* introduce new abstraction layer for GPU devices
* add unit test for device abstraction
* fix ruff
* convert TorchDeviceSelect into a stateless class
* move logic to select context-specific execution device into context API
* add mock hardware environments to pytest
* remove dangling mocker fixture
* fix unit test for running on non-CUDA systems
* remove unimplemented get_execution_device() call
* remove autocast precision
* Multiple changes:
1. Remove TorchDeviceSelect.get_execution_device(), as well as calls to
context.models.get_execution_device().
2. Rename TorchDeviceSelect to TorchDevice
3. Added back the legacy public API defined in `invocation_api`, including
choose_precision().
4. Added a config file migration script to accommodate removal of precision=autocast.
* add deprecation warnings to choose_torch_device() and choose_precision()
* fix test crash
* remove app_config argument from choose_torch_device() and choose_torch_dtype()
---------
Co-authored-by: Lincoln Stein <lstein@gmail.com>
`LatentsField` objects have an optional `seed` field. This should only be populated when the latents are noise, generated from a seed.
`DenoiseLatentsInvocation` needs a seed value for scheduler initialization. It's used in a few places, and there is some logic for determining the seed to use with a series of fallbacks:
- Use the seed from the noise (a `LatentsField` object)
- Use the seed from the latents (a `LatentsField` object - normally it won't have a seed)
- Use `0` as a final fallback
In `DenoisLatentsInvocation`, we set the seed in the `LatentsOutput`, even though the output latents are not noise.
This is normally fine, but when we use refiner, we re-use the those same latents for the refiner denoise. This causes that characteristic same-seed-fried look on the refiner pass.
Simple fix - do not set the field in the output latents.
The previous algorithm errored if the image wasn't divisible by the tile size. I've reimplemented it from scratch to mitigate this issue.
The new algorithm is simpler. We create a pool of tiles, then use them to create an image composed completely of tiles. If there is any awkwardly sized space on the edge of the image, the tiles are cropped to fit.
Finally, paste the original image over the tile image.
I've added a jupyter notebook to do a smoke test of infilling methods, and 10 test images.
The other infill algorithms can be easily tested with the notebook on the same images, though I didn't set that up yet.
Tested and confirmed this gives results just as good as the earlier infill, though of course they aren't the same due to the change in the algorithm.
We have had a few bugs with v4 related to file encodings, especially on Windows.
Windows uses its own character encodings instead of `utf-8`, often `cp1252`. Some characters cannot be decoded using `utf-8`, causing `UnicodeDecodeError`.
There are a couple places where this can cause problems:
- In the installer bootstrap, we install or upgrade `pip` and decode the result, using `subprocess`.
The input to this includes the user's home dir. In #6105, the user had one of the problematic characters in their username. `subprocess` attempts and fails to decode the username, which crashes the installer.
To fix this, we need to use `locale.getpreferredencoding()` when executing the command.
- Similarly, in the model install service and config class, we attempt to load a yaml config file. If a problematic character is in the path to the file (which often includes the user's home dir), we can get the same error.
One example is #6129 in which the models.yaml migration fails.
To fix this, we need to open the file with `locale.getpreferredencoding()`.
Compare the installed paths to determine if the model is already installed. Fixes an issue where installed models showed up as uninstalled or vice-versa. Related to relative vs absolute path handling.
Renaming the model file to the model name introduces unnecessary contraints on model names.
For example, a model name can technically be any length, but a model _filename_ cannot be too long.
There are also constraints on valid characters for filenames which shouldn't be applied to model record names.
I believe the old behaviour is a holdover from the old system.
Setting to 'auto' works only for InvokeAI config and auto detects the SD model but will override if user explicitly sets it. If auto used with checkpoint models, we raise an error. Checkpoints will always need to set to non-auto.
Previously, exceptions raised as custom nodes are initialized were fatal errors, causing the app to exit.
With this change, any error on import is caught and the error message printed. App continues to start up without the node.
For example, a custom node that isn't updated for v4.0.0 may raise an error on import if it is attempting to import things that no longer exist.
Prefer an early return/continue to reduce the indentation of the processor loop. Easier to read.
There are other ways to improve its structure but at first glance, they seem to involve changing the logic in scarier ways.
This must not have been tested after the processors were unified. Needed to shift the logic around so the resume event is handled correctly. Clear and easy fix.
* pass model config to _load_model
* make conversion work again
* do not write diffusers to disk when convert_cache set to 0
* adding same model to cache twice is a no-op, not an assertion error
* fix issues identified by psychedelicious during pr review
* following conversion, avoid redundant read of cached submodels
* fix error introduced while merging
---------
Co-authored-by: Lincoln Stein <lstein@gmail.com>
We switched all model paths to be absolute in #5900. In hindsight, this is a mistake, because it makes the `models_dir` non-portable.
This change reverts to the previous model pathing:
- Invoke-managed models (in the `models_dir`) are stored with relative paths
- Non-invoke-managed models (outside the `models_dir`, i.e. in-place installed models) still have absolute paths.
## Why absolute paths make things non-portable
Let's say my `models_dir` is `/media/rhino/invokeai/models/`. In the DB, all model paths will be absolute children of this path, like this:
- `/media/rhino/invokeai/models/sd-1/main/model1.ckpt`
I want to change my `models_dir` to `/home/bat/invokeai/models/`. I update my `invokeai.yaml` file and physically move the files to that directory.
On startup, the app checks for missing models. Because all of my model paths were absolute, they now point to a nonexistent path. All models are broken.
There are a couple options to recover from this situation, neither of which are reasonable:
1. The user must manually update every model's path. Unacceptable UX.
2. On startup, we check for missing models. For each missing model, we compare its path with the last-known models dir. If there is a match, we replace that portion of the path with the new models dir. Then we re-check to see if the path exists. If it does, we update the models DB entry. Brittle and requires a new DB entry for last-known models dir.
It's better to use relative paths for Invoke-managed models.
These two changes are interrelated.
## Autoimport
The autoimport feature can be easily replicated using the scan folder tab in the model manager. Removing the implicit autoimport reduces surface area and unifies all model installation into the UI.
This functionality is removed, and the `autoimport_dir` config setting is removed.
## Startup model dir scanning
We scanned the invoke-managed models dir on startup and took certain actions:
- Register orphaned model files
- Remove model records from the db when the model path doesn't exist
### Orphaned model files
We should never have orphaned model files during normal use - we manage the models directory, and we only delete files when the user requests it.
During testing or development, when a fresh DB or memory DB is used, we could end up with orphaned models that should be registered.
Instead of always scanning for orphaned models and registering them, we now only do the scan if the new `scan_models_on_startup` config flag is set.
The description for this setting indicates it is intended for use for testing only.
### Remove records for missing model files
This functionality could unexpectedly wipe models from the db.
For example, if your models dir was on external media, and that media was inaccessible during startup, the scan would see all your models as missing and delete them from the db.
The "proactive" scan is removed. Instead, we will scan for missing models and log a warning if we find a model whose path doesn't exist. No possibility for data loss.
I had added this because I mistakenly believed the HF token was required to download HF models.
Turns out this is not the case, and the vast majority of HF models do not need the API token to download.
Previously we only handled expected error types. If a different error was raised, the install job would end up in an unexpected state where it has failed and isn't doing anything, but its status is still running.
This indirectly prevents the installer threads from exiting - they are waiting for all jobs to be completed, including the failed-but-still-running job.
We need to handle any error here to prevent this.
- Enriched dependencies to not just be a string - allows reuse of a dependency as a starter model _and_ dependency of another model. For example, all the SDXL models have the fp16 VAE as a dependency, but you can also download it on its own.
- Looked at popular models on the major model sites to select the list. No SD2 models. All hosted on HF.
* Fix minor bugs involving model manager handling of model paths
- Leave models found in the `autoimport` directory there. Do not move them
into the `models` hierarchy.
- If model name, type or base is updated and model is in the `models` directory,
update its path as appropriate.
- On startup during model scanning, if a model's path is a symbolic link, then resolve
to an absolute path before deciding it is a new model that must be hashed and
registered. (This prevents needless hashing at startup time).
* fix issue with dropped suffix
---------
Co-authored-by: Lincoln Stein <lstein@gmail.com>
Add class `DefaultInvokeAIAppConfig`, which inherits from `InvokeAIAppConfig`. When instantiated, this class does not parse environment variables, so it outputs a "clean" default config. That's the only difference.
Then, we can use this new class in the 3 places:
- When creating the example config file (no env vars should be here)
- When migrating a v3 config (we want to instantiate the migrated config without env vars, so that when we write it out, they are not written to disk)
- When creating a fresh config file (i.e. on first run with an uninitialized root or new config file path - no env vars here!)
For SSDs, `blake3` is about 10x faster than `blake3_single` - 3 files/second vs 30 files/second.
For spinning HDDs, `blake3` is about 100x slower than `blake3_single` - 300 seconds/file vs 3 seconds/file.
For external drives, `blake3` is always worse, but the difference is highly variable. For external spinning drives, it's probably way worse than internal.
The least offensive algorithm is `blake3_single`, and it's still _much_ faster than any other algorithm.
Some processors, like Canny, didn't use `detect_resolution`. The resultant control images were then resized by the processors from 512x512 to the desired dimensions. The result is that the control images are the right size, but very low quality.
Using detect_resolution fixes this.
This allows users to create simple "profiles" via separate `invokeai.yaml` files.
- Remove `InvokeAIAppConfig.set_root()`, it's extraneous
- Remove `InvokeAIAppConfig.merge_from_file()`, it's extraneous
- Add `--config` to the app arg parser, add `InvokeAIAppConfig._config_file`, and consume in the config singleton getter
- `InvokeAIAppConfig.init_file_path` -> `InvokeAIAppConfig.config_file_path`
The models from INITIAL_MODELS.yaml have been recreated as a structured python object. This data is served on a new route. The model sources are compared against currently-installed models to determine if they are already installed or not.
This flag acts as a proxy for the `get_config()` function to determine if the full application is running.
If it was, the config will set the root, do HF login, etc.
If not (e.g. it's called by an external script), all that stuff will be skipped.
HF login, legacy yaml confs, and default init file are all handled during app setup.
All directories are created as they are needed by the app.
No need to check for a valid root dir - we will make it if it doesn't exist.
This provides a simple way to provide a HF token. If HF reports no valid token, one is prompted for until a valid token is provided, or the user presses Ctrl + C to cancel.
Use the util function to calculate ram cache size on startup. This way, the `ram` setting will always be optimized for a system, even if they add or remove RAM. In other words, the default value is now dynamic.
When running the configurator, the `legacy_models_conf_path` was stripped when saving the config file. Then the migration logic didn't fire correctly, and the custom models.yaml paths weren't migrated into the db.
- Rework the logic to migrate this path by adding it to the config object as a normal field that is not excluded from serialization.
- Rearrange the models.yaml migration logic to remove the legacy path after migrating, then write the config file. This way, the legacy path doesn't stick around.
- Move the schema version into the config object.
- Back up the config file before attempting migration.
- Add tests to cover this edge case
Hold onto `conf_path` temporarily while migrating `invokeai.yaml` so that it gets migrated correctly as the model installer starts up. Stashed as `legacy_models_yaml_path` in the config, excluded from serialization.
We have two problems with how argparse is being utilized:
- We parse CLI args as the `api_app.py` file is read. This causes a problem pytest, which has an incompatible set of CLI args. Some tests import the FastAPI app, which triggers the config to parse CLI args, which receives the pytest args and fails.
- We've repeatedly had problems when something that uses the config is imported before the CLI args are parsed. When this happens, the root dir may not be set correctly, so we attempt to operate on incorrect paths.
To resolve these issues, we need to lift CLI arg parsing outside of the application code, but still let the application access the CLI args. We can create a external app entrypoint to do this.
- `InvokeAIArgs` is a simple helper class that parses CLI args and stores the result.
- `run_app()` is the new entrypoint. It first parses CLI args, then runs `invoke_api` to start the app.
The `invokeai-web` project script and `invokeai-web.py` dev script now call `run_app()` instead of `invoke_api()`.
The first time `get_config()` is called to get the singleton config object, it retrieves the args from `InvokeAIArgs`, sets the root dir if provided, then merges settings in from `invokeai.yaml`.
CLI arg parsing is now safely insulated from application code, but still accessible. And we don't need to worry about import order having an impact on anything, because by the time the app is running, we have already parsed CLI args. Whew!
This fixes an issue with `test_images.py`, which tests the bulk images routers and imports the whole FastAPI app. This triggers the config logic which fails on the test runner, because it has no `invokeai.yaml`.
Also probably just good for graceful fallback.
- `write_file` requires an destination file path
- `read_config` -> `merge_from_file`, if no path is provided, reads from `self.init_file_path`
- update app, tests to use new methods
- fix configurator, was overwriting config file data unexpectedly
Tweak the name of it so that incoming configs with the old default value of 6 have the setting stripped out. The result is all configs will now have the new, much better default value of 1.
Having this all in the `get_config` function makes testing hard. Move these two functions to their own methods, and call them on app startup explicitly.
- Remove OmegaConf. It functioned as an intermediary data format, between YAML/argparse and pydantic. It's not necessary - we can parse YAML or CLI args directly with pydantic.
- Remove dynamic CLI args. Only `root` is explicitly supported. This greatly simplifies config handling. Configuration is done by editing the YAML file. Frequently-used args can be added if there is a demand.
- A separate arg parser is created to handle the slimmed-down CLI args. It's run immediately in the `invokeai-web` script to handle `--version` and `--help`. It is also used inside the singleton config getter (see below).
- Remove categories from the config. Our settings model is mostly flat. Handling categories adds complexity for both us and users - we have to handle transforming a flat config to categorized config (and vice-versa), while users have to be careful with indentation in their YAML file.
- Add a `meta` key to the config file. Currently, this holds the config schema version only. It is not a part of the config object itself.
- Remove legacy settings that are no longer referenced, or were effectively no-op settings when referenced in code.
- Implement simple migration logic to for v3 configs. If migration is successful, the v3 config file is backed up to `invokeai.yaml.bak` and the new config written to `invokeai.yaml`.
- Previously, the singleton config was accessed by calling `InvokeAIAppConfig.get_config()`. This returned an instance of `InvokeAIAppConfig`, which _also_ has the `get_config` function. This created to a confusing situation where you weren't sure if you needed to call `get_config` or just use the config object. This method is replaced by a standalone `get_config` function which returns a singleton config object.
- Wrap CLI arg parsing (for `root`) and loading/migrating `invokeai.yaml` into the new `get_config()` function.
- Move `generate_config_docstrings` into standalone utility function.
- Make `root` a private attr (`_root`). This reduces the temptation to directly modify and or use this sensitive field and ensures it is neither serialized nor read from input data. Use `root_path` to access the resolved root path, or `set_root` to set the root to something.
- No longer install core conversion models. Use the HuggingFace cache to load
them if and when needed.
- Call directly into the diffusers library to perform conversions with only shallow
wrappers around them to massage arguments, etc.
- At root configuration time, do not create all the possible model subdirectories,
but let them be created and populated at model install time.
- Remove checks for missing core conversion files, since they are no
longer installed.
In the client, a controlnet or t2i adapter has two images:
- The source control image: the image the user selected (required)
- The processed control image: the user's image after we've processed it (optional)
The processed image is optional because a user may provide a pre-processed image.
We only actually use one of these images when building the graph, and until this change, we only stored one of the in image metadata. This created a situation where only a processed image was stored in metadata - say, a canny edge map - and the user-selected image wasn't provided.
By adding the processed image to metadata, we can recall both the control image and optional processed image.
This commit is followed by a UI-facing changes to support the change.
BLAKE3 has poor performance on spinning disks when parallelized. See https://github.com/BLAKE3-team/BLAKE3/issues/31
- Replace `skip_model_hash` setting with `hashing_algorithm`. Any algorithm we support is accepted.
- Add `random` algorithm: hashes a UUID with BLAKE3 to create a random "hash". Equivalent to the previous skip functionality.
- Add `blake3_single` algorithm: hashes on a single thread using BLAKE3, fixes the aforementioned performance issue
- Update model probe to accept the algorithm to hash with as an optional arg, defaulting to `blake3`
- Update all calls of the probe to use the app's configured hashing algorithm
- Update an external script that probes models
- Update tests
- Move ModelHash into its own module to avoid circuclar import issues
We were stripping the file extension from file models when moving them in `_sync_model_path`. For example, `some_model.safetensors` would be moved to `some_model`, which of course breaks things.
Instead of using the model's name as the new path, use the model's path's last segment. This is the same behaviour for directories, but for files, it retains the file extension.
- If the metadata yaml has an invalid version, exist the app. If we don't, the app will crawl the models dir and add models to the db without having first parsed `models.yaml`. This should not happen often, as the vast majority of users are on v3.0.0 models.yaml files.
- Fix off-by-one error with models count (need to pop the `__metadata__` stanza
- After a successful migration, rename `models.yaml` to `models.yaml.bak` to prevent the migration logic from re-running on subsequent app startups.
The old logic to check if a model needed to be moved relied on the model path being a relative path. Paths are now absolute, causing this check to fail. We then assumed the paths were different and moved the model from its current location to, well, its current location.
Use more resilient method to check if a model should be moved.
mkdocs can autogenerate python class docs from its docstrings. Our config is a pydantic model.
It's tedious and error-prone to duplicate docstrings from the pydantic field descriptions to the class docstrings.
- Add helper function to generate a mkdocs-compatible docstring from the InvokeAIAppConfig class fields
Recently the schema for models was changed to a generic `ModelField`, and the UI was unable to derive the type of those fields. This didn't affect functionality, but it did break the styling of handles.
Add `ui_type` to the affected fields and update the UI to use the correct capitalizations.
A list of regex and token pairs is accepted. As a file is downloaded by the model installer, the URL is tested against the provided regex/token pairs. The token for the first matching regex is used during download, added as a bearer token.
When we change a model image, its URL remains the same. The browser will aggressively cache the image. The easiest way to fix this is to append a random query parameter to the URL whenever we build a model config in the API.
- All models are identified by a key and optionally a submodel type via new model `ModelField`. Previously, a few model types had their own class, but not all of them. This inconsistency just added complexity without any benefit.
- Update all invocation to use the new format.
- In the node API, models are loaded by key or an instance of `ModelField` as a convenience.
- Add an enriched model schema for metadata. It includes key, hash, name, base and type.
In order for delete by match to work, we need the whole invocation output to be stringified.
For some reason, the serialization of the output was set to only include the `type` field. It should instead include the whole output.
I don't understand how this ever worked unless pydantic had different serialization behaviour in v1 (though it appears to have been the same).
Closes#5805
Rename MM routes to be consistent:
- "import" -> "install"
- "model_record" -> "model"
Comment several unused routes while I work (may end up removing them?):
- list model summary (we use the search route instead)
- add model record
- convert model
- merge models
- Metadata is merged with the config. We can simplify the MM substantially and remove the handling for metadata.
- Per discussion, we don't have an ETA for frontend implementation of tags, and with the realization that the tags from CivitAI are largely useless, there's no reason to keep tags in the MM right now. When we are ready to implement tags on the frontend, we can refer back to the implementation here and use it if it supports the design.
- Fix all tests.
* UI in MM to create trigger phrases
* add scheduler and vaePrecision to config
* UI for configuring default settings for models'
* hook MM default model settings up to API
* add button to set default settings in parameters
* pull out trigger phrases
* back-end for default settings
* lint
* remove log;
gi
* ruff
* ruff format
---------
Co-authored-by: Mary Hipp <maryhipp@Marys-MacBook-Air.local>
- When installing, model keys are now calculated from the model contents.
- .safetensors, .ckpt and other single file models are hashed with sha1
- The contents of diffusers directories are hashed using imohash (faster)
fixup yaml->sql db migration script to assign deterministic key
- this commit also detects and assigns the correct image encoder for
ip adapter models.
Gets the first model that matches the given name, base and type. Raises 404 if there isn't one.
This will be used for backwards compatibility with old metadata.
This was done in the frontend before but it's something the backend should handle.
The logic compares the found model paths to the path and source of all installed models. It excludes core models.
- Support extended HF repoid syntax in TUI. This allows
installation of subfolders and safetensors files, as in
`XpucT/Deliberate::Deliberate_v5.safetensors`
- Add `error` and `error_traceback` properties to the install
job objects.
- Rename the `heuristic_import` route to `heuristic_install`.
- Fix the example `config` input in the `heuristic_install` route.
Double underscores are used in the app but it doesn't actually do or convey anything that single underscores don't already do. Considered unpythonic except for actual dunder/magic methods.
Consolidate graph processing logic into session processor.
With graphs as the unit of work, and the session queue distributing graphs, we no longer need the invocation queue or processor.
Instead, the session processor dequeues the next session and processes it in a simple loop, greatly simplifying the app.
- Remove `graph_execution_manager` service.
- Remove `queue` (invocation queue) service.
- Remove `processor` (invocation processor) service.
- Remove queue-related logic from `Invoker`. It now only starts and stops the services, providing them with access to other services.
- Remove unused `invocation_retrieval_error` and `session_retrieval_error` events, these are no longer needed.
- Clean up stats service now that it is less coupled to the rest of the app.
- Refactor cancellation logic - cancellations now originate from session queue (i.e. HTTP cancel endpoint) and are emitted as events. Processor gets the events and sets the canceled event. Access to this event is provided to the invocation context for e.g. the step callback.
- Remove `sessions` router; it provided access to `graph_executions` but that no longer exists.
`GraphInvocation` is a node that can contain a whole graph. It is removed for a number of reasons:
1. This feature was unused (the UI doesn't support it) and there is no plan for it to be used.
The use-case it served is known in other node execution engines as "node groups" or "blocks" - a self-contained group of nodes, which has group inputs and outputs. This is a planned feature that will be handled client-side.
2. It adds substantial complexity to the graph processing logic. It's probably not enough to have a measurable performance impact but it does make it harder to work in the graph logic.
3. It allows for graphs to be recursive, and the improved invocations union handling does not play well with it. Actually, it works fine within `graph.py` but not in the tests for some reason. I do not understand why. There's probably a workaround, but I took this as encouragement to remove `GraphInvocation` from the app since we don't use it.
The change to `Graph.nodes` and `GraphExecutionState.results` validation requires some fanagling to get the OpenAPI schema generation to work. See new comments for a details.
We use pydantic to validate a union of valid invocations when instantiating a graph.
Previously, we constructed the union while creating the `Graph` class. This introduces a dependency on the order of imports.
For example, consider a setup where we have 3 invocations in the app:
- Python executes the module where `FirstInvocation` is defined, registering `FirstInvocation`.
- Python executes the module where `SecondInvocation` is defined, registering `SecondInvocation`.
- Python executes the module where `Graph` is defined. A union of invocations is created and used to define the `Graph.nodes` field. The union contains `FirstInvocation` and `SecondInvocation`.
- Python executes the module where `ThirdInvocation` is defined, registering `ThirdInvocation`.
- A graph is created that includes `ThirdInvocation`. Pydantic validates the graph using the union, which does not know about `ThirdInvocation`, raising a `ValidationError` about an unknown invocation type.
This scenario has been particularly problematic in tests, where we may create invocations dynamically. The test files have to be structured in such a way that the imports happen in the right order. It's a major pain.
This PR refactors the validation of graph nodes to resolve this issue:
- `BaseInvocation` gets a new method `get_typeadapter`. This builds a pydantic `TypeAdapter` for the union of all registered invocations, caching it after the first call.
- `Graph.nodes`'s type is widened to `dict[str, BaseInvocation]`. This actually is a nice bonus, because we get better type hints whenever we reference `some_graph.nodes`.
- A "plain" field validator takes over the validation logic for `Graph.nodes`. "Plain" validators totally override pydantic's own validation logic. The validator grabs the `TypeAdapter` from `BaseInvocation`, then validates each node with it. The validation is identical to the previous implementation - we get the same errors.
`BaseInvocationOutput` gets the same treatment.
- Replace AnyModelLoader with ModelLoaderRegistry
- Fix type check errors in multiple files
- Remove apparently unneeded `get_model_config_enum()` method from model manager
- Remove last vestiges of old model manager
- Updated tests and documentation
resolve conflict with seamless.py
- Rename old "model_management" directory to "model_management_OLD" in order to catch
dangling references to original model manager.
- Caught and fixed most dangling references (still checking)
- Rename lora, textual_inversion and model_patcher modules
- Introduce a RawModel base class to simplfy the Union returned by the
model loaders.
- Tidy up the model manager 2-related tests. Add useful fixtures, and
a finalizer to the queue and installer fixtures that will stop the
services and release threads.
- ModelMetadataStoreService is now injected into ModelRecordStoreService
(these two services are really joined at the hip, and should someday be merged)
- ModelRecordStoreService is now injected into ModelManagerService
- Reduced timeout value for the various installer and download wait*() methods
- Introduced a Mock modelmanager for testing
- Removed bare print() statement with _logger in the install helper backend.
- Removed unused code from model loader init file
- Made `locker` a private variable in the `LoadedModel` object.
- Fixed up model merge frontend (will be deprecated anyway!)
- Replace legacy model manager service with the v2 manager.
- Update invocations to use new load interface.
- Fixed many but not all type checking errors in the invocations. Most
were unrelated to model manager
- Updated routes. All the new routes live under the route tag
`model_manager_v2`. To avoid confusion with the old routes,
they have the URL prefix `/api/v2/models`. The old routes
have been de-registered.
- Added a pytest for the loader.
- Updated documentation in contributing/MODEL_MANAGER.md
- Implement new model loader and modify invocations and embeddings
- Finish implementation loaders for all models currently supported by
InvokeAI.
- Move lora, textual_inversion, and model patching support into
backend/embeddings.
- Restore support for model cache statistics collection (a little ugly,
needs work).
- Fixed up invocations that load and patch models.
- Move seamless and silencewarnings utils into better location
- Cache stat collection enabled.
- Implemented ONNX loading.
- Add ability to specify the repo version variant in installer CLI.
- If caller asks for a repo version that doesn't exist, will fall back
to empty version rather than raising an error.
We have two different classes named `ModelInfo` which might need to be used by API consumers. We need to export both but have to deal with this naming collision.
The `ModelInfo` I've renamed here is the one that is returned when a model is loaded. It's the object least likely to be used by API consumers.
Replace `delete_on_startup: bool` & associated logic with `ephemeral: bool` and `TemporaryDirectory`.
The temp dir is created inside of `output_dir`. For example, if `output_dir` is `invokeai/outputs/tensors/`, then the temp dir might be `invokeai/outputs/tensors/tmpvj35ht7b/`.
The temp dir is cleaned up when the service is stopped, or when it is GC'd if not properly stopped.
In the event of a catastrophic crash where the temp files are not cleaned up, the user can delete the tempdir themselves.
This situation may not occur in normal use, but if you kill the process, python cannot clean up the temp dir itself. This includes running the app in a debugger and killing the debugger process - something I do relatively often.
Tests updated.
- The default is to not delete on startup - feels safer.
- The two services using this class _do_ delete on startup.
- The class has "ephemeral" removed from its name.
- Tests & app updated for this change.
`_delete_all` logged how many items it deleted, and had to be called _after_ service start bc it needed access to logger.
Move the logger call to the startup method and return the the deleted stats from `_delete_all`. This lets `_delete_all` be called at any time.
Turns out they are just different enough in purpose that the implementations would be rather unintuitive. I've made a separate ObjectSerializer service to handle tensors and conditioning.
Refined the class a bit too.
Turns out `ItemStorageABC` was almost identical to `PickleStorageBase`. Instead of maintaining separate classes, we can use `ItemStorageABC` for both.
There's only one change needed - the `ItemStorageABC.set` method must return the newly stored item's ID. This allows us to let the service handle the responsibility of naming the item, but still create the requisite output objects during node execution.
The naming implementation is improved here. It extracts the name of the generic and appends a UUID to that string when saving items.
- New generic class `PickleStorageBase`, implements the same API as `LatentsStorageBase`, use for storing non-serializable data via pickling
- Implementation `PickleStorageTorch` uses `torch.save` and `torch.load`, same as `LatentsStorageDisk`
- Add `tensors: PickleStorageBase[torch.Tensor]` to `InvocationServices`
- Add `conditioning: PickleStorageBase[ConditioningFieldData]` to `InvocationServices`
- Remove `latents` service and all `LatentsStorage` classes
- Update `InvocationContext` and all usage of old `latents` service to use the new services/context wrapper methods
This class works the same way as `WithMetadata` - it simply adds a `board` field to the node. The context wrapper function is able to pull the board id from this. This allows image-outputting nodes to get a board field "for free", and have their outputs automatically saved to it.
This is a breaking change for node authors who may have a field called `board`, because it makes `board` a reserved field name. I'll look into how to avoid this - maybe by naming this invoke-managed field `_board` to avoid collisions?
Supporting changes:
- `WithBoard` is added to all image-outputting nodes, giving them the ability to save to board.
- Unused, duplicate `WithMetadata` and `WithWorkflow` classes are deleted from `baseinvocation.py`. The "real" versions are in `fields.py`.
- Remove `LinearUIOutputInvocation`. Now that all nodes that output images also have a `board` field by default, this node is no longer necessary. See comment here for context: https://github.com/invoke-ai/InvokeAI/pull/5491#discussion_r1480760629
- Without `LinearUIOutputInvocation`, the `ImagesInferface.update` method is no longer needed, and removed.
Note: This commit does not bump all node versions. I will ensure that is done correctly before merging the PR of which this commit is a part.
Note: A followup commit will implement the frontend changes to support this change.
- The config is already cached by the config class's `get_config()` method.
- The config mutates itself in its `root_path` property getter. Freezing the class makes any attempt to grab a path from the config error. Unfortunately this means we cannot easily freeze the class without fiddling with the inner workings of `InvokeAIAppConfig`, which is outside the scope here.
