            # incorrectly relativized to the root directory. It will now be relativized to the legacy conf directory.
+            if model_config_path:
+                if Path(model_config_path).is_relative_to(legacy_conf_path):
+                    new_config_path = Path(model_config_path).relative_to(legacy_conf_path)
+                elif Path(model_config_path).is_relative_to(legacy_conf_dir):
+                    new_config_path = Path(*Path(model_config_path).parts[1:])
+                else:
+                    new_config_path = None
+                if new_config_path:
+                    cursor.execute(
+                        """--sql
+                        UPDATE models
+                        SET config = json_set(config, '$.config_path', ?)
+                        WHERE id = ?;
+                        """,
+                        (str(new_config_path), model_id),
+                    )
+
+
+def build_migration_8(app_config: InvokeAIAppConfig) -> Migration:
+    """
+    Build the migration from database version 7 to 8.
+
+    This migration does the following:
+    - Removes the `model_config` table.
+    - Migrates absolute model & legacy config paths to be relative to the models directory.
+    """
+    migration_8 = Migration(
+        from_version=7,
+        to_version=8,
+        callback=Migration8Callback(app_config),
+    )
+
+    return migration_8

invokeai/app/services/shared/sqlite_migrator/migrations/migration_9.py

@@ -0,0 +1,29 @@
+import sqlite3
+
+from invokeai.app.services.shared.sqlite_migrator.sqlite_migrator_common import Migration
+
+
+class Migration9Callback:
+    def __call__(self, cursor: sqlite3.Cursor) -> None:
+        self._empty_session_queue(cursor)
+
+    def _empty_session_queue(self, cursor: sqlite3.Cursor) -> None:
+        """Empties the session queue. This is done to prevent any lingering session queue items from causing pydantic errors due to changed schemas."""
+
+        cursor.execute("DELETE FROM session_queue;")
+
+
+def build_migration_9() -> Migration:
+    """
+    Build the migration from database version 8 to 9.
+
+    This migration does the following:
+    - Empties the session queue. This is done to prevent any lingering session queue items from causing pydantic errors due to changed schemas.
+    """
+    migration_9 = Migration(
+        from_version=8,
+        to_version=9,
+        callback=Migration9Callback(),
+    )
+
+    return migration_9

invokeai/app/services/shared/sqlite_migrator/sqlite_migrator_impl.py

@@ -1,4 +1,6 @@
 import sqlite3
+from contextlib import closing
+from datetime import datetime
 from pathlib import Path
 from typing import Optional
 
@@ -32,6 +34,7 @@ class SqliteMigrator:
         self._db = db
         self._logger = db.logger
         self._migration_set = MigrationSet()
+        self._backup_path: Optional[Path] = None
 
     def register_migration(self, migration: Migration) -> None:
         """Registers a migration."""
@@ -55,6 +58,18 @@ class SqliteMigrator:
                 return False
 
             self._logger.info("Database update needed")
+
+            # Make a backup of the db if it needs to be updated and is a file db
+            if self._db.db_path is not None:
+                timestamp = datetime.now().strftime("%Y%m%d-%H%M%S")
+                self._backup_path = self._db.db_path.parent / f"{self._db.db_path.stem}_backup_{timestamp}.db"
+                self._logger.info(f"Backing up database to {str(self._backup_path)}")
+                # Use SQLite to do the backup
+                with closing(sqlite3.connect(self._backup_path)) as backup_conn:
+                    self._db.conn.backup(backup_conn)
+            else:
+                self._logger.info("Using in-memory database, no backup needed")
+
             next_migration = self._migration_set.get(from_version=self._get_current_version(cursor))
             while next_migration is not None:
                 self._run_migration(next_migration)

invokeai/backend/ip_adapter/ip_adapter.py

@@ -1,22 +1,31 @@
 # copied from https://github.com/tencent-ailab/IP-Adapter (Apache License 2.0)
 #   and modified as needed
 
-from typing import Optional, Union
+import pathlib
+from typing import List, Optional, TypedDict, Union
 
+import safetensors
+import safetensors.torch
 import torch
 from PIL import Image
 from transformers import CLIPImageProcessor, CLIPVisionModelWithProjection
 
 from invokeai.backend.ip_adapter.ip_attention_weights import IPAttentionWeights
 
-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 +34,7 @@ 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.
@@ -45,7 +54,7 @@ class ImageProjModel(torch.nn.Module):
         model.load_state_dict(state_dict)
         return model
 
-    def forward(self, image_embeds):
+    def forward(self, image_embeds: torch.Tensor):
         embeds = image_embeds
         clip_extra_context_tokens = self.proj(embeds).reshape(
             -1, self.clip_extra_context_tokens, self.cross_attention_dim
@@ -57,7 +66,7 @@ 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 +77,7 @@ 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.
@@ -87,21 +96,22 @@ class MLPProjModel(torch.nn.Module):
         model.load_state_dict(state_dict)
         return model
 
-    def forward(self, image_embeds):
+    def forward(self, image_embeds: torch.Tensor):
         clip_extra_context_tokens = self.proj(image_embeds)
         return clip_extra_context_tokens
 
 
-class IPAdapter(RawModel):
+class IPAdapter(torch.nn.Module):
     """IP-Adapter: https://arxiv.org/pdf/2308.06721.pdf"""
 
     def __init__(
         self,
-        state_dict: dict[str, torch.Tensor],
+        state_dict: IPAdapterStateDict,
         device: torch.device,
         dtype: torch.dtype = torch.float16,
         num_tokens: int = 4,
     ):
+        super().__init__()
         self.device = device
         self.dtype = dtype
 
@@ -129,24 +139,27 @@ 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]
+    ) -> Union[ImageProjModel, Resampler, MLPProjModel]:
         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):
-        if isinstance(pil_image, Image.Image):
-            pil_image = [pil_image]
+    def get_image_embeds(self, pil_image: List[Image.Image], image_encoder: CLIPVisionModelWithProjection):
         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)
-        uncond_image_prompt_embeds = self._image_proj_model(torch.zeros_like(clip_image_embeds))
-        return image_prompt_embeds, uncond_image_prompt_embeds
+        try:
+            image_prompt_embeds = self._image_proj_model(clip_image_embeds)
+            uncond_image_prompt_embeds = self._image_proj_model(torch.zeros_like(clip_image_embeds))
+            return image_prompt_embeds, uncond_image_prompt_embeds
+        except RuntimeError as e:
+            raise RuntimeError("Selected CLIP Vision Model is incompatible with the current IP Adapter") from e
 
 
 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]) -> Union[Resampler, MLPProjModel]:
         return Resampler.from_state_dict(
             state_dict=state_dict,
             depth=4,
@@ -157,31 +170,32 @@ class IPAdapterPlus(IPAdapter):
         ).to(self.device, dtype=self.dtype)
 
     @torch.inference_mode()
-    def get_image_embeds(self, pil_image, image_encoder: CLIPVisionModelWithProjection):
-        if isinstance(pil_image, Image.Image):
-            pil_image = [pil_image]
+    def get_image_embeds(self, pil_image: List[Image.Image], image_encoder: CLIPVisionModelWithProjection):
         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]
-        image_prompt_embeds = self._image_proj_model(clip_image_embeds)
         uncond_clip_image_embeds = image_encoder(torch.zeros_like(clip_image), output_hidden_states=True).hidden_states[
             -2
         ]
-        uncond_image_prompt_embeds = self._image_proj_model(uncond_clip_image_embeds)
-        return image_prompt_embeds, uncond_image_prompt_embeds
+        try:
+            image_prompt_embeds = self._image_proj_model(clip_image_embeds)
+            uncond_image_prompt_embeds = self._image_proj_model(uncond_clip_image_embeds)
+            return image_prompt_embeds, uncond_image_prompt_embeds
+        except RuntimeError as e:
+            raise RuntimeError("Selected CLIP Vision Model is incompatible with the current IP Adapter") from e
 
 
 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,
@@ -192,24 +206,48 @@ class IPAdapterPlusXL(IPAdapterPlus):
         ).to(self.device, dtype=self.dtype)
 
 
-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")
+def load_ip_adapter_tensors(ip_adapter_ckpt_path: pathlib.Path, device: str) -> IPAdapterStateDict:
+    state_dict: IPAdapterStateDict = {"ip_adapter": {}, "image_proj": {}}
 
-    if "proj.weight" in state_dict["image_proj"]:  # IPAdapter (with ImageProjModel).
+    if ip_adapter_ckpt_path.suffix == ".safetensors":
+        model = safetensors.torch.load_file(ip_adapter_ckpt_path, device=device)
+        for key in model.keys():
+            if key.startswith("image_proj."):
+                state_dict["image_proj"][key.replace("image_proj.", "")] = model[key]
+            elif key.startswith("ip_adapter."):
+                state_dict["ip_adapter"][key.replace("ip_adapter.", "")] = model[key]
+            else:
+                raise RuntimeError(f"Encountered unexpected IP Adapter state dict key: '{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")
+
+    return state_dict
+
+
+def build_ip_adapter(
+    ip_adapter_ckpt_path: pathlib.Path, device: torch.device, dtype: torch.dtype = torch.float16
+) -> Union[IPAdapter, IPAdapterPlus, IPAdapterPlusXL, IPAdapterPlus]:
+    state_dict = load_ip_adapter_tensors(ip_adapter_ckpt_path, device.type)
+
+    # IPAdapter (with ImageProjModel)
+    if "proj.weight" in state_dict["image_proj"]:
         return IPAdapter(state_dict, device=device, dtype=dtype)
-    elif "proj_in.weight" in state_dict["image_proj"]:  # IPAdaterPlus or IPAdapterPlusXL (with Resampler).
+
+    # IPAdaterPlus or IPAdapterPlusXL (with Resampler)
+    elif "proj_in.weight" in state_dict["image_proj"]:
         cross_attention_dim = state_dict["ip_adapter"]["1.to_k_ip.weight"].shape[-1]
         if cross_attention_dim == 768:
-            # SD1 IP-Adapter Plus
-            return IPAdapterPlus(state_dict, device=device, dtype=dtype)
+            return IPAdapterPlus(state_dict, device=device, dtype=dtype)  # SD1 IP-Adapter Plus
         elif cross_attention_dim == 2048:
-            # SDXL IP-Adapter Plus
-            return IPAdapterPlusXL(state_dict, device=device, dtype=dtype)
+            return IPAdapterPlusXL(state_dict, device=device, dtype=dtype)  # SDXL IP-Adapter Plus
         else:
             raise Exception(f"Unsupported IP-Adapter Plus cross-attention dimension: {cross_attention_dim}.")
-    elif "proj.0.weight" in state_dict["image_proj"]:  # IPAdapterFull (with MLPProjModel).
+
+    # IPAdapterFull (with MLPProjModel)
+    elif "proj.0.weight" in state_dict["image_proj"]:
         return IPAdapterFull(state_dict, device=device, dtype=dtype)
+
+    # Unrecognized IP Adapter Architectures
     else:
         raise ValueError(f"'{ip_adapter_ckpt_path}' has an unrecognized IP-Adapter model architecture.")

invokeai/backend/ip_adapter/resampler.py

@@ -9,8 +9,8 @@ import torch.nn as nn
 
 
 # FFN
-def FeedForward(dim, mult=4):
-    inner_dim = int(dim * mult)
+def FeedForward(dim: int, mult: int = 4):
+    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):
-    bs, length, width = x.shape
+def reshape_tensor(x: torch.Tensor, heads: int):
+    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,
-        depth=8,
-        dim_head=64,
-        heads=16,
-        num_queries=8,
-        embedding_dim=768,
-        output_dim=1024,
-        ff_mult=4,
+        dim: int = 1024,
+        depth: int = 8,
+        dim_head: int = 64,
+        heads: int = 16,
+        num_queries: int = 8,
+        embedding_dim: int = 768,
+        output_dim: int = 1024,
+        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/lora.py

@@ -1,624 +0,0 @@
-# Copyright (c) 2024 The InvokeAI Development team
-"""LoRA model support."""
-
-import bisect
-from pathlib import Path
-from typing import Dict, List, Optional, Tuple, Union
-
-import torch
-from safetensors.torch import load_file
-from typing_extensions import Self
-
-from invokeai.backend.model_manager import BaseModelType
-
-from .raw_model import RawModel
-
-
-class LoRALayerBase:
-    # rank: Optional[int]
-    # alpha: Optional[float]
-    # bias: Optional[torch.Tensor]
-    # layer_key: str
-
-    # @property
-    # def scale(self):
-    #    return self.alpha / self.rank if (self.alpha and self.rank) else 1.0
-
-    def __init__(
-        self,
-        layer_key: str,
-        values: Dict[str, torch.Tensor],
-    ):
-        if "alpha" in values:
-            self.alpha = values["alpha"].item()
-        else:
-            self.alpha = None
-
-        if "bias_indices" in values and "bias_values" in values and "bias_size" in values:
-            self.bias: Optional[torch.Tensor] = torch.sparse_coo_tensor(
-                values["bias_indices"],
-                values["bias_values"],
-                tuple(values["bias_size"]),
-            )
-
-        else:
-            self.bias = None
-
-        self.rank = None  # set in layer implementation
-        self.layer_key = layer_key
-
-    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
-        raise NotImplementedError()
-
-    def calc_size(self) -> int:
-        model_size = 0
-        for val in [self.bias]:
-            if val is not None:
-                model_size += val.nelement() * val.element_size()
-        return model_size
-
-    def to(
-        self,
-        device: Optional[torch.device] = None,
-        dtype: Optional[torch.dtype] = None,
-    ) -> None:
-        if self.bias is not None:
-            self.bias = self.bias.to(device=device, dtype=dtype)
-
-
-# TODO: find and debug lora/locon with bias
-class LoRALayer(LoRALayerBase):
-    # up: torch.Tensor
-    # mid: Optional[torch.Tensor]
-    # down: torch.Tensor
-
-    def __init__(
-        self,
-        layer_key: str,
-        values: Dict[str, torch.Tensor],
-    ):
-        super().__init__(layer_key, values)
-
-        self.up = values["lora_up.weight"]
-        self.down = values["lora_down.weight"]
-        if "lora_mid.weight" in values:
-            self.mid: Optional[torch.Tensor] = values["lora_mid.weight"]
-        else:
-            self.mid = None
-
-        self.rank = self.down.shape[0]
-
-    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
-        if self.mid is not None:
-            up = self.up.reshape(self.up.shape[0], self.up.shape[1])
-            down = self.down.reshape(self.down.shape[0], self.down.shape[1])
-            weight = torch.einsum("m n w h, i m, n j -> i j w h", self.mid, up, down)
-        else:
-            weight = self.up.reshape(self.up.shape[0], -1) @ self.down.reshape(self.down.shape[0], -1)
-
-        return weight
-
-    def calc_size(self) -> int:
-        model_size = super().calc_size()
-        for val in [self.up, self.mid, self.down]:
-            if val is not None:
-                model_size += val.nelement() * val.element_size()
-        return model_size
-
-    def to(
-        self,
-        device: Optional[torch.device] = None,
-        dtype: Optional[torch.dtype] = None,
-    ) -> None:
-        super().to(device=device, dtype=dtype)
-
-        self.up = self.up.to(device=device, dtype=dtype)
-        self.down = self.down.to(device=device, dtype=dtype)
-
-        if self.mid is not None:
-            self.mid = self.mid.to(device=device, dtype=dtype)
-
-
-class LoHALayer(LoRALayerBase):
-    # w1_a: torch.Tensor
-    # w1_b: torch.Tensor
-    # w2_a: torch.Tensor
-    # w2_b: torch.Tensor
-    # t1: Optional[torch.Tensor] = None
-    # t2: Optional[torch.Tensor] = None
-
-    def __init__(self, layer_key: str, values: Dict[str, torch.Tensor]):
-        super().__init__(layer_key, values)
-
-        self.w1_a = values["hada_w1_a"]
-        self.w1_b = values["hada_w1_b"]
-        self.w2_a = values["hada_w2_a"]
-        self.w2_b = values["hada_w2_b"]
-
-        if "hada_t1" in values:
-            self.t1: Optional[torch.Tensor] = values["hada_t1"]
-        else:
-            self.t1 = None
-
-        if "hada_t2" in values:
-            self.t2: Optional[torch.Tensor] = values["hada_t2"]
-        else:
-            self.t2 = None
-
-        self.rank = self.w1_b.shape[0]
-
-    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
-        if self.t1 is None:
-            weight: torch.Tensor = (self.w1_a @ self.w1_b) * (self.w2_a @ self.w2_b)
-
-        else:
-            rebuild1 = torch.einsum("i j k l, j r, i p -> p r k l", self.t1, self.w1_b, self.w1_a)
-            rebuild2 = torch.einsum("i j k l, j r, i p -> p r k l", self.t2, self.w2_b, self.w2_a)
-            weight = rebuild1 * rebuild2
-
-        return weight
-
-    def calc_size(self) -> int:
-        model_size = super().calc_size()
-        for val in [self.w1_a, self.w1_b, self.w2_a, self.w2_b, self.t1, self.t2]:
-            if val is not None:
-                model_size += val.nelement() * val.element_size()
-        return model_size
-
-    def to(
-        self,
-        device: Optional[torch.device] = None,
-        dtype: Optional[torch.dtype] = None,
-    ) -> None:
-        super().to(device=device, dtype=dtype)
-
-        self.w1_a = self.w1_a.to(device=device, dtype=dtype)
-        self.w1_b = self.w1_b.to(device=device, dtype=dtype)
-        if self.t1 is not None:
-            self.t1 = self.t1.to(device=device, dtype=dtype)
-
-        self.w2_a = self.w2_a.to(device=device, dtype=dtype)
-        self.w2_b = self.w2_b.to(device=device, dtype=dtype)
-        if self.t2 is not None:
-            self.t2 = self.t2.to(device=device, dtype=dtype)
-
-
-class LoKRLayer(LoRALayerBase):
-    # w1: Optional[torch.Tensor] = None
-    # w1_a: Optional[torch.Tensor] = None
-    # w1_b: Optional[torch.Tensor] = None
-    # w2: Optional[torch.Tensor] = None
-    # w2_a: Optional[torch.Tensor] = None
-    # w2_b: Optional[torch.Tensor] = None
-    # t2: Optional[torch.Tensor] = None
-
-    def __init__(
-        self,
-        layer_key: str,
-        values: Dict[str, torch.Tensor],
-    ):
-        super().__init__(layer_key, values)
-
-        if "lokr_w1" in values:
-            self.w1: Optional[torch.Tensor] = values["lokr_w1"]
-            self.w1_a = None
-            self.w1_b = None
-        else:
-            self.w1 = None
-            self.w1_a = values["lokr_w1_a"]
-            self.w1_b = values["lokr_w1_b"]
-
-        if "lokr_w2" in values:
-            self.w2: Optional[torch.Tensor] = values["lokr_w2"]
-            self.w2_a = None
-            self.w2_b = None
-        else:
-            self.w2 = None
-            self.w2_a = values["lokr_w2_a"]
-            self.w2_b = values["lokr_w2_b"]
-
-        if "lokr_t2" in values:
-            self.t2: Optional[torch.Tensor] = values["lokr_t2"]
-        else:
-            self.t2 = None
-
-        if "lokr_w1_b" in values:
-            self.rank = values["lokr_w1_b"].shape[0]
-        elif "lokr_w2_b" in values:
-            self.rank = values["lokr_w2_b"].shape[0]
-        else:
-            self.rank = None  # unscaled
-
-    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
-        w1: Optional[torch.Tensor] = self.w1
-        if w1 is None:
-            assert self.w1_a is not None
-            assert self.w1_b is not None
-            w1 = self.w1_a @ self.w1_b
-
-        w2 = self.w2
-        if w2 is None:
-            if self.t2 is None:
-                assert self.w2_a is not None
-                assert self.w2_b is not None
-                w2 = self.w2_a @ self.w2_b
-            else:
-                w2 = torch.einsum("i j k l, i p, j r -> p r k l", self.t2, self.w2_a, self.w2_b)
-
-        if len(w2.shape) == 4:
-            w1 = w1.unsqueeze(2).unsqueeze(2)
-        w2 = w2.contiguous()
-        assert w1 is not None
-        assert w2 is not None
-        weight = torch.kron(w1, w2)
-
-        return weight
-
-    def calc_size(self) -> int:
-        model_size = super().calc_size()
-        for val in [self.w1, self.w1_a, self.w1_b, self.w2, self.w2_a, self.w2_b, self.t2]:
-            if val is not None:
-                model_size += val.nelement() * val.element_size()
-        return model_size
-
-    def to(
-        self,
-        device: Optional[torch.device] = None,
-        dtype: Optional[torch.dtype] = None,
-    ) -> None:
-        super().to(device=device, dtype=dtype)
-
-        if self.w1 is not None:
-            self.w1 = self.w1.to(device=device, dtype=dtype)
-        else:
-            assert self.w1_a is not None
-            assert self.w1_b is not None
-            self.w1_a = self.w1_a.to(device=device, dtype=dtype)
-            self.w1_b = self.w1_b.to(device=device, dtype=dtype)
-
-        if self.w2 is not None:
-            self.w2 = self.w2.to(device=device, dtype=dtype)
-        else:
-            assert self.w2_a is not None
-            assert self.w2_b is not None
-            self.w2_a = self.w2_a.to(device=device, dtype=dtype)
-            self.w2_b = self.w2_b.to(device=device, dtype=dtype)
-
-        if self.t2 is not None:
-            self.t2 = self.t2.to(device=device, dtype=dtype)
-
-
-class FullLayer(LoRALayerBase):
-    # weight: torch.Tensor
-
-    def __init__(
-        self,
-        layer_key: str,
-        values: Dict[str, torch.Tensor],
-    ):
-        super().__init__(layer_key, values)
-
-        self.weight = values["diff"]
-
-        if len(values.keys()) > 1:
-            _keys = list(values.keys())
-            _keys.remove("diff")
-            raise NotImplementedError(f"Unexpected keys in lora diff layer: {_keys}")
-
-        self.rank = None  # unscaled
-
-    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
-        return self.weight
-
-    def calc_size(self) -> int:
-        model_size = super().calc_size()
-        model_size += self.weight.nelement() * self.weight.element_size()
-        return model_size
-
-    def to(
-        self,
-        device: Optional[torch.device] = None,
-        dtype: Optional[torch.dtype] = None,
-    ) -> None:
-        super().to(device=device, dtype=dtype)
-
-        self.weight = self.weight.to(device=device, dtype=dtype)
-
-
-class IA3Layer(LoRALayerBase):
-    # weight: torch.Tensor
-    # on_input: torch.Tensor
-
-    def __init__(
-        self,
-        layer_key: str,
-        values: Dict[str, torch.Tensor],
-    ):
-        super().__init__(layer_key, values)
-
-        self.weight = values["weight"]
-        self.on_input = values["on_input"]
-
-        self.rank = None  # unscaled
-
-    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
-        weight = self.weight
-        if not self.on_input:
-            weight = weight.reshape(-1, 1)
-        assert orig_weight is not None
-        return orig_weight * weight
-
-    def calc_size(self) -> int:
-        model_size = super().calc_size()
-        model_size += self.weight.nelement() * self.weight.element_size()
-        model_size += self.on_input.nelement() * self.on_input.element_size()
-        return model_size
-
-    def to(
-        self,
-        device: Optional[torch.device] = None,
-        dtype: Optional[torch.dtype] = None,
-    ):
-        super().to(device=device, dtype=dtype)
-
-        self.weight = self.weight.to(device=device, dtype=dtype)
-        self.on_input = self.on_input.to(device=device, dtype=dtype)
-
-
-AnyLoRALayer = Union[LoRALayer, LoHALayer, LoKRLayer, FullLayer, IA3Layer]
-
-
-class LoRAModelRaw(RawModel):  # (torch.nn.Module):
-    _name: str
-    layers: Dict[str, AnyLoRALayer]
-
-    def __init__(
-        self,
-        name: str,
-        layers: Dict[str, AnyLoRALayer],
-    ):
-        self._name = name
-        self.layers = layers
-
-    @property
-    def name(self) -> str:
-        return self._name
-
-    def to(
-        self,
-        device: Optional[torch.device] = None,
-        dtype: Optional[torch.dtype] = None,
-    ) -> None:
-        # TODO: try revert if exception?
-        for _key, layer in self.layers.items():
-            layer.to(device=device, dtype=dtype)
-
-    def calc_size(self) -> int:
-        model_size = 0
-        for _, layer in self.layers.items():
-            model_size += layer.calc_size()
-        return model_size
-
-    @classmethod
-    def _convert_sdxl_keys_to_diffusers_format(cls, state_dict: Dict[str, torch.Tensor]) -> Dict[str, torch.Tensor]:
-        """Convert the keys of an SDXL LoRA state_dict to diffusers format.
-
-        The input state_dict can be in either Stability AI format or diffusers format. If the state_dict is already in
-        diffusers format, then this function will have no effect.
-
-        This function is adapted from:
-        https://github.com/bmaltais/kohya_ss/blob/2accb1305979ba62f5077a23aabac23b4c37e935/networks/lora_diffusers.py#L385-L409
-
-        Args:
-            state_dict (Dict[str, Tensor]): The SDXL LoRA state_dict.
-
-        Raises:
-            ValueError: If state_dict contains an unrecognized key, or not all keys could be converted.
-
-        Returns:
-            Dict[str, Tensor]: The diffusers-format state_dict.
-        """
-        converted_count = 0  # The number of Stability AI keys converted to diffusers format.
-        not_converted_count = 0  # The number of keys that were not converted.
-
-        # Get a sorted list of Stability AI UNet keys so that we can efficiently search for keys with matching prefixes.
-        # For example, we want to efficiently find `input_blocks_4_1` in the list when searching for
-        # `input_blocks_4_1_proj_in`.
-        stability_unet_keys = list(SDXL_UNET_STABILITY_TO_DIFFUSERS_MAP)
-        stability_unet_keys.sort()
-
-        new_state_dict = {}
-        for full_key, value in state_dict.items():
-            if full_key.startswith("lora_unet_"):
-                search_key = full_key.replace("lora_unet_", "")
-                # Use bisect to find the key in stability_unet_keys that *may* match the search_key's prefix.
-                position = bisect.bisect_right(stability_unet_keys, search_key)
-                map_key = stability_unet_keys[position - 1]
-                # Now, check if the map_key *actually* matches the search_key.
-                if search_key.startswith(map_key):
-                    new_key = full_key.replace(map_key, SDXL_UNET_STABILITY_TO_DIFFUSERS_MAP[map_key])
-                    new_state_dict[new_key] = value
-                    converted_count += 1
-                else:
-                    new_state_dict[full_key] = value
-                    not_converted_count += 1
-            elif full_key.startswith("lora_te1_") or full_key.startswith("lora_te2_"):
-                # The CLIP text encoders have the same keys in both Stability AI and diffusers formats.
-                new_state_dict[full_key] = value
-                continue
-            else:
-                raise ValueError(f"Unrecognized SDXL LoRA key prefix: '{full_key}'.")
-
-        if converted_count > 0 and not_converted_count > 0:
-            raise ValueError(
-                f"The SDXL LoRA could only be partially converted to diffusers format. converted={converted_count},"
-                f" not_converted={not_converted_count}"
-            )
-
-        return new_state_dict
-
-    @classmethod
-    def from_checkpoint(
-        cls,
-        file_path: Union[str, Path],
-        device: Optional[torch.device] = None,
-        dtype: Optional[torch.dtype] = None,
-        base_model: Optional[BaseModelType] = None,
-    ) -> Self:
-        device = device or torch.device("cpu")
-        dtype = dtype or torch.float32
-
-        if isinstance(file_path, str):
-            file_path = Path(file_path)
-
-        model = cls(
-            name=file_path.stem,
-            layers={},
-        )
-
-        if file_path.suffix == ".safetensors":
-            sd = load_file(file_path.absolute().as_posix(), device="cpu")
-        else:
-            sd = torch.load(file_path, map_location="cpu")
-
-        state_dict = cls._group_state(sd)
-
-        if base_model == BaseModelType.StableDiffusionXL:
-            state_dict = cls._convert_sdxl_keys_to_diffusers_format(state_dict)
-
-        for layer_key, values in state_dict.items():
-            # lora and locon
-            if "lora_down.weight" in values:
-                layer: AnyLoRALayer = LoRALayer(layer_key, values)
-
-            # loha
-            elif "hada_w1_b" in values:
-                layer = LoHALayer(layer_key, values)
-
-            # lokr
-            elif "lokr_w1_b" in values or "lokr_w1" in values:
-                layer = LoKRLayer(layer_key, values)
-
-            # diff
-            elif "diff" in values:
-                layer = FullLayer(layer_key, values)
-
-            # ia3
-            elif "weight" in values and "on_input" in values:
-                layer = IA3Layer(layer_key, values)
-
-            else:
-                print(f">> Encountered unknown lora layer module in {model.name}: {layer_key} - {list(values.keys())}")
-                raise Exception("Unknown lora format!")
-
-            # lower memory consumption by removing already parsed layer values
-            state_dict[layer_key].clear()
-
-            layer.to(device=device, dtype=dtype)
-            model.layers[layer_key] = layer
-
-        return model
-
-    @staticmethod
-    def _group_state(state_dict: Dict[str, torch.Tensor]) -> Dict[str, Dict[str, torch.Tensor]]:
-        state_dict_groupped: Dict[str, Dict[str, torch.Tensor]] = {}
-
-        for key, value in state_dict.items():
-            stem, leaf = key.split(".", 1)
-            if stem not in state_dict_groupped:
-                state_dict_groupped[stem] = {}
-            state_dict_groupped[stem][leaf] = value
-
-        return state_dict_groupped
-
-
-# code from
-# https://github.com/bmaltais/kohya_ss/blob/2accb1305979ba62f5077a23aabac23b4c37e935/networks/lora_diffusers.py#L15C1-L97C32
-def make_sdxl_unet_conversion_map() -> List[Tuple[str, str]]:
-    """Create a dict mapping state_dict keys from Stability AI SDXL format to diffusers SDXL format."""
-    unet_conversion_map_layer = []
-
-    for i in range(3):  # num_blocks is 3 in sdxl
-        # loop over downblocks/upblocks
-        for j in range(2):
-            # loop over resnets/attentions for downblocks
-            hf_down_res_prefix = f"down_blocks.{i}.resnets.{j}."
-            sd_down_res_prefix = f"input_blocks.{3*i + j + 1}.0."
-            unet_conversion_map_layer.append((sd_down_res_prefix, hf_down_res_prefix))
-
-            if i < 3:
-                # no attention layers in down_blocks.3
-                hf_down_atn_prefix = f"down_blocks.{i}.attentions.{j}."
-                sd_down_atn_prefix = f"input_blocks.{3*i + j + 1}.1."
-                unet_conversion_map_layer.append((sd_down_atn_prefix, hf_down_atn_prefix))
-
-        for j in range(3):
-            # loop over resnets/attentions for upblocks
-            hf_up_res_prefix = f"up_blocks.{i}.resnets.{j}."
-            sd_up_res_prefix = f"output_blocks.{3*i + j}.0."
-            unet_conversion_map_layer.append((sd_up_res_prefix, hf_up_res_prefix))
-
-            # if i > 0: commentout for sdxl
-            # no attention layers in up_blocks.0
-            hf_up_atn_prefix = f"up_blocks.{i}.attentions.{j}."
-            sd_up_atn_prefix = f"output_blocks.{3*i + j}.1."
-            unet_conversion_map_layer.append((sd_up_atn_prefix, hf_up_atn_prefix))
-
-        if i < 3:
-            # no downsample in down_blocks.3
-            hf_downsample_prefix = f"down_blocks.{i}.downsamplers.0.conv."
-            sd_downsample_prefix = f"input_blocks.{3*(i+1)}.0.op."
-            unet_conversion_map_layer.append((sd_downsample_prefix, hf_downsample_prefix))
-
-            # no upsample in up_blocks.3
-            hf_upsample_prefix = f"up_blocks.{i}.upsamplers.0."
-            sd_upsample_prefix = f"output_blocks.{3*i + 2}.{2}."  # change for sdxl
-            unet_conversion_map_layer.append((sd_upsample_prefix, hf_upsample_prefix))
-
-    hf_mid_atn_prefix = "mid_block.attentions.0."
-    sd_mid_atn_prefix = "middle_block.1."
-    unet_conversion_map_layer.append((sd_mid_atn_prefix, hf_mid_atn_prefix))
-
-    for j in range(2):
-        hf_mid_res_prefix = f"mid_block.resnets.{j}."
-        sd_mid_res_prefix = f"middle_block.{2*j}."
-        unet_conversion_map_layer.append((sd_mid_res_prefix, hf_mid_res_prefix))
-
-    unet_conversion_map_resnet = [
-        # (stable-diffusion, HF Diffusers)
-        ("in_layers.0.", "norm1."),
-        ("in_layers.2.", "conv1."),
-        ("out_layers.0.", "norm2."),
-        ("out_layers.3.", "conv2."),
-        ("emb_layers.1.", "time_emb_proj."),
-        ("skip_connection.", "conv_shortcut."),
-    ]
-
-    unet_conversion_map = []
-    for sd, hf in unet_conversion_map_layer:
-        if "resnets" in hf:
-            for sd_res, hf_res in unet_conversion_map_resnet:
-                unet_conversion_map.append((sd + sd_res, hf + hf_res))
-        else:
-            unet_conversion_map.append((sd, hf))
-
-    for j in range(2):
-        hf_time_embed_prefix = f"time_embedding.linear_{j+1}."
-        sd_time_embed_prefix = f"time_embed.{j*2}."
-        unet_conversion_map.append((sd_time_embed_prefix, hf_time_embed_prefix))
-
-    for j in range(2):
-        hf_label_embed_prefix = f"add_embedding.linear_{j+1}."
-        sd_label_embed_prefix = f"label_emb.0.{j*2}."
-        unet_conversion_map.append((sd_label_embed_prefix, hf_label_embed_prefix))
-
-    unet_conversion_map.append(("input_blocks.0.0.", "conv_in."))
-    unet_conversion_map.append(("out.0.", "conv_norm_out."))
-    unet_conversion_map.append(("out.2.", "conv_out."))
-
-    return unet_conversion_map
-
-
-SDXL_UNET_STABILITY_TO_DIFFUSERS_MAP = {
-    sd.rstrip(".").replace(".", "_"): hf.rstrip(".").replace(".", "_") for sd, hf in make_sdxl_unet_conversion_map()
-}

invokeai/backend/lora/full_layer.py

@@ -0,0 +1,42 @@
+from typing import Dict, Optional
+
+import torch
+
+from invokeai.backend.lora.lora_layer_base import LoRALayerBase
+
+
+class FullLayer(LoRALayerBase):
+    # weight: torch.Tensor
+
+    def __init__(
+        self,
+        layer_key: str,
+        values: Dict[str, torch.Tensor],
+    ):
+        super().__init__(layer_key, values)
+
+        self.weight = values["diff"]
+
+        if len(values.keys()) > 1:
+            _keys = list(values.keys())
+            _keys.remove("diff")
+            raise NotImplementedError(f"Unexpected keys in lora diff layer: {_keys}")
+
+        self.rank = None  # unscaled
+
+    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
+        return self.weight
+
+    def calc_size(self) -> int:
+        model_size = super().calc_size()
+        model_size += self.weight.nelement() * self.weight.element_size()
+        return model_size
+
+    def to(
+        self,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ) -> None:
+        super().to(device=device, dtype=dtype)
+
+        self.weight = self.weight.to(device=device, dtype=dtype)

invokeai/backend/lora/ia3_layer.py

@@ -0,0 +1,45 @@
+from typing import Dict, Optional
+
+import torch
+
+from invokeai.backend.lora.lora_layer_base import LoRALayerBase
+
+
+class IA3Layer(LoRALayerBase):
+    # weight: torch.Tensor
+    # on_input: torch.Tensor
+
+    def __init__(
+        self,
+        layer_key: str,
+        values: Dict[str, torch.Tensor],
+    ):
+        super().__init__(layer_key, values)
+
+        self.weight = values["weight"]
+        self.on_input = values["on_input"]
+
+        self.rank = None  # unscaled
+
+    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
+        weight = self.weight
+        if not self.on_input:
+            weight = weight.reshape(-1, 1)
+        assert orig_weight is not None
+        return orig_weight * weight
+
+    def calc_size(self) -> int:
+        model_size = super().calc_size()
+        model_size += self.weight.nelement() * self.weight.element_size()
+        model_size += self.on_input.nelement() * self.on_input.element_size()
+        return model_size
+
+    def to(
+        self,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ):
+        super().to(device=device, dtype=dtype)
+
+        self.weight = self.weight.to(device=device, dtype=dtype)
+        self.on_input = self.on_input.to(device=device, dtype=dtype)

invokeai/backend/lora/loha_layer.py

@@ -0,0 +1,69 @@
+from typing import Dict, Optional
+
+import torch
+
+from invokeai.backend.lora.lora_layer_base import LoRALayerBase
+
+
+class LoHALayer(LoRALayerBase):
+    # w1_a: torch.Tensor
+    # w1_b: torch.Tensor
+    # w2_a: torch.Tensor
+    # w2_b: torch.Tensor
+    # t1: Optional[torch.Tensor] = None
+    # t2: Optional[torch.Tensor] = None
+
+    def __init__(self, layer_key: str, values: Dict[str, torch.Tensor]):
+        super().__init__(layer_key, values)
+
+        self.w1_a = values["hada_w1_a"]
+        self.w1_b = values["hada_w1_b"]
+        self.w2_a = values["hada_w2_a"]
+        self.w2_b = values["hada_w2_b"]
+
+        if "hada_t1" in values:
+            self.t1: Optional[torch.Tensor] = values["hada_t1"]
+        else:
+            self.t1 = None
+
+        if "hada_t2" in values:
+            self.t2: Optional[torch.Tensor] = values["hada_t2"]
+        else:
+            self.t2 = None
+
+        self.rank = self.w1_b.shape[0]
+
+    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
+        if self.t1 is None:
+            weight: torch.Tensor = (self.w1_a @ self.w1_b) * (self.w2_a @ self.w2_b)
+
+        else:
+            rebuild1 = torch.einsum("i j k l, j r, i p -> p r k l", self.t1, self.w1_b, self.w1_a)
+            rebuild2 = torch.einsum("i j k l, j r, i p -> p r k l", self.t2, self.w2_b, self.w2_a)
+            weight = rebuild1 * rebuild2
+
+        return weight
+
+    def calc_size(self) -> int:
+        model_size = super().calc_size()
+        for val in [self.w1_a, self.w1_b, self.w2_a, self.w2_b, self.t1, self.t2]:
+            if val is not None:
+                model_size += val.nelement() * val.element_size()
+        return model_size
+
+    def to(
+        self,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ) -> None:
+        super().to(device=device, dtype=dtype)
+
+        self.w1_a = self.w1_a.to(device=device, dtype=dtype)
+        self.w1_b = self.w1_b.to(device=device, dtype=dtype)
+        if self.t1 is not None:
+            self.t1 = self.t1.to(device=device, dtype=dtype)
+
+        self.w2_a = self.w2_a.to(device=device, dtype=dtype)
+        self.w2_b = self.w2_b.to(device=device, dtype=dtype)
+        if self.t2 is not None:
+            self.t2 = self.t2.to(device=device, dtype=dtype)

invokeai/backend/lora/lokr_layer.py

@@ -0,0 +1,110 @@
+from typing import Dict, Optional
+
+import torch
+
+from invokeai.backend.lora.lora_layer_base import LoRALayerBase
+
+
+class LoKRLayer(LoRALayerBase):
+    # w1: Optional[torch.Tensor] = None
+    # w1_a: Optional[torch.Tensor] = None
+    # w1_b: Optional[torch.Tensor] = None
+    # w2: Optional[torch.Tensor] = None
+    # w2_a: Optional[torch.Tensor] = None
+    # w2_b: Optional[torch.Tensor] = None
+    # t2: Optional[torch.Tensor] = None
+
+    def __init__(
+        self,
+        layer_key: str,
+        values: Dict[str, torch.Tensor],
+    ):
+        super().__init__(layer_key, values)
+
+        if "lokr_w1" in values:
+            self.w1: Optional[torch.Tensor] = values["lokr_w1"]
+            self.w1_a = None
+            self.w1_b = None
+        else:
+            self.w1 = None
+            self.w1_a = values["lokr_w1_a"]
+            self.w1_b = values["lokr_w1_b"]
+
+        if "lokr_w2" in values:
+            self.w2: Optional[torch.Tensor] = values["lokr_w2"]
+            self.w2_a = None
+            self.w2_b = None
+        else:
+            self.w2 = None
+            self.w2_a = values["lokr_w2_a"]
+            self.w2_b = values["lokr_w2_b"]
+
+        if "lokr_t2" in values:
+            self.t2: Optional[torch.Tensor] = values["lokr_t2"]
+        else:
+            self.t2 = None
+
+        if "lokr_w1_b" in values:
+            self.rank = values["lokr_w1_b"].shape[0]
+        elif "lokr_w2_b" in values:
+            self.rank = values["lokr_w2_b"].shape[0]
+        else:
+            self.rank = None  # unscaled
+
+    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
+        w1: Optional[torch.Tensor] = self.w1
+        if w1 is None:
+            assert self.w1_a is not None
+            assert self.w1_b is not None
+            w1 = self.w1_a @ self.w1_b
+
+        w2 = self.w2
+        if w2 is None:
+            if self.t2 is None:
+                assert self.w2_a is not None
+                assert self.w2_b is not None
+                w2 = self.w2_a @ self.w2_b
+            else:
+                w2 = torch.einsum("i j k l, i p, j r -> p r k l", self.t2, self.w2_a, self.w2_b)
+
+        if len(w2.shape) == 4:
+            w1 = w1.unsqueeze(2).unsqueeze(2)
+        w2 = w2.contiguous()
+        assert w1 is not None
+        assert w2 is not None
+        weight = torch.kron(w1, w2)
+
+        return weight
+
+    def calc_size(self) -> int:
+        model_size = super().calc_size()
+        for val in [self.w1, self.w1_a, self.w1_b, self.w2, self.w2_a, self.w2_b, self.t2]:
+            if val is not None:
+                model_size += val.nelement() * val.element_size()
+        return model_size
+
+    def to(
+        self,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ) -> None:
+        super().to(device=device, dtype=dtype)
+
+        if self.w1 is not None:
+            self.w1 = self.w1.to(device=device, dtype=dtype)
+        else:
+            assert self.w1_a is not None
+            assert self.w1_b is not None
+            self.w1_a = self.w1_a.to(device=device, dtype=dtype)
+            self.w1_b = self.w1_b.to(device=device, dtype=dtype)
+
+        if self.w2 is not None:
+            self.w2 = self.w2.to(device=device, dtype=dtype)
+        else:
+            assert self.w2_a is not None
+            assert self.w2_b is not None
+            self.w2_a = self.w2_a.to(device=device, dtype=dtype)
+            self.w2_b = self.w2_b.to(device=device, dtype=dtype)
+
+        if self.t2 is not None:
+            self.t2 = self.t2.to(device=device, dtype=dtype)

invokeai/backend/lora/lora_layer.py

@@ -0,0 +1,81 @@
+from typing import Optional
+
+import torch
+
+from invokeai.backend.lora.lora_layer_base import LoRALayerBase
+
+
+class LoRALayer(LoRALayerBase):
+    def __init__(
+        self,
+        layer_key: str,
+        values: dict[str, torch.Tensor],
+    ):
+        super().__init__(layer_key, values)
+
+        self.up = values["lora_up.weight"]
+        self.down = values["lora_down.weight"]
+
+        self.mid: Optional[torch.Tensor] = values.get("lora_mid.weight", None)
+        self.dora_scale: Optional[torch.Tensor] = values.get("dora_scale", None)
+        self.rank = self.down.shape[0]
+
+    def _apply_dora(self, orig_weight: torch.Tensor, lora_weight: torch.Tensor) -> torch.Tensor:
+        """Apply DoRA to the weight matrix.
+
+        This function is based roughly on the reference implementation in PEFT, but handles scaling in a slightly
+        different way:
+        https://github.com/huggingface/peft/blob/26726bf1ddee6ca75ed4e1bfd292094526707a78/src/peft/tuners/lora/layer.py#L421-L433
+
+        """
+        # Merge the original weight with the LoRA weight.
+        merged_weight = orig_weight + lora_weight
+
+        # Calculate the vector-wise L2 norm of the weight matrix across each column vector.
+        weight_norm: torch.Tensor = torch.linalg.norm(merged_weight, dim=1)
+
+        dora_factor = self.dora_scale / weight_norm
+        new_weight = dora_factor * merged_weight
+
+        # TODO(ryand): This is wasteful. We already have the final weight, but we calculate the diff, because that is
+        # what the `get_weight()` API is expected to return. If we do refactor this, we'll have to give some thought to
+        # how lora weight scaling should be applied - having the full weight diff makes this easy.
+        weight_diff = new_weight - orig_weight
+        return weight_diff
+
+    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
+        if self.mid is not None:
+            up = self.up.reshape(self.up.shape[0], self.up.shape[1])
+            down = self.down.reshape(self.down.shape[0], self.down.shape[1])
+            weight = torch.einsum("m n w h, i m, n j -> i j w h", self.mid, up, down)
+        else:
+            weight = self.up.reshape(self.up.shape[0], -1) @ self.down.reshape(self.down.shape[0], -1)
+
+        if self.dora_scale is not None:
+            assert orig_weight is not None
+            weight = self._apply_dora(orig_weight, weight)
+
+        return weight
+
+    def calc_size(self) -> int:
+        model_size = super().calc_size()
+        for val in [self.up, self.mid, self.down]:
+            if val is not None:
+                model_size += val.nelement() * val.element_size()
+        return model_size
+
+    def to(
+        self,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ) -> None:
+        super().to(device=device, dtype=dtype)
+
+        self.up = self.up.to(device=device, dtype=dtype)
+        self.down = self.down.to(device=device, dtype=dtype)
+
+        if self.mid is not None:
+            self.mid = self.mid.to(device=device, dtype=dtype)
+
+        if self.dora_scale is not None:
+            self.dora_scale = self.dora_scale.to(device=device, dtype=dtype)

invokeai/backend/lora/lora_layer_base.py

@@ -0,0 +1,55 @@
+from typing import Dict, Optional
+
+import torch
+
+
+class LoRALayerBase:
+    # rank: Optional[int]
+    # alpha: Optional[float]
+    # bias: Optional[torch.Tensor]
+    # layer_key: str
+
+    # @property
+    # def scale(self):
+    #    return self.alpha / self.rank if (self.alpha and self.rank) else 1.0
+
+    def __init__(
+        self,
+        layer_key: str,
+        values: Dict[str, torch.Tensor],
+    ):
+        if "alpha" in values:
+            self.alpha = values["alpha"].item()
+        else:
+            self.alpha = None
+
+        if "bias_indices" in values and "bias_values" in values and "bias_size" in values:
+            self.bias: Optional[torch.Tensor] = torch.sparse_coo_tensor(
+                values["bias_indices"],
+                values["bias_values"],
+                tuple(values["bias_size"]),
+            )
+
+        else:
+            self.bias = None
+
+        self.rank = None  # set in layer implementation
+        self.layer_key = layer_key
+
+    def get_weight(self, orig_weight: Optional[torch.Tensor]) -> torch.Tensor:
+        raise NotImplementedError()
+
+    def calc_size(self) -> int:
+        model_size = 0
+        for val in [self.bias]:
+            if val is not None:
+                model_size += val.nelement() * val.element_size()
+        return model_size
+
+    def to(
+        self,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ) -> None:
+        if self.bias is not None:
+            self.bias = self.bias.to(device=device, dtype=dtype)

invokeai/backend/lora/lora_model.py

@@ -0,0 +1,111 @@
+from pathlib import Path
+from typing import Optional, Union
+
+import torch
+
+from invokeai.backend.lora.full_layer import FullLayer
+from invokeai.backend.lora.ia3_layer import IA3Layer
+from invokeai.backend.lora.loha_layer import LoHALayer
+from invokeai.backend.lora.lokr_layer import LoKRLayer
+from invokeai.backend.lora.lora_layer import LoRALayer
+from invokeai.backend.lora.sdxl_state_dict_utils import convert_sdxl_keys_to_diffusers_format
+from invokeai.backend.model_manager import BaseModelType
+from invokeai.backend.util.serialization import load_state_dict
+
+AnyLoRALayer = Union[LoRALayer, LoHALayer, LoKRLayer, FullLayer, IA3Layer]
+
+
+class LoRAModelRaw(torch.nn.Module):
+    def __init__(
+        self,
+        name: str,
+        layers: dict[str, AnyLoRALayer],
+    ):
+        super().__init__()
+        self._name = name
+        self.layers = layers
+
+    @property
+    def name(self) -> str:
+        return self._name
+
+    def to(
+        self,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ) -> None:
+        # TODO: try revert if exception?
+        for _key, layer in self.layers.items():
+            layer.to(device=device, dtype=dtype)
+
+    def calc_size(self) -> int:
+        model_size = 0
+        for _, layer in self.layers.items():
+            model_size += layer.calc_size()
+        return model_size
+
+    @classmethod
+    def from_checkpoint(
+        cls,
+        file_path: Union[str, Path],
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+        base_model: Optional[BaseModelType] = None,
+    ):
+        device = device or torch.device("cpu")
+        dtype = dtype or torch.float32
+
+        file_path = Path(file_path)
+
+        model_name = file_path.stem
+
+        sd = load_state_dict(file_path, device=str(device))
+        state_dict = cls._group_state(sd)
+
+        if base_model == BaseModelType.StableDiffusionXL:
+            state_dict = convert_sdxl_keys_to_diffusers_format(state_dict)
+
+        layers: dict[str, AnyLoRALayer] = {}
+        for layer_key, values in state_dict.items():
+            # lora and locon
+            if "lora_down.weight" in values:
+                layer: AnyLoRALayer = LoRALayer(layer_key, values)
+
+            # loha
+            elif "hada_w1_b" in values:
+                layer = LoHALayer(layer_key, values)
+
+            # lokr
+            elif "lokr_w1_b" in values or "lokr_w1" in values:
+                layer = LoKRLayer(layer_key, values)
+
+            # diff
+            elif "diff" in values:
+                layer = FullLayer(layer_key, values)
+
+            # ia3
+            elif "weight" in values and "on_input" in values:
+                layer = IA3Layer(layer_key, values)
+
+            else:
+                raise ValueError(f"Unknown lora layer module in {model_name}: {layer_key}: {list(values.keys())}")
+
+            # lower memory consumption by removing already parsed layer values
+            state_dict[layer_key].clear()
+
+            layer.to(device=device, dtype=dtype)
+            layers[layer_key] = layer
+
+        return cls(name=model_name, layers=layers)
+
+    @staticmethod
+    def _group_state(state_dict: dict[str, torch.Tensor]) -> dict[str, dict[str, torch.Tensor]]:
+        state_dict_groupped: dict[str, dict[str, torch.Tensor]] = {}
+
+        for key, value in state_dict.items():
+            stem, leaf = key.split(".", 1)
+            if stem not in state_dict_groupped:
+                state_dict_groupped[stem] = {}
+            state_dict_groupped[stem][leaf] = value
+
+        return state_dict_groupped

invokeai/backend/lora/lora_model_patcher.py

@@ -0,0 +1,137 @@
+from contextlib import contextmanager
+from typing import Iterator, Tuple
+
+import torch
+from diffusers.models.unets.unet_2d_condition import UNet2DConditionModel
+from transformers import CLIPTextModel
+
+from invokeai.backend.lora.lora_model import LoRAModelRaw
+from invokeai.backend.model_manager.any_model_type import AnyModel
+
+
+class LoraModelPatcher:
+    @staticmethod
+    def _resolve_lora_key(model: torch.nn.Module, lora_key: str, prefix: str) -> Tuple[str, torch.nn.Module]:
+        assert "." not in lora_key
+
+        if not lora_key.startswith(prefix):
+            raise Exception(f"lora_key with invalid prefix: {lora_key}, {prefix}")
+
+        module = model
+        module_key = ""
+        key_parts = lora_key[len(prefix) :].split("_")
+
+        submodule_name = key_parts.pop(0)
+
+        while len(key_parts) > 0:
+            try:
+                module = module.get_submodule(submodule_name)
+                module_key += "." + submodule_name
+                submodule_name = key_parts.pop(0)
+            except Exception:
+                submodule_name += "_" + key_parts.pop(0)
+
+        module = module.get_submodule(submodule_name)
+        module_key = (module_key + "." + submodule_name).lstrip(".")
+
+        return (module_key, module)
+
+    @classmethod
+    @contextmanager
+    def apply_lora_unet(
+        cls,
+        unet: UNet2DConditionModel,
+        loras: Iterator[Tuple[LoRAModelRaw, float]],
+    ):
+        with cls.apply_lora(unet, loras, "lora_unet_"):
+            yield
+
+    @classmethod
+    @contextmanager
+    def apply_lora_text_encoder(
+        cls,
+        text_encoder: CLIPTextModel,
+        loras: Iterator[Tuple[LoRAModelRaw, float]],
+    ):
+        with cls.apply_lora(text_encoder, loras, "lora_te_"):
+            yield
+
+    @classmethod
+    @contextmanager
+    def apply_sdxl_lora_text_encoder(
+        cls,
+        text_encoder: CLIPTextModel,
+        loras: Iterator[Tuple[LoRAModelRaw, float]],
+    ):
+        with cls.apply_lora(text_encoder, loras, "lora_te1_"):
+            yield
+
+    @classmethod
+    @contextmanager
+    def apply_sdxl_lora_text_encoder2(
+        cls,
+        text_encoder: CLIPTextModel,
+        loras: Iterator[Tuple[LoRAModelRaw, float]],
+    ):
+        with cls.apply_lora(text_encoder, loras, "lora_te2_"):
+            yield
+
+    @classmethod
+    @contextmanager
+    def apply_lora(
+        cls,
+        model: AnyModel,
+        loras: Iterator[Tuple[LoRAModelRaw, float]],
+        prefix: str,
+    ):
+        original_weights = {}
+        try:
+            with torch.no_grad():
+                for lora, lora_weight in loras:
+                    # assert lora.device.type == "cpu"
+                    for layer_key, layer in lora.layers.items():
+                        if not layer_key.startswith(prefix):
+                            continue
+
+                        # TODO(ryand): A non-negligible amount of time is currently spent resolving LoRA keys. This
+                        # should be improved in the following ways:
+                        # 1. The key mapping could be more-efficiently pre-computed. This would save time every time a
+                        #    LoRA model is applied.
+                        # 2. From an API perspective, there's no reason that the `LoraModelPatcher` should be aware of
+                        #    the intricacies of Stable Diffusion key resolution. It should just expect the input LoRA
+                        #    weights to have valid keys.
+                        assert isinstance(model, torch.nn.Module)
+                        module_key, module = cls._resolve_lora_key(model, layer_key, prefix)
+
+                        # All of the LoRA weight calculations will be done on the same device as the module weight.
+                        # (Performance will be best if this is a CUDA device.)
+                        device = module.weight.device
+                        dtype = module.weight.dtype
+
+                        if module_key not in original_weights:
+                            original_weights[module_key] = module.weight.detach().to(device="cpu", copy=True)
+
+                        layer_scale = layer.alpha / layer.rank if (layer.alpha and layer.rank) else 1.0
+
+                        # We intentionally move to the target device first, then cast. Experimentally, this was found to
+                        # be significantly faster for 16-bit CPU tensors being moved to a CUDA device than doing the
+                        # same thing in a single call to '.to(...)'.
+                        layer.to(device=device)
+                        layer.to(dtype=torch.float32)
+                        # TODO(ryand): Using torch.autocast(...) over explicit casting may offer a speed benefit on CUDA
+                        # devices here. Experimentally, it was found to be very slow on CPU. More investigation needed.
+                        layer_weight = layer.get_weight(module.weight) * (lora_weight * layer_scale)
+                        layer.to(device=torch.device("cpu"))
+
+                        if module.weight.shape != layer_weight.shape:
+                            layer_weight = layer_weight.reshape(module.weight.shape)
+
+                        module.weight += layer_weight.to(dtype=dtype)
+
+            yield  # wait for context manager exit
+
+        finally:
+            assert hasattr(model, "get_submodule")  # mypy not picking up fact that torch.nn.Module has get_submodule()
+            with torch.no_grad():
+                for module_key, weight in original_weights.items():
+                    model.get_submodule(module_key).weight.copy_(weight)

invokeai/backend/lora/sdxl_state_dict_utils.py

@@ -0,0 +1,157 @@
+import bisect
+from typing import TypeVar
+
+
+def make_sdxl_unet_conversion_map() -> list[tuple[str, str]]:
+    """Create a dict mapping state_dict keys from Stability AI SDXL format to diffusers SDXL format.
+
+    Ported from:
+    https://github.com/bmaltais/kohya_ss/blob/2accb1305979ba62f5077a23aabac23b4c37e935/networks/lora_diffusers.py#L15C1-L97C32
+    """
+    unet_conversion_map_layer: list[tuple[str, str]] = []
+
+    for i in range(3):  # num_blocks is 3 in sdxl
+        # loop over downblocks/upblocks
+        for j in range(2):
+            # loop over resnets/attentions for downblocks
+            hf_down_res_prefix = f"down_blocks.{i}.resnets.{j}."
+            sd_down_res_prefix = f"input_blocks.{3*i + j + 1}.0."
+            unet_conversion_map_layer.append((sd_down_res_prefix, hf_down_res_prefix))
+
+            if i < 3:
+                # no attention layers in down_blocks.3
+                hf_down_atn_prefix = f"down_blocks.{i}.attentions.{j}."
+                sd_down_atn_prefix = f"input_blocks.{3*i + j + 1}.1."
+                unet_conversion_map_layer.append((sd_down_atn_prefix, hf_down_atn_prefix))
+
+        for j in range(3):
+            # loop over resnets/attentions for upblocks
+            hf_up_res_prefix = f"up_blocks.{i}.resnets.{j}."
+            sd_up_res_prefix = f"output_blocks.{3*i + j}.0."
+            unet_conversion_map_layer.append((sd_up_res_prefix, hf_up_res_prefix))
+
+            # if i > 0: commentout for sdxl
+            # no attention layers in up_blocks.0
+            hf_up_atn_prefix = f"up_blocks.{i}.attentions.{j}."
+            sd_up_atn_prefix = f"output_blocks.{3*i + j}.1."
+            unet_conversion_map_layer.append((sd_up_atn_prefix, hf_up_atn_prefix))
+
+        if i < 3:
+            # no downsample in down_blocks.3
+            hf_downsample_prefix = f"down_blocks.{i}.downsamplers.0.conv."
+            sd_downsample_prefix = f"input_blocks.{3*(i+1)}.0.op."
+            unet_conversion_map_layer.append((sd_downsample_prefix, hf_downsample_prefix))
+
+            # no upsample in up_blocks.3
+            hf_upsample_prefix = f"up_blocks.{i}.upsamplers.0."
+            sd_upsample_prefix = f"output_blocks.{3*i + 2}.{2}."  # change for sdxl
+            unet_conversion_map_layer.append((sd_upsample_prefix, hf_upsample_prefix))
+
+    hf_mid_atn_prefix = "mid_block.attentions.0."
+    sd_mid_atn_prefix = "middle_block.1."
+    unet_conversion_map_layer.append((sd_mid_atn_prefix, hf_mid_atn_prefix))
+
+    for j in range(2):
+        hf_mid_res_prefix = f"mid_block.resnets.{j}."
+        sd_mid_res_prefix = f"middle_block.{2*j}."
+        unet_conversion_map_layer.append((sd_mid_res_prefix, hf_mid_res_prefix))
+
+    unet_conversion_map_resnet = [
+        # (stable-diffusion, HF Diffusers)
+        ("in_layers.0.", "norm1."),
+        ("in_layers.2.", "conv1."),
+        ("out_layers.0.", "norm2."),
+        ("out_layers.3.", "conv2."),
+        ("emb_layers.1.", "time_emb_proj."),
+        ("skip_connection.", "conv_shortcut."),
+    ]
+
+    unet_conversion_map: list[tuple[str, str]] = []
+    for sd, hf in unet_conversion_map_layer:
+        if "resnets" in hf:
+            for sd_res, hf_res in unet_conversion_map_resnet:
+                unet_conversion_map.append((sd + sd_res, hf + hf_res))
+        else:
+            unet_conversion_map.append((sd, hf))
+
+    for j in range(2):
+        hf_time_embed_prefix = f"time_embedding.linear_{j+1}."
+        sd_time_embed_prefix = f"time_embed.{j*2}."
+        unet_conversion_map.append((sd_time_embed_prefix, hf_time_embed_prefix))
+
+    for j in range(2):
+        hf_label_embed_prefix = f"add_embedding.linear_{j+1}."
+        sd_label_embed_prefix = f"label_emb.0.{j*2}."
+        unet_conversion_map.append((sd_label_embed_prefix, hf_label_embed_prefix))
+
+    unet_conversion_map.append(("input_blocks.0.0.", "conv_in."))
+    unet_conversion_map.append(("out.0.", "conv_norm_out."))
+    unet_conversion_map.append(("out.2.", "conv_out."))
+
+    return unet_conversion_map
+
+
+SDXL_UNET_STABILITY_TO_DIFFUSERS_MAP = {
+    sd.rstrip(".").replace(".", "_"): hf.rstrip(".").replace(".", "_") for sd, hf in make_sdxl_unet_conversion_map()
+}
+
+
+T = TypeVar("T")
+
+
+def convert_sdxl_keys_to_diffusers_format(state_dict: dict[str, T]) -> dict[str, T]:
+    """Convert the keys of an SDXL LoRA state_dict to diffusers format.
+
+    The input state_dict can be in either Stability AI format or diffusers format. If the state_dict is already in
+    diffusers format, then this function will have no effect.
+
+    This function is adapted from:
+    https://github.com/bmaltais/kohya_ss/blob/2accb1305979ba62f5077a23aabac23b4c37e935/networks/lora_diffusers.py#L385-L409
+
+    Args:
+        state_dict (dict[str, Tensor]): The SDXL LoRA state_dict.
+
+    Raises:
+        ValueError: If state_dict contains an unrecognized key, or not all keys could be converted.
+
+    Returns:
+        dict[str, Tensor]: The diffusers-format state_dict.
+    """
+    converted_count = 0  # The number of Stability AI keys converted to diffusers format.
+    not_converted_count = 0  # The number of keys that were not converted.
+
+    # Get a sorted list of Stability AI UNet keys so that we can efficiently search for keys with matching prefixes.
+    # For example, we want to efficiently find `input_blocks_4_1` in the list when searching for
+    # `input_blocks_4_1_proj_in`.
+    stability_unet_keys = list(SDXL_UNET_STABILITY_TO_DIFFUSERS_MAP)
+    stability_unet_keys.sort()
+
+    new_state_dict: dict[str, T] = {}
+    for full_key, value in state_dict.items():
+        if full_key.startswith("lora_unet_"):
+            search_key = full_key.replace("lora_unet_", "")
+            # Use bisect to find the key in stability_unet_keys that *may* match the search_key's prefix.
+            position = bisect.bisect_right(stability_unet_keys, search_key)
+            map_key = stability_unet_keys[position - 1]
+            # Now, check if the map_key *actually* matches the search_key.
+            if search_key.startswith(map_key):
+                new_key = full_key.replace(map_key, SDXL_UNET_STABILITY_TO_DIFFUSERS_MAP[map_key])
+                new_state_dict[new_key] = value
+                converted_count += 1
+            else:
+                new_state_dict[full_key] = value
+                not_converted_count += 1
+        elif full_key.startswith("lora_te1_") or full_key.startswith("lora_te2_"):
+            # The CLIP text encoders have the same keys in both Stability AI and diffusers formats.
+            new_state_dict[full_key] = value
+            continue
+        else:
+            raise ValueError(f"Unrecognized SDXL LoRA key prefix: '{full_key}'.")
+
+    if converted_count > 0 and not_converted_count > 0:
+        raise ValueError(
+            f"The SDXL LoRA could only be partially converted to diffusers format. converted={converted_count},"
+            f" not_converted={not_converted_count}"
+        )
+
+    return new_state_dict

invokeai/backend/model_manager/__init__.py

@@ -1,7 +1,6 @@
 """Re-export frequently-used symbols from the Model Manager backend."""
 
 from .config import (
-    AnyModel,
     AnyModelConfig,
     BaseModelType,
     InvalidModelConfigException,
@@ -18,7 +17,6 @@ from .probe import ModelProbe
 from .search import ModelSearch
 
 __all__ = [
-    "AnyModel",
     "AnyModelConfig",
     "BaseModelType",
     "ModelRepoVariant",
@@ -33,42 +31,3 @@ __all__ = [
     "SchedulerPredictionType",
     "SubModelType",
 ]
-
-########## to help populate the openapi_schema with format enums for each config ###########
-# This code is no longer necessary?
-# leave it here just in case
-#
-# import inspect
-# from enum import Enum
-# from typing import Any, Iterable, Dict, get_args, Set
-# def _expand(something: Any) -> Iterable[type]:
-#     if isinstance(something, type):
-#         yield something
-#     else:
-#         for x in get_args(something):
-#             for y in _expand(x):
-#                 yield y
-
-# def _find_format(cls: type) -> Iterable[Enum]:
-#     if hasattr(inspect, "get_annotations"):
-#         fields = inspect.get_annotations(cls)
-#     else:
-#         fields = cls.__annotations__
-#     if "format" in fields:
-#         for x in get_args(fields["format"]):
-#             yield x
-#     for parent_class in cls.__bases__:
-#         for x in _find_format(parent_class):
-#             yield x
-#     return None
-
-# def get_model_config_formats() -> Dict[str, Set[Enum]]:
-#     result: Dict[str, Set[Enum]] = {}
-#     for model_config in _expand(AnyModelConfig):
-#         for field in _find_format(model_config):
-#             if field is None:
-#                 continue
-#             if not result.get(model_config.__qualname__):
-#                 result[model_config.__qualname__] = set()
-#             result[model_config.__qualname__].add(field)
-#     return result

invokeai/backend/model_manager/any_model_type.py

@@ -0,0 +1,12 @@
+from typing import Union
+
+import torch
+from diffusers.models.modeling_utils import ModelMixin
+
+from invokeai.backend.ip_adapter.ip_adapter import IPAdapter
+from invokeai.backend.lora.lora_model import LoRAModelRaw
+from invokeai.backend.onnx.onnx_runtime import IAIOnnxRuntimeModel
+from invokeai.backend.textual_inversion import TextualInversionModelRaw
+
+# ModelMixin is the base class for all diffusers and transformers models
+AnyModel = Union[ModelMixin, torch.nn.Module, IPAdapter, LoRAModelRaw, TextualInversionModelRaw, IAIOnnxRuntimeModel]

invokeai/backend/model_manager/config.py

@@ -24,20 +24,12 @@ import time
 from enum import Enum
 from typing import Literal, Optional, Type, TypeAlias, Union
 
-import torch
-from diffusers.models.modeling_utils import ModelMixin
 from pydantic import BaseModel, ConfigDict, Discriminator, Field, Tag, TypeAdapter
 from typing_extensions import Annotated, Any, Dict
 
 from invokeai.app.invocations.constants import SCHEDULER_NAME_VALUES
 from invokeai.app.util.misc import uuid_string
 
-from ..raw_model import RawModel
-
-# ModelMixin is the base class for all diffusers and transformers models
-# RawModel is the InvokeAI wrapper class for ip_adapters, loras, textual_inversion and onnx runtime
-AnyModel = Union[ModelMixin, RawModel, torch.nn.Module]
-
 
 class InvalidModelConfigException(Exception):
     """Exception for when config parser doesn't recognized this combination of model type and format."""
@@ -323,10 +315,13 @@ class MainDiffusersConfig(DiffusersConfigBase, MainConfigBase):
         return Tag(f"{ModelType.Main.value}.{ModelFormat.Diffusers.value}")
 
 
-class IPAdapterConfig(ModelConfigBase):
-    """Model config for IP Adaptor format models."""
-
+class IPAdapterBaseConfig(ModelConfigBase):
     type: Literal[ModelType.IPAdapter] = ModelType.IPAdapter
+
+
+class IPAdapterInvokeAIConfig(IPAdapterBaseConfig):
+    """Model config for IP Adapter diffusers format models."""
+
     image_encoder_model_id: str
     format: Literal[ModelFormat.InvokeAI]
 
@@ -335,6 +330,16 @@ 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 +395,8 @@ AnyModelConfig = Annotated[
         Annotated[LoRADiffusersConfig, LoRADiffusersConfig.get_tag()],
         Annotated[TextualInversionFileConfig, TextualInversionFileConfig.get_tag()],
         Annotated[TextualInversionFolderConfig, TextualInversionFolderConfig.get_tag()],
-        Annotated[IPAdapterConfig, IPAdapterConfig.get_tag()],
+        Annotated[IPAdapterInvokeAIConfig, IPAdapterInvokeAIConfig.get_tag()],
+        Annotated[IPAdapterCheckpointConfig, IPAdapterCheckpointConfig.get_tag()],
         Annotated[T2IAdapterConfig, T2IAdapterConfig.get_tag()],
         Annotated[CLIPVisionDiffusersConfig, CLIPVisionDiffusersConfig.get_tag()],
     ],

invokeai/backend/model_manager/convert_ckpt_to_diffusers.py

@@ -3,10 +3,10 @@
 """Conversion script for the Stable Diffusion checkpoints."""
 
 from pathlib import Path
-from typing import Dict
+from typing import Optional
 
 import torch
-from diffusers import AutoencoderKL
+from diffusers.models.autoencoders.autoencoder_kl import AutoencoderKL
 from diffusers.pipelines.stable_diffusion.convert_from_ckpt import (
     convert_ldm_vae_checkpoint,
     create_vae_diffusers_config,
@@ -15,11 +15,14 @@ from diffusers.pipelines.stable_diffusion.convert_from_ckpt import (
 )
 from omegaconf import DictConfig
 
+from invokeai.backend.model_manager.any_model_type import AnyModel
+
 
 def convert_ldm_vae_to_diffusers(
-    checkpoint: Dict[str, torch.Tensor],
+    checkpoint: torch.Tensor | dict[str, torch.Tensor],
     vae_config: DictConfig,
     image_size: int,
+    dump_path: Optional[Path] = None,
     precision: torch.dtype = torch.float16,
 ) -> AutoencoderKL:
     """Convert a checkpoint-style VAE into a Diffusers VAE"""
@@ -28,16 +31,21 @@ def convert_ldm_vae_to_diffusers(
 
     vae = AutoencoderKL(**vae_config)
     vae.load_state_dict(converted_vae_checkpoint)
-    return vae.to(precision)
+    vae.to(precision)
+
+    if dump_path:
+        vae.save_pretrained(dump_path, safe_serialization=True)
+
+    return vae
 
 
 def convert_ckpt_to_diffusers(
     checkpoint_path: str | Path,
-    dump_path: str | Path,
+    dump_path: Optional[str | Path] = None,
     precision: torch.dtype = torch.float16,
     use_safetensors: bool = True,
     **kwargs,
-):
+) -> AnyModel:
     """
     Takes all the arguments of download_from_original_stable_diffusion_ckpt(),
     and in addition a path-like object indicating the location of the desired diffusers
@@ -47,18 +55,20 @@ def convert_ckpt_to_diffusers(
     pipe = pipe.to(precision)
 
     # TO DO: save correct repo variant
-    pipe.save_pretrained(
-        dump_path,
-        safe_serialization=use_safetensors,
-    )
+    if dump_path:
+        pipe.save_pretrained(
+            dump_path,
+            safe_serialization=use_safetensors,
+        )
+    return pipe
 
 
 def convert_controlnet_to_diffusers(
     checkpoint_path: Path,
-    dump_path: Path,
+    dump_path: Optional[Path] = None,
     precision: torch.dtype = torch.float16,
     **kwargs,
-):
+) -> AnyModel:
     """
     Takes all the arguments of download_controlnet_from_original_ckpt(),
     and in addition a path-like object indicating the location of the desired diffusers
@@ -68,4 +78,6 @@ def convert_controlnet_to_diffusers(
     pipe = pipe.to(precision)
 
     # TO DO: save correct repo variant
-    pipe.save_pretrained(dump_path, safe_serialization=True)
+    if dump_path:
+        pipe.save_pretrained(dump_path, safe_serialization=True)
+    return pipe

invokeai/backend/model_manager/load/convert_cache/convert_cache_default.py

@@ -19,11 +19,20 @@ class ModelConvertCache(ModelConvertCacheBase):
         self._cache_path = cache_path
         self._max_size = max_size
 
+        # adjust cache size at startup in case it has been changed
+        if self._cache_path.exists():
+            self.make_room(0.0)
+
     @property
     def max_size(self) -> float:
         """Return the maximum size of this cache directory (GB)."""
         return self._max_size
 
+    @max_size.setter
+    def max_size(self, value: float) -> None:
+        """Set the maximum size of this cache directory (GB)."""
+        self._max_size = value
+
     def cache_path(self, key: str) -> Path:
         """Return the path for a model with the indicated key."""
         return self._cache_path / key

invokeai/backend/model_manager/load/load_base.py

@@ -10,8 +10,8 @@ from pathlib import Path
 from typing import Any, Optional
 
 from invokeai.app.services.config import InvokeAIAppConfig
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.model_manager.config import (
-    AnyModel,
     AnyModelConfig,
     SubModelType,
 )
@@ -83,3 +83,15 @@ class ModelLoaderBase(ABC):
     ) -> int:
         """Return size in bytes of the model, calculated before loading."""
         pass
+
+    @property
+    @abstractmethod
+    def convert_cache(self) -> ModelConvertCacheBase:
+        """Return the convert cache associated with this loader."""
+        pass
+
+    @property
+    @abstractmethod
+    def ram_cache(self) -> ModelCacheBase[AnyModel]:
+        """Return the ram cache associated with this loader."""
+        pass

invokeai/backend/model_manager/load/load_default.py

@@ -3,16 +3,15 @@
 
 from logging import Logger
 from pathlib import Path
-from typing import Optional, Tuple
+from typing import Optional
 
 from invokeai.app.services.config import InvokeAIAppConfig
 from invokeai.backend.model_manager import (
-    AnyModel,
     AnyModelConfig,
     InvalidModelConfigException,
-    ModelRepoVariant,
     SubModelType,
 )
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.model_manager.config import DiffusersConfigBase, ModelType
 from invokeai.backend.model_manager.load.convert_cache import ModelConvertCacheBase
 from invokeai.backend.model_manager.load.load_base import LoadedModel, ModelLoaderBase
@@ -54,51 +53,43 @@ class ModelLoader(ModelLoaderBase):
         if model_config.type is ModelType.Main and not submodel_type:
             raise InvalidModelConfigException("submodel_type is required when loading a main model")
 
-        model_path, model_config, submodel_type = self._get_model_path(model_config, submodel_type)
+        model_path = self._get_model_path(model_config)
 
         if not model_path.exists():
             raise InvalidModelConfigException(f"Files for model '{model_config.name}' not found at {model_path}")
 
-        model_path = self._convert_if_needed(model_config, model_path, submodel_type)
-        locker = self._load_if_needed(model_config, model_path, submodel_type)
+        with skip_torch_weight_init():
+            locker = self._convert_and_load(model_config, model_path, submodel_type)
         return LoadedModel(config=model_config, _locker=locker)
 
-    def _get_model_path(
-        self, config: AnyModelConfig, submodel_type: Optional[SubModelType] = None
-    ) -> Tuple[Path, AnyModelConfig, Optional[SubModelType]]:
+    @property
+    def convert_cache(self) -> ModelConvertCacheBase:
+        """Return the convert cache associated with this loader."""
+        return self._convert_cache
+
+    @property
+    def ram_cache(self) -> ModelCacheBase[AnyModel]:
+        """Return the ram cache associated with this loader."""
+        return self._ram_cache
+
+    def _get_model_path(self, config: AnyModelConfig) -> Path:
         model_base = self._app_config.models_path
-        result = (model_base / config.path).resolve(), config, submodel_type
-        return result
+        return (model_base / config.path).resolve()
 
-    def _convert_if_needed(
-        self, config: AnyModelConfig, model_path: Path, submodel_type: Optional[SubModelType] = None
-    ) -> Path:
-        cache_path: Path = self._convert_cache.cache_path(config.key)
-
-        if not self._needs_conversion(config, model_path, cache_path):
-            return cache_path if cache_path.exists() else model_path
-
-        self._convert_cache.make_room(self.get_size_fs(config, model_path, submodel_type))
-        return self._convert_model(config, model_path, cache_path)
-
-    def _needs_conversion(self, config: AnyModelConfig, model_path: Path, dest_path: Path) -> bool:
-        return False
-
-    def _load_if_needed(
+    def _convert_and_load(
         self, config: AnyModelConfig, model_path: Path, submodel_type: Optional[SubModelType] = None
     ) -> ModelLockerBase:
-        # TO DO: This is not thread safe!
         try:
             return self._ram_cache.get(config.key, submodel_type)
         except IndexError:
             pass
 
-        model_variant = getattr(config, "repo_variant", None)
-        self._ram_cache.make_room(self.get_size_fs(config, model_path, submodel_type))
-
-        # This is where the model is actually loaded!
-        with skip_torch_weight_init():
-            loaded_model = self._load_model(model_path, model_variant=model_variant, submodel_type=submodel_type)
+        cache_path: Path = self._convert_cache.cache_path(config.key)
+        if self._needs_conversion(config, model_path, cache_path):
+            loaded_model = self._do_convert(config, model_path, cache_path, submodel_type)
+        else:
+            config.path = str(cache_path) if cache_path.exists() else str(self._get_model_path(config))
+            loaded_model = self._load_model(config, submodel_type)
 
         self._ram_cache.put(
             config.key,
@@ -123,15 +114,34 @@ class ModelLoader(ModelLoaderBase):
             variant=config.repo_variant if isinstance(config, DiffusersConfigBase) else None,
         )
 
+    def _do_convert(
+        self, config: AnyModelConfig, model_path: Path, cache_path: Path, submodel_type: Optional[SubModelType] = None
+    ) -> AnyModel:
+        self.convert_cache.make_room(calc_model_size_by_fs(model_path))
+        pipeline = self._convert_model(config, model_path, cache_path if self.convert_cache.max_size > 0 else None)
+        if submodel_type:
+            # Proactively load the various submodels into the RAM cache so that we don't have to re-convert
+            # the entire pipeline every time a new submodel is needed.
+            for subtype in SubModelType:
+                if subtype == submodel_type:
+                    continue
+                if submodel := getattr(pipeline, subtype.value, None):
+                    self._ram_cache.put(
+                        config.key, submodel_type=subtype, model=submodel, size=calc_model_size_by_data(submodel)
+                    )
+        return getattr(pipeline, submodel_type.value) if submodel_type else pipeline
+
+    def _needs_conversion(self, config: AnyModelConfig, model_path: Path, dest_path: Path) -> bool:
+        return False
+
     # This needs to be implemented in subclasses that handle checkpoints
-    def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Path) -> Path:
+    def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Optional[Path] = None) -> AnyModel:
         raise NotImplementedError
 
     # This needs to be implemented in the subclass
     def _load_model(
         self,
-        model_path: Path,
-        model_variant: Optional[ModelRepoVariant] = None,
+        config: AnyModelConfig,
         submodel_type: Optional[SubModelType] = None,
     ) -> AnyModel:
         raise NotImplementedError

invokeai/backend/model_manager/load/model_cache/model_cache_base.py

@@ -14,7 +14,8 @@ from typing import Dict, Generic, Optional, TypeVar
 
 import torch
 
-from invokeai.backend.model_manager.config import AnyModel, SubModelType
+from invokeai.backend.model_manager.any_model_type import AnyModel
+from invokeai.backend.model_manager.config import SubModelType
 
 
 class ModelLockerBase(ABC):

invokeai/backend/model_manager/load/model_cache/model_cache_default.py

@@ -28,7 +28,8 @@ from typing import Dict, List, Optional
 
 import torch
 
-from invokeai.backend.model_manager import AnyModel, SubModelType
+from invokeai.backend.model_manager import SubModelType
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.model_manager.load.memory_snapshot import MemorySnapshot, get_pretty_snapshot_diff
 from invokeai.backend.util.devices import choose_torch_device
 from invokeai.backend.util.logging import InvokeAILogger
@@ -122,6 +123,11 @@ class ModelCache(ModelCacheBase[AnyModel]):
         """Return the cap on cache size."""
         return self._max_cache_size
 
+    @max_cache_size.setter
+    def max_cache_size(self, value: float) -> None:
+        """Set the cap on cache size."""
+        self._max_cache_size = value
+
     @property
     def stats(self) -> Optional[CacheStats]:
         """Return collected CacheStats object."""
@@ -157,8 +163,9 @@ class ModelCache(ModelCacheBase[AnyModel]):
     ) -> None:
         """Store model under key and optional submodel_type."""
         key = self._make_cache_key(key, submodel_type)
-        assert key not in self._cached_models
-
+        if key in self._cached_models:
+            return
+        self.make_room(size)
         cache_record = CacheRecord(key, model, size)
         self._cached_models[key] = cache_record
         self._cache_stack.append(key)
@@ -405,6 +412,8 @@ class ModelCache(ModelCacheBase[AnyModel]):
             #
             # Keep in mind that gc is only responsible for handling reference cycles. Most objects should be cleaned up
             # immediately when their reference count hits 0.
+            if self.stats:
+                self.stats.cleared = models_cleared
             gc.collect()
 
         torch.cuda.empty_cache()
@@ -421,4 +430,8 @@ class ModelCache(ModelCacheBase[AnyModel]):
         )
         free_mem, _ = torch.cuda.mem_get_info(torch.device(vram_device))
         if needed_size > free_mem:
-            raise torch.cuda.OutOfMemoryError
+            needed_gb = round(needed_size / GIG, 2)
+            free_gb = round(free_mem / GIG, 2)
+            raise torch.cuda.OutOfMemoryError(
+                f"Insufficient VRAM to load model, requested {needed_gb}GB but only had {free_gb}GB free"
+            )

invokeai/backend/model_manager/load/model_cache/model_locker.py

@@ -4,7 +4,7 @@ Base class and implementation of a class that moves models in and out of VRAM.
 
 import torch
 
-from invokeai.backend.model_manager import AnyModel
+from invokeai.backend.model_manager.any_model_type import AnyModel
 
 from .model_cache_base import CacheRecord, ModelCacheBase, ModelLockerBase
 

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

@@ -2,6 +2,7 @@
 """Class for ControlNet model loading in InvokeAI."""
 
 from pathlib import Path
+from typing import Optional
 
 from invokeai.backend.model_manager import (
     AnyModelConfig,
@@ -9,6 +10,7 @@ from invokeai.backend.model_manager import (
     ModelFormat,
     ModelType,
 )
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.model_manager.config import CheckpointConfigBase
 from invokeai.backend.model_manager.convert_ckpt_to_diffusers import convert_controlnet_to_diffusers
 
@@ -33,7 +35,7 @@ class ControlNetLoader(GenericDiffusersLoader):
         else:
             return True
 
-    def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Path) -> Path:
+    def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Optional[Path] = None) -> AnyModel:
         assert isinstance(config, CheckpointConfigBase)
         image_size = (
             512
@@ -44,8 +46,8 @@ class ControlNetLoader(GenericDiffusersLoader):
         )
 
         self._logger.info(f"Converting {model_path} to diffusers format")
-        with open(self._app_config.root_path / config.config_path, "r") as config_stream:
-            convert_controlnet_to_diffusers(
+        with open(self._app_config.legacy_conf_path / config.config_path, "r") as config_stream:
+            result = convert_controlnet_to_diffusers(
                 model_path,
                 output_path,
                 original_config_file=config_stream,
@@ -53,4 +55,4 @@ class ControlNetLoader(GenericDiffusersLoader):
                 precision=self._torch_dtype,
                 from_safetensors=model_path.suffix == ".safetensors",
             )
-        return output_path
+        return result

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

@@ -9,14 +9,15 @@ from diffusers.configuration_utils import ConfigMixin
 from diffusers.models.modeling_utils import ModelMixin
 
 from invokeai.backend.model_manager import (
-    AnyModel,
+    AnyModelConfig,
     BaseModelType,
     InvalidModelConfigException,
     ModelFormat,
-    ModelRepoVariant,
     ModelType,
     SubModelType,
 )
+from invokeai.backend.model_manager.any_model_type import AnyModel
+from invokeai.backend.model_manager.config import DiffusersConfigBase
 
 from .. import ModelLoader, ModelLoaderRegistry
 
@@ -28,14 +29,15 @@ class GenericDiffusersLoader(ModelLoader):
 
     def _load_model(
         self,
-        model_path: Path,
-        model_variant: Optional[ModelRepoVariant] = None,
+        config: AnyModelConfig,
         submodel_type: Optional[SubModelType] = None,
     ) -> AnyModel:
+        model_path = Path(config.path)
         model_class = self.get_hf_load_class(model_path)
         if submodel_type is not None:
             raise Exception(f"There are no submodels in models of type {model_class}")
-        variant = model_variant.value if model_variant else None
+        repo_variant = config.repo_variant if isinstance(config, DiffusersConfigBase) else None
+        variant = repo_variant.value if repo_variant else None
         try:
             result: AnyModel = model_class.from_pretrained(model_path, torch_dtype=self._torch_dtype, variant=variant)
         except OSError as e:

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

@@ -7,31 +7,26 @@ from typing import Optional
 import torch
 
 from invokeai.backend.ip_adapter.ip_adapter import build_ip_adapter
-from invokeai.backend.model_manager import (
-    AnyModel,
-    BaseModelType,
-    ModelFormat,
-    ModelRepoVariant,
-    ModelType,
-    SubModelType,
-)
+from invokeai.backend.model_manager import AnyModelConfig, BaseModelType, ModelFormat, ModelType, SubModelType
+from invokeai.backend.model_manager.any_model_type import AnyModel
 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."""
 
     def _load_model(
         self,
-        model_path: Path,
-        model_variant: Optional[ModelRepoVariant] = None,
+        config: AnyModelConfig,
         submodel_type: Optional[SubModelType] = None,
     ) -> AnyModel:
         if submodel_type is not None:
             raise ValueError("There are no submodels in an IP-Adapter model.")
+        model_path = Path(config.path)
         model = build_ip_adapter(
-            ip_adapter_ckpt_path=str(model_path / "ip_adapter.bin"),
+            ip_adapter_ckpt_path=model_path,
             device=torch.device("cpu"),
             dtype=self._torch_dtype,
         )

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

@@ -3,19 +3,18 @@
 
 from logging import Logger
 from pathlib import Path
-from typing import Optional, Tuple
+from typing import Optional
 
 from invokeai.app.services.config import InvokeAIAppConfig
-from invokeai.backend.lora import LoRAModelRaw
+from invokeai.backend.lora.lora_model import LoRAModelRaw
 from invokeai.backend.model_manager import (
-    AnyModel,
     AnyModelConfig,
     BaseModelType,
     ModelFormat,
-    ModelRepoVariant,
     ModelType,
     SubModelType,
 )
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.model_manager.load.convert_cache import ModelConvertCacheBase
 from invokeai.backend.model_manager.load.model_cache.model_cache_base import ModelCacheBase
 
@@ -41,12 +40,12 @@ class LoRALoader(ModelLoader):
 
     def _load_model(
         self,
-        model_path: Path,
-        model_variant: Optional[ModelRepoVariant] = None,
+        config: AnyModelConfig,
         submodel_type: Optional[SubModelType] = None,
     ) -> AnyModel:
         if submodel_type is not None:
             raise ValueError("There are no submodels in a LoRA model.")
+        model_path = Path(config.path)
         assert self._model_base is not None
         model = LoRAModelRaw.from_checkpoint(
             file_path=model_path,
@@ -56,12 +55,9 @@ class LoRALoader(ModelLoader):
         return model
 
     # override
-    def _get_model_path(
-        self, config: AnyModelConfig, submodel_type: Optional[SubModelType] = None
-    ) -> Tuple[Path, AnyModelConfig, Optional[SubModelType]]:
-        self._model_base = (
-            config.base
-        )  # cheating a little - we remember this variable for using in the subsequent call to _load_model()
+    def _get_model_path(self, config: AnyModelConfig) -> Path:
+        # cheating a little - we remember this variable for using in the subsequent call to _load_model()
+        self._model_base = config.base
 
         model_base_path = self._app_config.models_path
         model_path = model_base_path / config.path
@@ -73,5 +69,4 @@ class LoRALoader(ModelLoader):
                     model_path = path
                     break
 
-        result = model_path.resolve(), config, submodel_type
-        return result
+        return model_path.resolve()

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

@@ -6,13 +6,13 @@ from pathlib import Path
 from typing import Optional
 
 from invokeai.backend.model_manager import (
-    AnyModel,
+    AnyModelConfig,
     BaseModelType,
     ModelFormat,
-    ModelRepoVariant,
     ModelType,
     SubModelType,
 )
+from invokeai.backend.model_manager.any_model_type import AnyModel
 
 from .. import ModelLoaderRegistry
 from .generic_diffusers import GenericDiffusersLoader
@@ -25,18 +25,19 @@ class OnnyxDiffusersModel(GenericDiffusersLoader):
 
     def _load_model(
         self,
-        model_path: Path,
-        model_variant: Optional[ModelRepoVariant] = None,
+        config: AnyModelConfig,
         submodel_type: Optional[SubModelType] = None,
     ) -> AnyModel:
         if not submodel_type is not None:
             raise Exception("A submodel type must be provided when loading onnx pipelines.")
+        model_path = Path(config.path)
         load_class = self.get_hf_load_class(model_path, submodel_type)
-        variant = model_variant.value if model_variant else None
+        repo_variant = getattr(config, "repo_variant", None)
+        variant = repo_variant.value if repo_variant else None
         model_path = model_path / submodel_type.value
         result: AnyModel = load_class.from_pretrained(
             model_path,
             torch_dtype=self._torch_dtype,
             variant=variant,
-        )  # type: ignore
+        )
         return result

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

@@ -5,16 +5,20 @@ from pathlib import Path
 from typing import Optional
 
 from invokeai.backend.model_manager import (
-    AnyModel,
     AnyModelConfig,
     BaseModelType,
     ModelFormat,
-    ModelRepoVariant,
     ModelType,
     SchedulerPredictionType,
     SubModelType,
 )
-from invokeai.backend.model_manager.config import CheckpointConfigBase, MainCheckpointConfig, ModelVariantType
+from invokeai.backend.model_manager.any_model_type import AnyModel
+from invokeai.backend.model_manager.config import (
+    CheckpointConfigBase,
+    DiffusersConfigBase,
+    MainCheckpointConfig,
+    ModelVariantType,
+)
 from invokeai.backend.model_manager.convert_ckpt_to_diffusers import convert_ckpt_to_diffusers
 
 from .. import ModelLoaderRegistry
@@ -41,14 +45,15 @@ class StableDiffusionDiffusersModel(GenericDiffusersLoader):
 
     def _load_model(
         self,
-        model_path: Path,
-        model_variant: Optional[ModelRepoVariant] = None,
+        config: AnyModelConfig,
         submodel_type: Optional[SubModelType] = None,
     ) -> AnyModel:
         if not submodel_type is not None:
             raise Exception("A submodel type must be provided when loading main pipelines.")
+        model_path = Path(config.path)
         load_class = self.get_hf_load_class(model_path, submodel_type)
-        variant = model_variant.value if model_variant else None
+        repo_variant = config.repo_variant if isinstance(config, DiffusersConfigBase) else None
+        variant = repo_variant.value if repo_variant else None
         model_path = model_path / submodel_type.value
         try:
             result: AnyModel = load_class.from_pretrained(
@@ -78,7 +83,7 @@ class StableDiffusionDiffusersModel(GenericDiffusersLoader):
         else:
             return True
 
-    def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Path) -> Path:
+    def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Optional[Path] = None) -> AnyModel:
         assert isinstance(config, MainCheckpointConfig)
         base = config.base
 
@@ -94,11 +99,11 @@ class StableDiffusionDiffusersModel(GenericDiffusersLoader):
 
         self._logger.info(f"Converting {model_path} to diffusers format")
 
-        convert_ckpt_to_diffusers(
+        loaded_model = convert_ckpt_to_diffusers(
             model_path,
             output_path,
             model_type=self.model_base_to_model_type[base],
-            original_config_file=self._app_config.root_path / config.config_path,
+            original_config_file=self._app_config.legacy_conf_path / config.config_path,
             extract_ema=True,
             from_safetensors=model_path.suffix == ".safetensors",
             precision=self._torch_dtype,
@@ -108,4 +113,4 @@ class StableDiffusionDiffusersModel(GenericDiffusersLoader):
             load_safety_checker=False,
             num_in_channels=VARIANT_TO_IN_CHANNEL_MAP[config.variant],
         )
-        return output_path
+        return loaded_model

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

@@ -2,17 +2,16 @@
 """Class for TI model loading in InvokeAI."""
 
 from pathlib import Path
-from typing import Optional, Tuple
+from typing import Optional
 
 from invokeai.backend.model_manager import (
-    AnyModel,
     AnyModelConfig,
     BaseModelType,
     ModelFormat,
-    ModelRepoVariant,
     ModelType,
     SubModelType,
 )
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.textual_inversion import TextualInversionModelRaw
 
 from .. import ModelLoader, ModelLoaderRegistry
@@ -27,22 +26,19 @@ class TextualInversionLoader(ModelLoader):
 
     def _load_model(
         self,
-        model_path: Path,
-        model_variant: Optional[ModelRepoVariant] = None,
+        config: AnyModelConfig,
         submodel_type: Optional[SubModelType] = None,
     ) -> AnyModel:
         if submodel_type is not None:
             raise ValueError("There are no submodels in a TI model.")
         model = TextualInversionModelRaw.from_checkpoint(
-            file_path=model_path,
+            file_path=config.path,
             dtype=self._torch_dtype,
         )
         return model
 
     # override
-    def _get_model_path(
-        self, config: AnyModelConfig, submodel_type: Optional[SubModelType] = None
-    ) -> Tuple[Path, AnyModelConfig, Optional[SubModelType]]:
+    def _get_model_path(self, config: AnyModelConfig) -> Path:
         model_path = self._app_config.models_path / config.path
 
         if config.format == ModelFormat.EmbeddingFolder:
@@ -53,4 +49,4 @@ class TextualInversionLoader(ModelLoader):
         if not path.exists():
             raise OSError(f"The embedding file at {path} was not found")
 
-        return path, config, submodel_type
+        return path

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

@@ -2,6 +2,7 @@
 """Class for VAE model loading in InvokeAI."""
 
 from pathlib import Path
+from typing import Optional
 
 import torch
 from omegaconf import DictConfig, OmegaConf
@@ -13,6 +14,7 @@ from invokeai.backend.model_manager import (
     ModelFormat,
     ModelType,
 )
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.model_manager.config import CheckpointConfigBase
 from invokeai.backend.model_manager.convert_ckpt_to_diffusers import convert_ldm_vae_to_diffusers
 
@@ -38,13 +40,13 @@ class VAELoader(GenericDiffusersLoader):
         else:
             return True
 
-    def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Path) -> Path:
+    def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Optional[Path] = None) -> AnyModel:
         # TODO(MM2): check whether sdxl VAE models convert.
         if config.base not in {BaseModelType.StableDiffusion1, BaseModelType.StableDiffusion2}:
             raise Exception(f"VAE conversion not supported for model type: {config.base}")
         else:
             assert isinstance(config, CheckpointConfigBase)
-            config_file = self._app_config.root_path / config.config_path
+            config_file = self._app_config.legacy_conf_path / config.config_path
 
         if model_path.suffix == ".safetensors":
             checkpoint = safetensors_load_file(model_path, device="cpu")
@@ -63,6 +65,6 @@ class VAELoader(GenericDiffusersLoader):
             vae_config=ckpt_config,
             image_size=512,
             precision=self._torch_dtype,
+            dump_path=output_path,
         )
-        vae_model.save_pretrained(output_path, safe_serialization=True)
-        return output_path
+        return vae_model

invokeai/backend/model_manager/load/model_util.py

@@ -8,7 +8,7 @@ from typing import Optional
 import torch
 from diffusers import DiffusionPipeline
 
-from invokeai.backend.model_manager.config import AnyModel
+from invokeai.backend.model_manager.any_model_type import AnyModel
 from invokeai.backend.onnx.onnx_runtime import IAIOnnxRuntimeModel
 
 

invokeai/backend/model_manager/load/optimizations.py

@@ -17,7 +17,7 @@ def skip_torch_weight_init() -> Generator[None, None, None]:
     completely unnecessary if the intent is to load checkpoint weights from disk for the layer. This context manager
     monkey-patches common torch layers to skip the weight initialization step.
     """
-    torch_modules = [torch.nn.Linear, torch.nn.modules.conv._ConvNd, torch.nn.Embedding]
+    torch_modules = [torch.nn.Linear, torch.nn.modules.conv._ConvNd, torch.nn.Embedding, torch.nn.LayerNorm]
     saved_functions = [hasattr(m, "reset_parameters") and m.reset_parameters for m in torch_modules]
 
     try:

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()):
@@ -323,7 +324,7 @@ class ModelProbe(object):
         with SilenceWarnings():
             if model_path.suffix.endswith((".ckpt", ".pt", ".pth", ".bin")):
                 cls._scan_model(model_path.name, model_path)
-                model = torch.load(model_path)
+                model = torch.load(model_path, map_location="cpu")
                 assert isinstance(model, dict)
                 return model
             else:
@@ -527,8 +528,25 @@ 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
+            cross_attention_dim = checkpoint["ip_adapter.1.to_k_ip.weight"].shape[-1]
+            if cross_attention_dim == 768:
+                return BaseModelType.StableDiffusion1
+            elif cross_attention_dim == 1024:
+                return BaseModelType.StableDiffusion2
+            elif cross_attention_dim == 2048:
+                return BaseModelType.StableDiffusionXL
+            else:
+                raise InvalidModelConfigException(
+                    f"IP-Adapter had unexpected cross-attention dimension: {cross_attention_dim}."
+                )
+        raise InvalidModelConfigException(f"{self.model_path}: Unable to determine base type")
 
 
 class CLIPVisionCheckpointProbe(CheckpointProbeBase):
@@ -768,7 +786,7 @@ 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)

invokeai/backend/model_patcher.py

@@ -13,157 +13,14 @@ from diffusers import OnnxRuntimeModel, UNet2DConditionModel
 from transformers import CLIPTextModel, CLIPTextModelWithProjection, CLIPTokenizer
 
 from invokeai.app.shared.models import FreeUConfig
-from invokeai.backend.model_manager import AnyModel
+from invokeai.backend.lora.lora_model import LoRAModelRaw
 from invokeai.backend.model_manager.load.optimizations import skip_torch_weight_init
 from invokeai.backend.onnx.onnx_runtime import IAIOnnxRuntimeModel
 
-from .lora import LoRAModelRaw
 from .textual_inversion import TextualInversionManager, TextualInversionModelRaw
 
-"""
-loras = [
-    (lora_model1, 0.7),
-    (lora_model2, 0.4),
-]
-with LoRAHelper.apply_lora_unet(unet, loras):
-    # unet with applied loras
-# unmodified unet
 
-"""
-
-
-# TODO: rename smth like ModelPatcher and add TI method?
 class ModelPatcher:
-    @staticmethod
-    def _resolve_lora_key(model: torch.nn.Module, lora_key: str, prefix: str) -> Tuple[str, torch.nn.Module]:
-        assert "." not in lora_key
-
-        if not lora_key.startswith(prefix):
-            raise Exception(f"lora_key with invalid prefix: {lora_key}, {prefix}")
-
-        module = model
-        module_key = ""
-        key_parts = lora_key[len(prefix) :].split("_")
-
-        submodule_name = key_parts.pop(0)
-
-        while len(key_parts) > 0:
-            try:
-                module = module.get_submodule(submodule_name)
-                module_key += "." + submodule_name
-                submodule_name = key_parts.pop(0)
-            except Exception:
-                submodule_name += "_" + key_parts.pop(0)
-
-        module = module.get_submodule(submodule_name)
-        module_key = (module_key + "." + submodule_name).lstrip(".")
-
-        return (module_key, module)
-
-    @classmethod
-    @contextmanager
-    def apply_lora_unet(
-        cls,
-        unet: UNet2DConditionModel,
-        loras: Iterator[Tuple[LoRAModelRaw, float]],
-    ) -> None:
-        with cls.apply_lora(unet, loras, "lora_unet_"):
-            yield
-
-    @classmethod
-    @contextmanager
-    def apply_lora_text_encoder(
-        cls,
-        text_encoder: CLIPTextModel,
-        loras: Iterator[Tuple[LoRAModelRaw, float]],
-    ) -> None:
-        with cls.apply_lora(text_encoder, loras, "lora_te_"):
-            yield
-
-    @classmethod
-    @contextmanager
-    def apply_sdxl_lora_text_encoder(
-        cls,
-        text_encoder: CLIPTextModel,
-        loras: List[Tuple[LoRAModelRaw, float]],
-    ) -> None:
-        with cls.apply_lora(text_encoder, loras, "lora_te1_"):
-            yield
-
-    @classmethod
-    @contextmanager
-    def apply_sdxl_lora_text_encoder2(
-        cls,
-        text_encoder: CLIPTextModel,
-        loras: List[Tuple[LoRAModelRaw, float]],
-    ) -> None:
-        with cls.apply_lora(text_encoder, loras, "lora_te2_"):
-            yield
-
-    @classmethod
-    @contextmanager
-    def apply_lora(
-        cls,
-        model: AnyModel,
-        loras: Iterator[Tuple[LoRAModelRaw, float]],
-        prefix: str,
-    ) -> None:
-        original_weights = {}
-        try:
-            with torch.no_grad():
-                for lora, lora_weight in loras:
-                    # assert lora.device.type == "cpu"
-                    for layer_key, layer in lora.layers.items():
-                        if not layer_key.startswith(prefix):
-                            continue
-
-                        # TODO(ryand): A non-negligible amount of time is currently spent resolving LoRA keys. This
-                        # should be improved in the following ways:
-                        # 1. The key mapping could be more-efficiently pre-computed. This would save time every time a
-                        #    LoRA model is applied.
-                        # 2. From an API perspective, there's no reason that the `ModelPatcher` should be aware of the
-                        #    intricacies of Stable Diffusion key resolution. It should just expect the input LoRA
-                        #    weights to have valid keys.
-                        assert isinstance(model, torch.nn.Module)
-                        module_key, module = cls._resolve_lora_key(model, layer_key, prefix)
-
-                        # All of the LoRA weight calculations will be done on the same device as the module weight.
-                        # (Performance will be best if this is a CUDA device.)
-                        device = module.weight.device
-                        dtype = module.weight.dtype
-
-                        if module_key not in original_weights:
-                            original_weights[module_key] = module.weight.detach().to(device="cpu", copy=True)
-
-                        layer_scale = layer.alpha / layer.rank if (layer.alpha and layer.rank) else 1.0
-
-                        # We intentionally move to the target device first, then cast. Experimentally, this was found to
-                        # be significantly faster for 16-bit CPU tensors being moved to a CUDA device than doing the
-                        # same thing in a single call to '.to(...)'.
-                        layer.to(device=device)
-                        layer.to(dtype=torch.float32)
-                        # TODO(ryand): Using torch.autocast(...) over explicit casting may offer a speed benefit on CUDA
-                        # devices here. Experimentally, it was found to be very slow on CPU. More investigation needed.
-                        layer_weight = layer.get_weight(module.weight) * (lora_weight * layer_scale)
-                        layer.to(device=torch.device("cpu"))
-
-                        assert isinstance(layer_weight, torch.Tensor)  # mypy thinks layer_weight is a float|Any ??!
-                        if module.weight.shape != layer_weight.shape:
-                            # TODO: debug on lycoris
-                            assert hasattr(layer_weight, "reshape")
-                            layer_weight = layer_weight.reshape(module.weight.shape)
-
-                        assert isinstance(layer_weight, torch.Tensor)  # mypy thinks layer_weight is a float|Any ??!
-                        module.weight += layer_weight.to(dtype=dtype)
-
-            yield  # wait for context manager exit
-
-        finally:
-            assert hasattr(model, "get_submodule")  # mypy not picking up fact that torch.nn.Module has get_submodule()
-            with torch.no_grad():
-                for module_key, weight in original_weights.items():
-                    model.get_submodule(module_key).weight.copy_(weight)
-
     @classmethod
     @contextmanager
     def apply_ti(

invokeai/backend/onnx/onnx_runtime.py

@@ -6,17 +6,16 @@ from typing import Any, List, Optional, Tuple, Union
 
 import numpy as np
 import onnx
+import torch
 from onnx import numpy_helper
 from onnxruntime import InferenceSession, SessionOptions, get_available_providers
 
-from ..raw_model import RawModel
-
 ONNX_WEIGHTS_NAME = "model.onnx"
 
 
 # NOTE FROM LS: This was copied from Stalker's original implementation.
 # I have not yet gone through and fixed all the type hints
-class IAIOnnxRuntimeModel(RawModel):
+class IAIOnnxRuntimeModel(torch.nn.Module):
     class _tensor_access:
         def __init__(self, model):  # type: ignore
             self.model = model
@@ -103,7 +102,7 @@ class IAIOnnxRuntimeModel(RawModel):
 
         self.proto = onnx.load(model_path, load_external_data=False)
         """
-
+        super().__init__()
         self.proto = onnx.load(model_path, load_external_data=True)
         # self.data = dict()
         # for tensor in self.proto.graph.initializer:

invokeai/backend/raw_model.py

@@ -1,15 +0,0 @@
-"""Base class for 'Raw' models.
-
-The RawModel class is the base class of LoRAModelRaw and TextualInversionModelRaw,
-and is used for type checking of calls to the model patcher. Its main purpose
-is to avoid a circular import issues when lora.py tries to import BaseModelType
-from invokeai.backend.model_manager.config, and the latter tries to import LoRAModelRaw
-from lora.py.
-
-The term 'raw' was introduced to describe a wrapper around a torch.nn.Module
-that adds additional methods and attributes.
-"""
-
-
-class RawModel:
-    """Base class for 'Raw' model wrappers."""

invokeai/backend/stable_diffusion/seamless.py

@@ -28,6 +28,10 @@ def _conv_forward_asymmetric(self, input, weight, bias):
 
 @contextmanager
 def set_seamless(model: Union[UNet2DConditionModel, AutoencoderKL, AutoencoderTiny], seamless_axes: List[str]):
+    if not seamless_axes:
+        yield
+        return
+
     # Callable: (input: Tensor, weight: Tensor, bias: Optional[Tensor]) -> Tensor
     to_restore: list[tuple[nn.Conv2d | nn.ConvTranspose2d, Callable]] = []
     try:

invokeai/backend/textual_inversion.py

@@ -9,10 +9,8 @@ from safetensors.torch import load_file
 from transformers import CLIPTokenizer
 from typing_extensions import Self
 
-from .raw_model import RawModel
 
-
-class TextualInversionModelRaw(RawModel):
+class TextualInversionModelRaw(torch.nn.Module):
     embedding: torch.Tensor  # [n, 768]|[n, 1280]
     embedding_2: Optional[torch.Tensor] = None  # [n, 768]|[n, 1280]   - for SDXL models
 

invokeai/backend/util/db_maintenance.py

@@ -31,6 +31,9 @@ class ConfigMapper:
     YAML_FILENAME = "invokeai.yaml"
     DATABASE_FILENAME = "invokeai.db"
 
+    DEFAULT_OUTDIR = "outputs"
+    DEFAULT_DB_DIR = "databases"
+
     database_path = None
     database_backup_dir = None
     outputs_path = None
@@ -50,12 +53,18 @@ class ConfigMapper:
     def __load_from_root_config(self, invoke_root):
         """Validate a yaml path exists, confirm the user wants to use it and load config."""
         yaml_path = os.path.join(invoke_root, self.YAML_FILENAME)
+        if not os.path.exists(yaml_path):
+            print(f"Unable to find invokeai.yaml at {yaml_path}!")
+            return False
         if os.path.exists(yaml_path):
             db_dir, outdir = self.__load_paths_from_yaml_file(yaml_path)
 
-            if db_dir is None or outdir is None:
-                print("The invokeai.yaml file was found but is missing the db_dir and/or outdir setting!")
-                return False
+            if db_dir is None:
+                db_dir = self.DEFAULT_DB_DIR
+                print(f"The invokeai.yaml file was found but is missing the db_dir setting! Defaulting to {db_dir}")
+            if outdir is None:
+                outdir = self.DEFAULT_OUTDIR
+                print(f"The invokeai.yaml file was found but is missing the outdir setting! Defaulting to {outdir}")
 
             if os.path.isabs(db_dir):
                 self.database_path = os.path.join(db_dir, self.DATABASE_FILENAME)

invokeai/backend/util/serialization.py

@@ -0,0 +1,37 @@
+from pathlib import Path
+from typing import Any, Optional, Union
+
+import torch
+from safetensors.torch import load_file
+
+
+def state_dict_to(
+    state_dict: dict[str, torch.Tensor], device: Optional[torch.device] = None, dtype: Optional[torch.dtype] = None
+) -> dict[str, torch.Tensor]:
+    new_state_dict: dict[str, torch.Tensor] = {}
+    for k, v in state_dict.items():
+        new_state_dict[k] = v.to(device=device, dtype=dtype, non_blocking=True)
+    return new_state_dict
+
+
+def load_state_dict(file_path: Union[str, Path], device: str = "cpu") -> Any:
+    """Load a state_dict from a file that may be in either PyTorch or safetensors format. The file format is inferred
+    from the file extension.
+    """
+    file_path = Path(file_path)
+
+    if file_path.suffix == ".safetensors":
+        state_dict = load_file(
+            file_path,
+            device=device,
+        )
+    else:
+        # weights_only=True is used to address a security vulnerability that allows arbitrary code execution.
+        # This option was first introduced in https://github.com/pytorch/pytorch/pull/86812.
+        #
+        # mmap=True is used to both reduce memory usage and speed up loading. This setting causes torch.load() to more
+        # closely mirror the behaviour of safetensors.torch.load_file().  This option was first introduced in
+        # https://github.com/pytorch/pytorch/pull/102549. The discussion on that PR provides helpful context.
+        state_dict = torch.load(file_path, map_location=device, weights_only=True, mmap=True)
+
+    return state_dict

invokeai/frontend/web/package.json

@@ -94,6 +94,7 @@
     "reactflow": "^11.10.4",
     "redux-dynamic-middlewares": "^2.2.0",
     "redux-remember": "^5.1.0",
+    "rfdc": "^1.3.1",
     "roarr": "^7.21.1",
     "serialize-error": "^11.0.3",
     "socket.io-client": "^4.7.5",

invokeai/frontend/web/pnpm-lock.yaml

@@ -137,6 +137,9 @@ dependencies:
   redux-remember:
     specifier: ^5.1.0
     version: 5.1.0(redux@5.0.1)
+  rfdc:
+    specifier: ^1.3.1
+    version: 1.3.1
   roarr:
     specifier: ^7.21.1
     version: 7.21.1
@@ -12128,6 +12131,10 @@ packages:
     resolution: {integrity: sha512-/x8uIPdTafBqakK0TmPNJzgkLP+3H+yxpUJhCQHsLBg1rYEVNR2D8BRYNWQhVBjyOd7oo1dZRVzIkwMY2oqfYQ==}
     dev: true
 
+  /rfdc@1.3.1:
+    resolution: {integrity: sha512-r5a3l5HzYlIC68TpmYKlxWjmOP6wiPJ1vWv2HeLhNsRZMrCkxeqxiHlQ21oXmQ4F3SiryXBHhAD7JZqvOJjFmg==}
+    dev: false
+
   /rimraf@2.6.3:
     resolution: {integrity: sha512-mwqeW5XsA2qAejG46gYdENaxXjx9onRNCfn7L0duuP4hCuTIi/QO7PDK07KJfp1d+izWPrzEJDcSqBa0OZQriA==}
     hasBin: true

invokeai/frontend/web/public/locales/de.json

@@ -4,7 +4,7 @@
         "reportBugLabel": "Fehler melden",
         "settingsLabel": "Einstellungen",
         "img2img": "Bild zu Bild",
-        "nodes": "Knoten Editor",
+        "nodes": "Arbeitsabläufe",
         "upload": "Hochladen",
         "load": "Laden",
         "statusDisconnected": "Getrennt",
@@ -74,7 +74,8 @@
         "updated": "Aktualisiert",
         "copy": "Kopieren",
         "aboutHeading": "Nutzen Sie Ihre kreative Energie",
-        "toResolve": "Lösen"
+        "toResolve": "Lösen",
+        "add": "Hinzufügen"
     },
     "gallery": {
         "galleryImageSize": "Bildgröße",
@@ -104,11 +105,16 @@
         "dropToUpload": "$t(gallery.drop) zum hochladen",
         "dropOrUpload": "$t(gallery.drop) oder hochladen",
         "drop": "Ablegen",
-        "problemDeletingImages": "Problem beim Löschen der Bilder"
+        "problemDeletingImages": "Problem beim Löschen der Bilder",
+        "bulkDownloadRequested": "Download vorbereiten",
+        "bulkDownloadRequestedDesc": "Dein Download wird vorbereitet. Dies kann ein paar Momente dauern.",
+        "bulkDownloadRequestFailed": "Problem beim Download vorbereiten",
+        "bulkDownloadFailed": "Download fehlgeschlagen",
+        "alwaysShowImageSizeBadge": "Zeige immer Bilder Größe Abzeichen"
     },
     "hotkeys": {
         "keyboardShortcuts": "Tastenkürzel",
-        "appHotkeys": "App-Tastenkombinationen",
+        "appHotkeys": "App",
         "generalHotkeys": "Allgemein",
         "galleryHotkeys": "Galerie",
         "unifiedCanvasHotkeys": "Leinwand",
@@ -757,7 +763,9 @@
         "scheduler": "Planer",
         "noRecallParameters": "Es wurden keine Parameter zum Abrufen gefunden",
         "recallParameters": "Parameter wiederherstellen",
-        "cfgRescaleMultiplier": "$t(parameters.cfgRescaleMultiplier)"
+        "cfgRescaleMultiplier": "$t(parameters.cfgRescaleMultiplier)",
+        "allPrompts": "Alle Prompts",
+        "imageDimensions": "Bilder Auslösungen"
     },
     "popovers": {
         "noiseUseCPU": {
@@ -1068,5 +1076,10 @@
     },
     "dynamicPrompts": {
         "showDynamicPrompts": "Dynamische Prompts anzeigen"
+    },
+    "prompt": {
+        "noMatchingTriggers": "Keine passenden Auslöser",
+        "addPromptTrigger": "Auslöse Text hinzufügen",
+        "compatibleEmbeddings": "Kompatible Einbettungen"
     }
 }

invokeai/frontend/web/public/locales/en.json

@@ -217,6 +217,7 @@
         "saveControlImage": "Save Control Image",
         "scribble": "scribble",
         "selectModel": "Select a model",
+        "selectCLIPVisionModel": "Select a CLIP Vision model",
         "setControlImageDimensions": "Set Control Image Dimensions To W/H",
         "showAdvanced": "Show Advanced",
         "small": "Small",
@@ -655,6 +656,7 @@
         "install": "Install",
         "installAll": "Install All",
         "installRepo": "Install Repo",
+        "ipAdapters": "IP Adapters",
         "load": "Load",
         "localOnly": "local only",
         "manual": "Manual",

invokeai/frontend/web/public/locales/it.json

@@ -73,7 +73,8 @@
         "ai": "ia",
         "file": "File",
         "toResolve": "Da risolvere",
-        "add": "Aggiungi"
+        "add": "Aggiungi",
+        "loglevel": "Livello di log"
     },
     "gallery": {
         "galleryImageSize": "Dimensione dell'immagine",
@@ -934,7 +935,9 @@
         "base": "Base",
         "lineart": "Linea",
         "controlnet": "$t(controlnet.controlAdapter_one) #{{number}} ($t(common.controlNet))",
-        "mediapipeFace": "Mediapipe Volto"
+        "mediapipeFace": "Mediapipe Volto",
+        "ip_adapter": "$t(controlnet.controlAdapter_one) #{{number}} ($t(common.ipAdapter))",
+        "t2i_adapter": "$t(controlnet.controlAdapter_one) #{{number}} ($t(common.t2iAdapter))"
     },
     "queue": {
         "queueFront": "Aggiungi all'inizio della coda",
@@ -1490,7 +1493,8 @@
             "title": "Generazione"
         },
         "advanced": {
-            "title": "Avanzate"
+            "title": "Avanzate",
+            "options": "Opzioni $t(accordions.advanced.title)"
         },
         "image": {
             "title": "Immagine"

invokeai/frontend/web/public/locales/ru.json

@@ -75,7 +75,8 @@
         "copy": "Копировать",
         "localSystem": "Локальная система",
         "aboutDesc": "Используя Invoke для работы? Проверьте это:",
-        "add": "Добавить"
+        "add": "Добавить",
+        "loglevel": "Уровень логов"
     },
     "gallery": {
         "galleryImageSize": "Размер изображений",
@@ -1505,7 +1506,8 @@
             "title": "Генерация"
         },
         "advanced": {
-            "title": "Расширенные"
+            "title": "Расширенные",
+            "options": "Опции $t(accordions.advanced.title)"
         },
         "image": {
             "title": "Изображение"

invokeai/frontend/web/src/app/store/middleware/devtools/actionSanitizer.ts

@@ -1,7 +1,7 @@
 import type { UnknownAction } from '@reduxjs/toolkit';
+import { deepClone } from 'common/util/deepClone';
 import { isAnyGraphBuilt } from 'features/nodes/store/actions';
 import { nodeTemplatesBuilt } from 'features/nodes/store/nodesSlice';
-import { cloneDeep } from 'lodash-es';
 import { appInfoApi } from 'services/api/endpoints/appInfo';
 import type { Graph } from 'services/api/types';
 import { socketGeneratorProgress } from 'services/events/actions';
@@ -33,7 +33,7 @@ export const actionSanitizer = <A extends UnknownAction>(action: A): A => {
   }
 
   if (socketGeneratorProgress.match(action)) {
-    const sanitized = cloneDeep(action);
+    const sanitized = deepClone(action);
     if (sanitized.payload.data.progress_image) {
       sanitized.payload.data.progress_image.dataURL = '<Progress image omitted>';
     }

invokeai/frontend/web/src/app/store/middleware/listenerMiddleware/listeners/socketio/socketModelInstall.ts

@@ -43,6 +43,7 @@ export const addModelInstallEventListener = (startAppListening: AppStartListenin
         })
       );
       dispatch(api.util.invalidateTags([{ type: 'ModelConfig', id: LIST_TAG }]));
+      dispatch(api.util.invalidateTags([{ type: 'ModelScanFolderResults', id: LIST_TAG }]));
     },
   });
 

invokeai/frontend/web/src/common/components/OverlayScrollbars/constants.ts

@@ -1,4 +1,5 @@
-import { cloneDeep, merge } from 'lodash-es';
+import { deepClone } from 'common/util/deepClone';
+import { merge } from 'lodash-es';
 import { ClickScrollPlugin, OverlayScrollbars } from 'overlayscrollbars';
 import type { UseOverlayScrollbarsParams } from 'overlayscrollbars-react';
 
@@ -22,7 +23,7 @@ export const getOverlayScrollbarsParams = (
   overflowX: 'hidden' | 'scroll' = 'hidden',
   overflowY: 'hidden' | 'scroll' = 'scroll'
 ) => {
-  const params = cloneDeep(overlayScrollbarsParams);
+  const params = deepClone(overlayScrollbarsParams);
   merge(params, { options: { overflow: { y: overflowY, x: overflowX } } });
   return params;
 };

invokeai/frontend/web/src/common/util/deepClone.ts

@@ -0,0 +1,15 @@
+import rfdc from 'rfdc';
+const _rfdc = rfdc();
+
+/**
+ * Deep-clones an object using Really Fast Deep Clone.
+ * This is the fastest deep clone library on Chrome, but not the fastest on FF. Still, it's much faster than lodash
+ * and structuredClone, so it's the best all-around choice.
+ *
+ * Simple Benchmark: https://www.measurethat.net/Benchmarks/Show/30358/0/lodash-clonedeep-vs-jsonparsejsonstringify-vs-recursive
+ * Repo: https://github.com/davidmarkclements/rfdc
+ *
+ * @param obj The object to deep-clone
+ * @returns The cloned object
+ */
+export const deepClone = <T>(obj: T): T => _rfdc(obj);

invokeai/frontend/web/src/features/canvas/store/canvasSlice.ts

@@ -1,6 +1,7 @@
 import type { PayloadAction } from '@reduxjs/toolkit';
 import { createSlice } from '@reduxjs/toolkit';
 import type { PersistConfig, RootState } from 'app/store/store';
+import { deepClone } from 'common/util/deepClone';
 import { roundDownToMultiple, roundToMultiple } from 'common/util/roundDownToMultiple';
 import calculateCoordinates from 'features/canvas/util/calculateCoordinates';
 import calculateScale from 'features/canvas/util/calculateScale';
@@ -13,7 +14,7 @@ import { modelChanged } from 'features/parameters/store/generationSlice';
 import type { PayloadActionWithOptimalDimension } from 'features/parameters/store/types';
 import { getIsSizeOptimal, getOptimalDimension } from 'features/parameters/util/optimalDimension';
 import type { IRect, Vector2d } from 'konva/lib/types';
-import { clamp, cloneDeep } from 'lodash-es';
+import { clamp } from 'lodash-es';
 import type { RgbaColor } from 'react-colorful';
 import { queueApi } from 'services/api/endpoints/queue';
 import type { ImageDTO } from 'services/api/types';
@@ -36,7 +37,7 @@ import { CANVAS_GRID_SIZE_FINE } from './constants';
 /**
  * The maximum history length to keep in the past/future layer states.
  */
-const MAX_HISTORY = 128;
+const MAX_HISTORY = 100;
 
 const initialLayerState: CanvasLayerState = {
   objects: [],
@@ -121,7 +122,7 @@ export const canvasSlice = createSlice({
       state.brushSize = action.payload;
     },
     clearMask: (state) => {
-      state.pastLayerStates.push(cloneDeep(state.layerState));
+      pushToPrevLayerStates(state);
       state.layerState.objects = state.layerState.objects.filter((obj) => !isCanvasMaskLine(obj));
       state.futureLayerStates = [];
       state.shouldPreserveMaskedArea = false;
@@ -163,10 +164,10 @@ export const canvasSlice = createSlice({
         state.boundingBoxDimensions = newBoundingBoxDimensions;
         state.boundingBoxCoordinates = newBoundingBoxCoordinates;
 
-        state.pastLayerStates.push(cloneDeep(state.layerState));
+        pushToPrevLayerStates(state);
 
         state.layerState = {
-          ...cloneDeep(initialLayerState),
+          ...deepClone(initialLayerState),
           objects: [
             {
               kind: 'image',
@@ -261,11 +262,7 @@ export const canvasSlice = createSlice({
         return;
       }
 
-      state.pastLayerStates.push(cloneDeep(state.layerState));
-
-      if (state.pastLayerStates.length > MAX_HISTORY) {
-        state.pastLayerStates.shift();
-      }
+      pushToPrevLayerStates(state);
 
       state.layerState.stagingArea.images.push({
         kind: 'image',
@@ -279,13 +276,9 @@ export const canvasSlice = createSlice({
       state.futureLayerStates = [];
     },
     discardStagedImages: (state) => {
-      state.pastLayerStates.push(cloneDeep(state.layerState));
+      pushToPrevLayerStates(state);
 
-      if (state.pastLayerStates.length > MAX_HISTORY) {
-        state.pastLayerStates.shift();
-      }
-
-      state.layerState.stagingArea = cloneDeep(cloneDeep(initialLayerState)).stagingArea;
+      state.layerState.stagingArea = deepClone(initialLayerState.stagingArea);
 
       state.futureLayerStates = [];
       state.shouldShowStagingOutline = true;
@@ -294,18 +287,21 @@ export const canvasSlice = createSlice({
     },
     discardStagedImage: (state) => {
       const { images, selectedImageIndex } = state.layerState.stagingArea;
-      state.pastLayerStates.push(cloneDeep(state.layerState));
-
-      if (state.pastLayerStates.length > MAX_HISTORY) {
-        state.pastLayerStates.shift();
-      }
-
-      if (!images.length) {
-        return;
-      }
+      pushToPrevLayerStates(state);
 
       images.splice(selectedImageIndex, 1);
 
+      if (images.length === 0) {
+        pushToPrevLayerStates(state);
+
+        state.layerState.stagingArea = deepClone(initialLayerState.stagingArea);
+
+        state.futureLayerStates = [];
+        state.shouldShowStagingOutline = true;
+        state.shouldShowStagingImage = true;
+        state.batchIds = [];
+      }
+
       if (selectedImageIndex >= images.length) {
         state.layerState.stagingArea.selectedImageIndex = images.length - 1;
       }
@@ -320,11 +316,7 @@ export const canvasSlice = createSlice({
     addFillRect: (state) => {
       const { boundingBoxCoordinates, boundingBoxDimensions, brushColor } = state;
 
-      state.pastLayerStates.push(cloneDeep(state.layerState));
-
-      if (state.pastLayerStates.length > MAX_HISTORY) {
-        state.pastLayerStates.shift();
-      }
+      pushToPrevLayerStates(state);
 
       state.layerState.objects.push({
         kind: 'fillRect',
@@ -339,11 +331,7 @@ export const canvasSlice = createSlice({
     addEraseRect: (state) => {
       const { boundingBoxCoordinates, boundingBoxDimensions } = state;
 
-      state.pastLayerStates.push(cloneDeep(state.layerState));
-
-      if (state.pastLayerStates.length > MAX_HISTORY) {
-        state.pastLayerStates.shift();
-      }
+      pushToPrevLayerStates(state);
 
       state.layerState.objects.push({
         kind: 'eraseRect',
@@ -367,11 +355,7 @@ export const canvasSlice = createSlice({
       // set & then spread this to only conditionally add the "color" key
       const newColor = layer === 'base' && tool === 'brush' ? { color: brushColor } : {};
 
-      state.pastLayerStates.push(cloneDeep(state.layerState));
-
-      if (state.pastLayerStates.length > MAX_HISTORY) {
-        state.pastLayerStates.shift();
-      }
+      pushToPrevLayerStates(state);
 
       const newLine: CanvasMaskLine | CanvasBaseLine = {
         kind: 'line',
@@ -409,11 +393,7 @@ export const canvasSlice = createSlice({
         return;
       }
 
-      state.futureLayerStates.unshift(cloneDeep(state.layerState));
-
-      if (state.futureLayerStates.length > MAX_HISTORY) {
-        state.futureLayerStates.pop();
-      }
+      pushToFutureLayerStates(state);
 
       state.layerState = targetState;
     },
@@ -424,11 +404,7 @@ export const canvasSlice = createSlice({
         return;
       }
 
-      state.pastLayerStates.push(cloneDeep(state.layerState));
-
-      if (state.pastLayerStates.length > MAX_HISTORY) {
-        state.pastLayerStates.shift();
-      }
+      pushToPrevLayerStates(state);
 
       state.layerState = targetState;
     },
@@ -445,8 +421,8 @@ export const canvasSlice = createSlice({
       state.shouldShowIntermediates = action.payload;
     },
     resetCanvas: (state) => {
-      state.pastLayerStates.push(cloneDeep(state.layerState));
-      state.layerState = cloneDeep(initialLayerState);
+      pushToPrevLayerStates(state);
+      state.layerState = deepClone(initialLayerState);
       state.futureLayerStates = [];
       state.batchIds = [];
       state.boundingBoxCoordinates = {
@@ -540,11 +516,7 @@ export const canvasSlice = createSlice({
 
       const { images, selectedImageIndex } = state.layerState.stagingArea;
 
-      state.pastLayerStates.push(cloneDeep(state.layerState));
-
-      if (state.pastLayerStates.length > MAX_HISTORY) {
-        state.pastLayerStates.shift();
-      }
+      pushToPrevLayerStates(state);
 
       const imageToCommit = images[selectedImageIndex];
 
@@ -553,7 +525,7 @@ export const canvasSlice = createSlice({
           ...imageToCommit,
         });
       }
-      state.layerState.stagingArea = cloneDeep(initialLayerState).stagingArea;
+      state.layerState.stagingArea = deepClone(initialLayerState.stagingArea);
 
       state.futureLayerStates = [];
       state.shouldShowStagingOutline = true;
@@ -623,7 +595,7 @@ export const canvasSlice = createSlice({
       };
     },
     setMergedCanvas: (state, action: PayloadAction<CanvasImage>) => {
-      state.pastLayerStates.push(cloneDeep(state.layerState));
+      pushToPrevLayerStates(state);
 
       state.futureLayerStates = [];
 
@@ -743,3 +715,17 @@ export const canvasPersistConfig: PersistConfig<CanvasState> = {
   migrate: migrateCanvasState,
   persistDenylist: [],
 };
+
+const pushToPrevLayerStates = (state: CanvasState) => {
+  state.pastLayerStates.push(deepClone(state.layerState));
+  if (state.pastLayerStates.length > MAX_HISTORY) {
+    state.pastLayerStates = state.pastLayerStates.slice(-MAX_HISTORY);
+  }
+};
+
+const pushToFutureLayerStates = (state: CanvasState) => {
+  state.futureLayerStates.unshift(deepClone(state.layerState));
+  if (state.futureLayerStates.length > MAX_HISTORY) {
+    state.futureLayerStates = state.futureLayerStates.slice(0, MAX_HISTORY);
+  }
+};

invokeai/frontend/web/src/features/controlAdapters/components/parameters/ParamControlAdapterModel.tsx

@@ -1,12 +1,18 @@
-import { Combobox, FormControl, Tooltip } from '@invoke-ai/ui-library';
+import type { ComboboxOnChange, ComboboxOption } from '@invoke-ai/ui-library';
+import { Combobox, Flex, FormControl, Tooltip } from '@invoke-ai/ui-library';
 import { createMemoizedSelector } from 'app/store/createMemoizedSelector';
 import { useAppDispatch, useAppSelector } from 'app/store/storeHooks';
 import { useGroupedModelCombobox } from 'common/hooks/useGroupedModelCombobox';
+import { useControlAdapterCLIPVisionModel } from 'features/controlAdapters/hooks/useControlAdapterCLIPVisionModel';
 import { useControlAdapterIsEnabled } from 'features/controlAdapters/hooks/useControlAdapterIsEnabled';
 import { useControlAdapterModel } from 'features/controlAdapters/hooks/useControlAdapterModel';
 import { useControlAdapterModels } from 'features/controlAdapters/hooks/useControlAdapterModels';
 import { useControlAdapterType } from 'features/controlAdapters/hooks/useControlAdapterType';
-import { controlAdapterModelChanged } from 'features/controlAdapters/store/controlAdaptersSlice';
+import {
+  controlAdapterCLIPVisionModelChanged,
+  controlAdapterModelChanged,
+} from 'features/controlAdapters/store/controlAdaptersSlice';
+import type { CLIPVisionModel } from 'features/controlAdapters/store/types';
 import { selectGenerationSlice } from 'features/parameters/store/generationSlice';
 import { memo, useCallback, useMemo } from 'react';
 import { useTranslation } from 'react-i18next';
@@ -29,6 +35,7 @@ const ParamControlAdapterModel = ({ id }: ParamControlAdapterModelProps) => {
   const { modelConfig } = useControlAdapterModel(id);
   const dispatch = useAppDispatch();
   const currentBaseModel = useAppSelector((s) => s.generation.model?.base);
+  const currentCLIPVisionModel = useControlAdapterCLIPVisionModel(id);
   const mainModel = useAppSelector(selectMainModel);
   const { t } = useTranslation();
 
@@ -49,6 +56,16 @@ const ParamControlAdapterModel = ({ id }: ParamControlAdapterModelProps) => {
     [dispatch, id]
   );
 
+  const onCLIPVisionModelChange = useCallback<ComboboxOnChange>(
+    (v) => {
+      if (!v?.value) {
+        return;
+      }
+      dispatch(controlAdapterCLIPVisionModelChanged({ id, clipVisionModel: v.value as CLIPVisionModel }));
+    },
+    [dispatch, id]
+  );
+
   const selectedModel = useMemo(
     () => (modelConfig && controlAdapterType ? { ...modelConfig, model_type: controlAdapterType } : null),
     [controlAdapterType, modelConfig]
@@ -71,18 +88,51 @@ const ParamControlAdapterModel = ({ id }: ParamControlAdapterModelProps) => {
     isLoading,
   });
 
+  const clipVisionOptions = useMemo<ComboboxOption[]>(
+    () => [
+      { label: 'ViT-H', value: 'ViT-H' },
+      { label: 'ViT-G', value: 'ViT-G' },
+    ],
+    []
+  );
+
+  const clipVisionModel = useMemo(
+    () => clipVisionOptions.find((o) => o.value === currentCLIPVisionModel),
+    [clipVisionOptions, currentCLIPVisionModel]
+  );
+
   return (
-    <Tooltip label={value?.description}>
-      <FormControl isDisabled={!isEnabled} isInvalid={!value || mainModel?.base !== modelConfig?.base}>
-        <Combobox
-          options={options}
-          placeholder={t('controlnet.selectModel')}
-          value={value}
-          onChange={onChange}
-          noOptionsMessage={noOptionsMessage}
-        />
-      </FormControl>
-    </Tooltip>
+    <Flex sx={{ gap: 2 }}>
+      <Tooltip label={value?.description}>
+        <FormControl
+          isDisabled={!isEnabled}
+          isInvalid={!value || mainModel?.base !== modelConfig?.base}
+          sx={{ width: '100%' }}
+        >
+          <Combobox
+            options={options}
+            placeholder={t('controlnet.selectModel')}
+            value={value}
+            onChange={onChange}
+            noOptionsMessage={noOptionsMessage}
+          />
+        </FormControl>
+      </Tooltip>
+      {modelConfig?.type === 'ip_adapter' && modelConfig.format === 'checkpoint' && (
+        <FormControl
+          isDisabled={!isEnabled}
+          isInvalid={!value || mainModel?.base !== modelConfig?.base}
+          sx={{ width: 'max-content', minWidth: 28 }}
+        >
+          <Combobox
+            options={clipVisionOptions}
+            placeholder={t('controlnet.selectCLIPVisionModel')}
+            value={clipVisionModel}
+            onChange={onCLIPVisionModelChange}
+          />
+        </FormControl>
+      )}
+    </Flex>
   );
 };
 

invokeai/frontend/web/src/features/controlAdapters/hooks/useControlAdapterCLIPVisionModel.ts

@@ -0,0 +1,24 @@
+import { createMemoizedSelector } from 'app/store/createMemoizedSelector';
+import { useAppSelector } from 'app/store/storeHooks';
+import {
+  selectControlAdapterById,
+  selectControlAdaptersSlice,
+} from 'features/controlAdapters/store/controlAdaptersSlice';
+import { useMemo } from 'react';
+
+export const useControlAdapterCLIPVisionModel = (id: string) => {
+  const selector = useMemo(
+    () =>
+      createMemoizedSelector(selectControlAdaptersSlice, (controlAdapters) => {
+        const cn = selectControlAdapterById(controlAdapters, id);
+        if (cn && cn?.type === 'ip_adapter') {
+          return cn.clipVisionModel;
+        }
+      }),
+    [id]
+  );
+
+  const clipVisionModel = useAppSelector(selector);
+
+  return clipVisionModel;
+};

invokeai/frontend/web/src/features/controlAdapters/store/controlAdaptersSlice.ts

@@ -2,10 +2,11 @@ import type { PayloadAction, Update } from '@reduxjs/toolkit';
 import { createEntityAdapter, createSlice, isAnyOf } from '@reduxjs/toolkit';
 import { getSelectorsOptions } from 'app/store/createMemoizedSelector';
 import type { PersistConfig, RootState } from 'app/store/store';
+import { deepClone } from 'common/util/deepClone';
 import { buildControlAdapter } from 'features/controlAdapters/util/buildControlAdapter';
 import { buildControlAdapterProcessor } from 'features/controlAdapters/util/buildControlAdapterProcessor';
 import { zModelIdentifierField } from 'features/nodes/types/common';
-import { cloneDeep, merge, uniq } from 'lodash-es';
+import { merge, uniq } from 'lodash-es';
 import type { ControlNetModelConfig, IPAdapterModelConfig, T2IAdapterModelConfig } from 'services/api/types';
 import { socketInvocationError } from 'services/events/actions';
 import { v4 as uuidv4 } from 'uuid';
@@ -13,6 +14,7 @@ import { v4 as uuidv4 } from 'uuid';
 import { controlAdapterImageProcessed } from './actions';
 import { CONTROLNET_PROCESSORS } from './constants';
 import type {
+  CLIPVisionModel,
   ControlAdapterConfig,
   ControlAdapterProcessorType,
   ControlAdaptersState,
@@ -114,7 +116,7 @@ export const controlAdaptersSlice = createSlice({
         if (!controlAdapter) {
           return;
         }
-        const newControlAdapter = merge(cloneDeep(controlAdapter), {
+        const newControlAdapter = merge(deepClone(controlAdapter), {
           id: newId,
           isEnabled: true,
         });
@@ -243,6 +245,13 @@ export const controlAdaptersSlice = createSlice({
       }
       caAdapter.updateOne(state, { id, changes: { controlMode } });
     },
+    controlAdapterCLIPVisionModelChanged: (
+      state,
+      action: PayloadAction<{ id: string; clipVisionModel: CLIPVisionModel }>
+    ) => {
+      const { id, clipVisionModel } = action.payload;
+      caAdapter.updateOne(state, { id, changes: { clipVisionModel } });
+    },
     controlAdapterResizeModeChanged: (
       state,
       action: PayloadAction<{
@@ -270,7 +279,7 @@ export const controlAdaptersSlice = createSlice({
         return;
       }
 
-      const processorNode = merge(cloneDeep(cn.processorNode), params);
+      const processorNode = merge(deepClone(cn.processorNode), params);
 
       caAdapter.updateOne(state, {
         id,
@@ -293,7 +302,7 @@ export const controlAdaptersSlice = createSlice({
         return;
       }
 
-      const processorNode = cloneDeep(
+      const processorNode = deepClone(
         CONTROLNET_PROCESSORS[processorType].buildDefaults(cn.model?.base)
       ) as RequiredControlAdapterProcessorNode;
 
@@ -333,7 +342,7 @@ export const controlAdaptersSlice = createSlice({
       caAdapter.updateOne(state, update);
     },
     controlAdaptersReset: () => {
-      return cloneDeep(initialControlAdaptersState);
+      return deepClone(initialControlAdaptersState);
     },
     pendingControlImagesCleared: (state) => {
       state.pendingControlImages = [];
@@ -380,6 +389,7 @@ export const {
   controlAdapterProcessedImageChanged,
   controlAdapterIsEnabledChanged,
   controlAdapterModelChanged,
+  controlAdapterCLIPVisionModelChanged,
   controlAdapterWeightChanged,
   controlAdapterBeginStepPctChanged,
   controlAdapterEndStepPctChanged,
@@ -406,7 +416,7 @@ const migrateControlAdaptersState = (state: any): any => {
     state._version = 1;
   }
   if (state._version === 1) {
-    state = cloneDeep(initialControlAdaptersState);
+    state = deepClone(initialControlAdaptersState);
   }
   return state;
 };

invokeai/frontend/web/src/features/controlAdapters/store/types.ts

@@ -243,12 +243,15 @@ export type T2IAdapterConfig = {
   shouldAutoConfig: boolean;
 };
 
+export type CLIPVisionModel = 'ViT-H' | 'ViT-G';
+
 export type IPAdapterConfig = {
   type: 'ip_adapter';
   id: string;
   isEnabled: boolean;
   controlImage: string | null;
   model: ParameterIPAdapterModel | null;
+  clipVisionModel: CLIPVisionModel;
   weight: number;
   beginStepPct: number;
   endStepPct: number;

invokeai/frontend/web/src/features/controlAdapters/util/buildControlAdapter.ts

@@ -1,3 +1,4 @@
+import { deepClone } from 'common/util/deepClone';
 import { CONTROLNET_PROCESSORS } from 'features/controlAdapters/store/constants';
 import type {
   ControlAdapterConfig,
@@ -7,7 +8,7 @@ import type {
   RequiredCannyImageProcessorInvocation,
   T2IAdapterConfig,
 } from 'features/controlAdapters/store/types';
-import { cloneDeep, merge } from 'lodash-es';
+import { merge } from 'lodash-es';
 
 export const initialControlNet: Omit<ControlNetConfig, 'id'> = {
   type: 'controlnet',
@@ -45,6 +46,7 @@ export const initialIPAdapter: Omit<IPAdapterConfig, 'id'> = {
   isEnabled: true,
   controlImage: null,
   model: null,
+  clipVisionModel: 'ViT-H',
   weight: 1,
   beginStepPct: 0,
   endStepPct: 1,
@@ -57,11 +59,11 @@ export const buildControlAdapter = (
 ): ControlAdapterConfig => {
   switch (type) {
     case 'controlnet':
-      return merge(cloneDeep(initialControlNet), { id, ...overrides });
+      return merge(deepClone(initialControlNet), { id, ...overrides });
     case 't2i_adapter':
-      return merge(cloneDeep(initialT2IAdapter), { id, ...overrides });
+      return merge(deepClone(initialT2IAdapter), { id, ...overrides });
     case 'ip_adapter':
-      return merge(cloneDeep(initialIPAdapter), { id, ...overrides });
+      return merge(deepClone(initialIPAdapter), { id, ...overrides });
     default:
       throw new Error(`Unknown control adapter type: ${type}`);
   }

invokeai/frontend/web/src/features/lora/store/loraSlice.ts

@@ -1,9 +1,9 @@
 import type { PayloadAction } from '@reduxjs/toolkit';
 import { createSlice } from '@reduxjs/toolkit';
 import type { PersistConfig, RootState } from 'app/store/store';
+import { deepClone } from 'common/util/deepClone';
 import { zModelIdentifierField } from 'features/nodes/types/common';
 import type { ParameterLoRAModel } from 'features/parameters/types/parameterSchemas';
-import { cloneDeep } from 'lodash-es';
 import type { LoRAModelConfig } from 'services/api/types';
 
 export type LoRA = {
@@ -58,7 +58,7 @@ export const loraSlice = createSlice({
       }
       lora.isEnabled = isEnabled;
     },
-    lorasReset: () => cloneDeep(initialLoraState),
+    lorasReset: () => deepClone(initialLoraState),
   },
 });
 
@@ -74,7 +74,7 @@ const migrateLoRAState = (state: any): any => {
   }
   if (state._version === 1) {
     // Model type has changed, so we need to reset the state - too risky to migrate
-    state = cloneDeep(initialLoraState);
+    state = deepClone(initialLoraState);
   }
   return state;
 };

invokeai/frontend/web/src/features/metadata/util/parsers.ts

@@ -372,6 +372,7 @@ const parseIPAdapter: MetadataParseFunc<IPAdapterConfigMetadata> = async (metada
     type: 'ip_adapter',
     isEnabled: true,
     model: zModelIdentifierField.parse(ipAdapterModel),
+    clipVisionModel: 'ViT-H',
     controlImage: image?.image_name ?? null,
     weight: weight ?? initialIPAdapter.weight,
     beginStepPct: begin_step_percent ?? initialIPAdapter.beginStepPct,

invokeai/frontend/web/src/features/modelManagerV2/subpanels/AddModelPanel/ModelInstallQueue/ModelInstallQueueItem.tsx

@@ -87,6 +87,10 @@ export const ModelInstallQueueItem = (props: ModelListItemProps) => {
   }, [installJob.source]);
 
   const progressValue = useMemo(() => {
+    if (installJob.status === 'completed' || installJob.status === 'error' || installJob.status === 'cancelled') {
+      return 100;
+    }
+
     if (isNil(installJob.bytes) || isNil(installJob.total_bytes)) {
       return null;
     }
@@ -96,7 +100,7 @@ export const ModelInstallQueueItem = (props: ModelListItemProps) => {
     }
 
     return (installJob.bytes / installJob.total_bytes) * 100;
-  }, [installJob.bytes, installJob.total_bytes]);
+  }, [installJob.bytes, installJob.status, installJob.total_bytes]);
 
   return (
     <Flex gap={3} w="full" alignItems="center">

invokeai/frontend/web/src/features/modelManagerV2/subpanels/AddModelPanel/ScanFolder/ScanFolderResultItem.tsx

@@ -1,48 +1,19 @@
 import { Badge, Box, Flex, IconButton, Text } from '@invoke-ai/ui-library';
-import { useAppDispatch } from 'app/store/storeHooks';
-import { addToast } from 'features/system/store/systemSlice';
-import { makeToast } from 'features/system/util/makeToast';
 import { useCallback } from 'react';
 import { useTranslation } from 'react-i18next';
 import { PiPlusBold } from 'react-icons/pi';
 import type { ScanFolderResponse } from 'services/api/endpoints/models';
-import { useInstallModelMutation } from 'services/api/endpoints/models';
 
 type Props = {
   result: ScanFolderResponse[number];
+  installModel: (source: string) => void;
 };
-export const ScanModelResultItem = ({ result }: Props) => {
+export const ScanModelResultItem = ({ result, installModel }: Props) => {
   const { t } = useTranslation();
-  const dispatch = useAppDispatch();
 
-  const [installModel] = useInstallModelMutation();
-
-  const handleQuickAdd = useCallback(() => {
-    installModel({ source: result.path })
-      .unwrap()
-      .then((_) => {
-        dispatch(
-          addToast(
-            makeToast({
-              title: t('toast.modelAddedSimple'),
-              status: 'success',
-            })
-          )
-        );
-      })
-      .catch((error) => {
-        if (error) {
-          dispatch(
-            addToast(
-              makeToast({
-                title: `${error.data.detail} `,
-                status: 'error',
-              })
-            )
-          );
-        }
-      });
-  }, [installModel, result, dispatch, t]);
+  const handleInstall = useCallback(() => {
+    installModel(result.path);
+  }, [installModel, result]);
 
   return (
     <Flex alignItems="center" justifyContent="space-between" w="100%" gap={3}>
@@ -54,7 +25,7 @@ export const ScanModelResultItem = ({ result }: Props) => {
         {result.is_installed ? (
           <Badge>{t('common.installed')}</Badge>
         ) : (
-          <IconButton aria-label={t('modelManager.install')} icon={<PiPlusBold />} onClick={handleQuickAdd} size="sm" />
+          <IconButton aria-label={t('modelManager.install')} icon={<PiPlusBold />} onClick={handleInstall} size="sm" />
         )}
       </Box>
     </Flex>

invokeai/frontend/web/src/features/modelManagerV2/subpanels/AddModelPanel/ScanFolder/ScanFolderResults.tsx

@@ -1,7 +1,10 @@
 import {
   Button,
+  Checkbox,
   Divider,
   Flex,
+  FormControl,
+  FormLabel,
   Heading,
   IconButton,
   Input,
@@ -12,7 +15,7 @@ import { useAppDispatch } from 'app/store/storeHooks';
 import ScrollableContent from 'common/components/OverlayScrollbars/ScrollableContent';
 import { addToast } from 'features/system/store/systemSlice';
 import { makeToast } from 'features/system/util/makeToast';
-import type { ChangeEventHandler } from 'react';
+import type { ChangeEvent, ChangeEventHandler } from 'react';
 import { useCallback, useMemo, useState } from 'react';
 import { useTranslation } from 'react-i18next';
 import { PiXBold } from 'react-icons/pi';
@@ -28,7 +31,7 @@ export const ScanModelsResults = ({ results }: ScanModelResultsProps) => {
   const { t } = useTranslation();
   const [searchTerm, setSearchTerm] = useState('');
   const dispatch = useAppDispatch();
-
+  const [inplace, setInplace] = useState(true);
   const [installModel] = useInstallModelMutation();
 
   const filteredResults = useMemo(() => {
@@ -42,6 +45,10 @@ export const ScanModelsResults = ({ results }: ScanModelResultsProps) => {
     setSearchTerm(e.target.value.trim());
   }, []);
 
+  const onChangeInplace = useCallback((e: ChangeEvent<HTMLInputElement>) => {
+    setInplace(e.target.checked);
+  }, []);
+
   const clearSearch = useCallback(() => {
     setSearchTerm('');
   }, []);
@@ -51,7 +58,7 @@ export const ScanModelsResults = ({ results }: ScanModelResultsProps) => {
       if (result.is_installed) {
         continue;
       }
-      installModel({ source: result.path })
+      installModel({ source: result.path, inplace })
         .unwrap()
         .then((_) => {
           dispatch(
@@ -76,7 +83,37 @@ export const ScanModelsResults = ({ results }: ScanModelResultsProps) => {
           }
         });
     }
-  }, [installModel, filteredResults, dispatch, t]);
+  }, [filteredResults, installModel, inplace, dispatch, t]);
+
+  const handleInstallOne = useCallback(
+    (source: string) => {
+      installModel({ source, inplace })
+        .unwrap()
+        .then((_) => {
+          dispatch(
+            addToast(
+              makeToast({
+                title: t('toast.modelAddedSimple'),
+                status: 'success',
+              })
+            )
+          );
+        })
+        .catch((error) => {
+          if (error) {
+            dispatch(
+              addToast(
+                makeToast({
+                  title: `${error.data.detail} `,
+                  status: 'error',
+                })
+              )
+            );
+          }
+        });
+    },
+    [installModel, inplace, dispatch, t]
+  );
 
   return (
     <>
@@ -85,6 +122,10 @@ export const ScanModelsResults = ({ results }: ScanModelResultsProps) => {
         <Flex justifyContent="space-between" alignItems="center">
           <Heading size="sm">{t('modelManager.scanResults')}</Heading>
           <Flex alignItems="center" gap={3}>
+            <FormControl w="min-content">
+              <FormLabel m={0}>{t('modelManager.inplaceInstall')}</FormLabel>
+              <Checkbox isChecked={inplace} onChange={onChangeInplace} size="md" />
+            </FormControl>
             <Button size="sm" onClick={handleAddAll} isDisabled={filteredResults.length === 0}>
               {t('modelManager.installAll')}
             </Button>
@@ -116,7 +157,7 @@ export const ScanModelsResults = ({ results }: ScanModelResultsProps) => {
           <ScrollableContent>
             <Flex flexDir="column" gap={3}>
               {filteredResults.map((result) => (
-                <ScanModelResultItem key={result.path} result={result} />
+                <ScanModelResultItem key={result.path} result={result} installModel={handleInstallOne} />
               ))}
             </Flex>
           </ScrollableContent>

invokeai/frontend/web/src/features/modelManagerV2/subpanels/ModelManagerPanel/ModelListItem.tsx

@@ -90,11 +90,13 @@ const ModelListItem = (props: ModelListItemProps) => {
       cursor="pointer"
       onClick={handleSelectModel}
     >
-      <Flex gap={2} w="full" h="full">
+      <Flex gap={2} w="full" h="full" minW={0}>
         <ModelImage image_url={model.cover_image} />
-        <Flex gap={1} alignItems="flex-start" flexDir="column" w="full">
+        <Flex gap={1} alignItems="flex-start" flexDir="column" w="full" minW={0}>
           <Flex gap={2} w="full" alignItems="flex-start">
-            <Text fontWeight="semibold">{model.name}</Text>
+            <Text fontWeight="semibold" noOfLines={1} wordBreak="break-all">
+              {model.name}
+            </Text>
             <Spacer />
           </Flex>
           <Text variant="subtext" noOfLines={1}>

invokeai/frontend/web/src/features/modelManagerV2/subpanels/ModelPanel/Model.tsx

@@ -87,9 +87,9 @@ export const Model = () => {
     <Flex flexDir="column" gap={4}>
       <Flex alignItems="flex-start" gap={4}>
         <ModelImageUpload model_key={selectedModelKey} model_image={data.cover_image} />
-        <Flex flexDir="column" gap={1} flexGrow={1}>
+        <Flex flexDir="column" gap={1} flexGrow={1} minW={0}>
           <Flex gap={2}>
-            <Heading as="h2" fontSize="lg">
+            <Heading as="h2" fontSize="lg" noOfLines={1} wordBreak="break-all">
               {data.name}
             </Heading>
             <Spacer />
@@ -114,7 +114,7 @@ export const Model = () => {
             )}
           </Flex>
           {data.source && (
-            <Text variant="subtext">
+            <Text variant="subtext" noOfLines={1} wordBreak="break-all">
               {t('modelManager.source')}: {data?.source}
             </Text>
           )}

invokeai/frontend/web/src/features/modelManagerV2/subpanels/ModelPanel/ModelAttrView.tsx

@@ -9,7 +9,9 @@ export const ModelAttrView = ({ label, value }: Props) => {
   return (
     <FormControl flexDir="column" alignItems="flex-start" gap={0}>
       <FormLabel>{label}</FormLabel>
-      <Text fontSize="md">{value || '-'}</Text>
+      <Text fontSize="md" noOfLines={1} wordBreak="break-all">
+        {value || '-'}
+      </Text>
     </FormControl>
   );
 };

invokeai/frontend/web/src/features/modelManagerV2/subpanels/ModelPanel/ModelView.tsx

@@ -53,7 +53,7 @@ export const ModelView = () => {
             </>
           )}
 
-          {data.type === 'ip_adapter' && (
+          {data.type === 'ip_adapter' && data.format === 'invokeai' && (
             <Flex gap={2}>
               <ModelAttrView label={t('modelManager.imageEncoderModelId')} value={data.image_encoder_model_id} />
             </Flex>

invokeai/frontend/web/src/features/nodes/store/nodesSlice.ts

@@ -1,6 +1,7 @@
 import type { PayloadAction } from '@reduxjs/toolkit';
 import { createSlice, isAnyOf } from '@reduxjs/toolkit';
 import type { PersistConfig, RootState } from 'app/store/store';
+import { deepClone } from 'common/util/deepClone';
 import { workflowLoaded } from 'features/nodes/store/actions';
 import { SHARED_NODE_PROPERTIES } from 'features/nodes/types/constants';
 import type {
@@ -44,7 +45,7 @@ import {
 } from 'features/nodes/types/field';
 import type { AnyNode, InvocationTemplate, NodeExecutionState } from 'features/nodes/types/invocation';
 import { isInvocationNode, isNotesNode, zNodeStatus } from 'features/nodes/types/invocation';
-import { cloneDeep, forEach } from 'lodash-es';
+import { forEach } from 'lodash-es';
 import type {
   Connection,
   Edge,
@@ -571,8 +572,23 @@ export const nodesSlice = createSlice({
       );
     },
     selectionCopied: (state) => {
-      state.nodesToCopy = state.nodes.filter((n) => n.selected).map(cloneDeep);
-      state.edgesToCopy = state.edges.filter((e) => e.selected).map(cloneDeep);
+      const nodesToCopy: AnyNode[] = [];
+      const edgesToCopy: Edge[] = [];
+
+      for (const node of state.nodes) {
+        if (node.selected) {
+          nodesToCopy.push(deepClone(node));
+        }
+      }
+
+      for (const edge of state.edges) {
+        if (edge.selected) {
+          edgesToCopy.push(deepClone(edge));
+        }
+      }
+
+      state.nodesToCopy = nodesToCopy;
+      state.edgesToCopy = edgesToCopy;
 
       if (state.nodesToCopy.length > 0) {
         const averagePosition = { x: 0, y: 0 };
@@ -594,11 +610,21 @@ export const nodesSlice = createSlice({
     },
     selectionPasted: (state, action: PayloadAction<{ cursorPosition?: XYPosition }>) => {
       const { cursorPosition } = action.payload;
-      const newNodes = state.nodesToCopy.map(cloneDeep);
+      const newNodes: AnyNode[] = [];
+
+      for (const node of state.nodesToCopy) {
+        newNodes.push(deepClone(node));
+      }
+
       const oldNodeIds = newNodes.map((n) => n.data.id);
-      const newEdges = state.edgesToCopy
-        .filter((e) => oldNodeIds.includes(e.source) && oldNodeIds.includes(e.target))
-        .map(cloneDeep);
+
+      const newEdges: Edge[] = [];
+
+      for (const edge of state.edgesToCopy) {
+        if (oldNodeIds.includes(edge.source) && oldNodeIds.includes(edge.target)) {
+          newEdges.push(deepClone(edge));
+        }
+      }
 
       newEdges.forEach((e) => (e.selected = true));
 

invokeai/frontend/web/src/features/nodes/store/workflowSlice.ts

@@ -1,6 +1,7 @@
 import type { PayloadAction } from '@reduxjs/toolkit';
 import { createSlice } from '@reduxjs/toolkit';
 import type { PersistConfig, RootState } from 'app/store/store';
+import { deepClone } from 'common/util/deepClone';
 import { workflowLoaded } from 'features/nodes/store/actions';
 import { isAnyNodeOrEdgeMutation, nodeEditorReset, nodesChanged, nodesDeleted } from 'features/nodes/store/nodesSlice';
 import type {
@@ -11,7 +12,7 @@ import type {
 import type { FieldIdentifier } from 'features/nodes/types/field';
 import { isInvocationNode } from 'features/nodes/types/invocation';
 import type { WorkflowCategory, WorkflowV3 } from 'features/nodes/types/workflow';
-import { cloneDeep, isEqual, omit, uniqBy } from 'lodash-es';
+import { isEqual, omit, uniqBy } from 'lodash-es';
 
 const blankWorkflow: Omit<WorkflowV3, 'nodes' | 'edges'> = {
   name: '',
@@ -131,8 +132,8 @@ export const workflowSlice = createSlice({
       });
 
       return {
-        ...cloneDeep(initialWorkflowState),
-        ...cloneDeep(workflowExtra),
+        ...deepClone(initialWorkflowState),
+        ...deepClone(workflowExtra),
         originalExposedFieldValues,
         mode: state.mode,
       };
@@ -144,7 +145,7 @@ export const workflowSlice = createSlice({
       });
     });
 
-    builder.addCase(nodeEditorReset, () => cloneDeep(initialWorkflowState));
+    builder.addCase(nodeEditorReset, () => deepClone(initialWorkflowState));
 
     builder.addCase(nodesChanged, (state, action) => {
       // Not all changes to nodes should result in the workflow being marked touched

invokeai/frontend/web/src/features/nodes/util/graph/addIPAdapterToLinearGraph.ts

@@ -48,7 +48,7 @@ export const addIPAdapterToLinearGraph = async (
       if (!ipAdapter.model) {
         return;
       }
-      const { id, weight, model, beginStepPct, endStepPct, controlImage } = ipAdapter;
+      const { id, weight, model, clipVisionModel, beginStepPct, endStepPct, controlImage } = ipAdapter;
 
       assert(controlImage, 'IP Adapter image is required');
 
@@ -58,6 +58,7 @@ export const addIPAdapterToLinearGraph = async (
         is_intermediate: true,
         weight: weight,
         ip_adapter_model: model,
+        clip_vision_model: clipVisionModel,
         begin_step_percent: beginStepPct,
         end_step_percent: endStepPct,
         image: {
@@ -83,7 +84,7 @@ export const addIPAdapterToLinearGraph = async (
 };
 
 const buildIPAdapterMetadata = (ipAdapter: IPAdapterConfig): S['IPAdapterMetadataField'] => {
-  const { controlImage, beginStepPct, endStepPct, model, weight } = ipAdapter;
+  const { controlImage, beginStepPct, endStepPct, model, clipVisionModel, weight } = ipAdapter;
 
   assert(model, 'IP Adapter model is required');
 
@@ -99,6 +100,7 @@ const buildIPAdapterMetadata = (ipAdapter: IPAdapterConfig): S['IPAdapterMetadat
 
   return {
     ip_adapter_model: model,
+    clip_vision_model: clipVisionModel,
     weight,
     begin_step_percent: beginStepPct,
     end_step_percent: endStepPct,

invokeai/frontend/web/src/features/nodes/util/node/nodeUpdate.ts

@@ -1,8 +1,9 @@
+import { deepClone } from 'common/util/deepClone';
 import { satisfies } from 'compare-versions';
 import { NodeUpdateError } from 'features/nodes/types/error';
 import type { InvocationNode, InvocationTemplate } from 'features/nodes/types/invocation';
 import { zParsedSemver } from 'features/nodes/types/semver';
-import { cloneDeep, defaultsDeep, keys, pick } from 'lodash-es';
+import { defaultsDeep, keys, pick } from 'lodash-es';
 
 import { buildInvocationNode } from './buildInvocationNode';
 
@@ -50,7 +51,7 @@ export const updateNode = (node: InvocationNode, template: InvocationTemplate):
   // The updateability of a node, via semver comparison, relies on the this kind of recursive merge
   // being valid. We rely on the template's major version to be majorly incremented if this kind of
   // merge would result in an invalid node.
-  const clone = cloneDeep(node);
+  const clone = deepClone(node);
   clone.data.version = template.version;
   defaultsDeep(clone, defaults); // mutates!
 

invokeai/frontend/web/src/features/nodes/util/workflow/buildWorkflow.ts

@@ -1,11 +1,12 @@
 import { logger } from 'app/logging/logger';
+import { deepClone } from 'common/util/deepClone';
 import { parseify } from 'common/util/serialize';
 import type { NodesState, WorkflowsState } from 'features/nodes/store/types';
 import { isInvocationNode, isNotesNode } from 'features/nodes/types/invocation';
 import type { WorkflowV3 } from 'features/nodes/types/workflow';
 import { zWorkflowV3 } from 'features/nodes/types/workflow';
 import i18n from 'i18n';
-import { cloneDeep, pick } from 'lodash-es';
+import { pick } from 'lodash-es';
 import { fromZodError } from 'zod-validation-error';
 
 export type BuildWorkflowArg = {
@@ -30,7 +31,7 @@ const workflowKeys = [
 type BuildWorkflowFunction = (arg: BuildWorkflowArg) => WorkflowV3;
 
 export const buildWorkflowFast: BuildWorkflowFunction = ({ nodes, edges, workflow }: BuildWorkflowArg): WorkflowV3 => {
-  const clonedWorkflow = pick(cloneDeep(workflow), workflowKeys);
+  const clonedWorkflow = pick(deepClone(workflow), workflowKeys);
 
   const newWorkflow: WorkflowV3 = {
     ...clonedWorkflow,
@@ -43,14 +44,14 @@ export const buildWorkflowFast: BuildWorkflowFunction = ({ nodes, edges, workflo
       newWorkflow.nodes.push({
         id: node.id,
         type: node.type,
-        data: cloneDeep(node.data),
+        data: deepClone(node.data),
         position: { ...node.position },
       });
     } else if (isNotesNode(node) && node.type) {
       newWorkflow.nodes.push({
         id: node.id,
         type: node.type,
-        data: cloneDeep(node.data),
+        data: deepClone(node.data),
         position: { ...node.position },
       });
     }

invokeai/frontend/web/src/features/nodes/util/workflow/migrations.ts

@@ -1,4 +1,5 @@
 import { $store } from 'app/store/nanostores/store';
+import { deepClone } from 'common/util/deepClone';
 import { WorkflowMigrationError, WorkflowVersionError } from 'features/nodes/types/error';
 import type { FieldType } from 'features/nodes/types/field';
 import type { InvocationNodeData } from 'features/nodes/types/invocation';
@@ -11,7 +12,7 @@ import { zWorkflowV2 } from 'features/nodes/types/v2/workflow';
 import type { WorkflowV3 } from 'features/nodes/types/workflow';
 import { zWorkflowV3 } from 'features/nodes/types/workflow';
 import { t } from 'i18next';
-import { cloneDeep, forEach } from 'lodash-es';
+import { forEach } from 'lodash-es';
 import { z } from 'zod';
 
 /**
@@ -89,7 +90,7 @@ export const parseAndMigrateWorkflow = (data: unknown): WorkflowV3 => {
     throw new WorkflowVersionError(t('nodes.unableToGetWorkflowVersion'));
   }
 
-  let workflow = cloneDeep(data) as WorkflowV1 | WorkflowV2 | WorkflowV3;
+  let workflow = deepClone(data) as WorkflowV1 | WorkflowV2 | WorkflowV3;
 
   if (workflow.meta.version === '1.0.0') {
     const v1 = zWorkflowV1.parse(workflow);

invokeai/frontend/web/src/features/parameters/store/generationSlice.ts

@@ -280,6 +280,7 @@ const migrateGenerationState = (state: any): GenerationState => {
     // The signature of the model has changed, so we need to reset it
     state._version = 2;
     state.model = null;
+    state.canvasCoherenceMode = initialGenerationState.canvasCoherenceMode;
   }
   return state;
 };

invokeai/frontend/web/src/features/settingsAccordions/components/AdvancedSettingsAccordion/AdvancedSettingsAccordion.tsx

@@ -61,7 +61,7 @@ export const AdvancedSettingsAccordion = memo(() => {
 
   return (
     <StandaloneAccordion label={t('accordions.advanced.title')} badges={badges} isOpen={isOpen} onToggle={onToggle}>
-      <Flex gap={4} alignItems="center" p={4} flexDir="column">
+      <Flex gap={4} alignItems="center" p={4} flexDir="column" data-testid="advanced-settings-accordion">
         <Flex gap={4} w="full">
           <ParamVAEModelSelect />
           <ParamVAEPrecision />

invokeai/frontend/web/src/features/settingsAccordions/components/ControlSettingsAccordion/ControlSettingsAccordion.tsx

@@ -77,7 +77,7 @@ export const ControlSettingsAccordion: React.FC = memo(() => {
 
   return (
     <StandaloneAccordion label={t('accordions.control.title')} badges={badges} isOpen={isOpen} onToggle={onToggle}>
-      <Flex gap={2} p={4} flexDir="column">
+      <Flex gap={2} p={4} flexDir="column" data-testid="control-accordion">
         <ButtonGroup size="sm" w="full" justifyContent="space-between" variant="ghost" isAttached={false}>
           <Button
             tooltip={t('controlnet.addControlNet')}