* new OffloadingDevice loads one model at a time, on demand
* fixup! new OffloadingDevice loads one model at a time, on demand
* fix(prompt_to_embeddings): call the text encoder directly instead of its forward method
allowing any associated hooks to run with it.
* more attempts to get things on the right device from the offloader
* more attempts to get things on the right device from the offloader
* make offloading methods an explicit part of the pipeline interface
* inlining some calls where device is only used once
* ensure model group is ready after pipeline.to is called
* fixup! Strategize slicing based on free [V]RAM (#2572)
* doc(offloading): docstrings for offloading.ModelGroup
* doc(offloading): docstrings for offloading-related pipeline methods
* refactor(offloading): s/SimpleModelGroup/FullyLoadedModelGroup
* refactor(offloading): s/HotSeatModelGroup/LazilyLoadedModelGroup
to frame it is the same terms as "FullyLoadedModelGroup"
---------
Co-authored-by: Damian Stewart <null@damianstewart.com>
Tensors with diffusers no longer have to be multiples of 8. This broke Perlin noise generation. We now generate noise for the next largest multiple of 8 and return a cropped result. Fixes#2674.
`generator` now asks `InvokeAIDiffuserComponent` to do postprocessing work on latents after every step. Thresholding - now implemented as replacing latents outside of the threshold with random noise - is called at this point. This postprocessing step is also where we can hook up symmetry and other image latent manipulations in the future.
Note: code at this layer doesn't need to worry about MPS as relevant torch functions are wrapped and made MPS-safe by `generator.py`.
1. Now works with sites that produce lots of redirects, such as CIVITAI
2. Derive name of destination model file from HTTP Content-Disposition header,
if present.
3. Swap \\ for / in file paths provided by users, to hopefully fix issues with
Windows.
- The invokeai-ti and invokeai-merge scripts will crash if there is not enough space
in the console to fit the user interface (even after responsive formatting).
- This PR intercepts the errors and prints a useful error message advising user to
make window larger.
- fix unused variables and f-strings found by pyflakes
- use global_converted_ckpts_dir() to find location of diffusers
- fixed bug in model_manager that was causing the description of converted
models to read "Optimized version of {model_name}'
Strategize slicing based on free [V]RAM when not using xformers. Free [V]RAM is evaluated at every generation. When there's enough memory, the entire generation occurs without slicing. If there is not enough free memory, we use diffusers' sliced attention.
This commit cleans up the code that did bulk imports of legacy model
files. The code has been refactored, and the user is now offered the
option of importing all the model files found in the directory, or
selecting which ones to import.
- This PR introduces a CLI prompt for the proper configuration file to
use when converting a ckpt file, in order to support both inpainting
and v2 models files.
- When user tries to directly !import a v2 model, it prints out a proper
warning that v2 ckpts are not directly supported.
rather than bypassing any path with diffusers in it, im specifically bypassing model.safetensors and diffusion_pytorch_model.safetensors both of which should be diffusers files in most cases.
- If CLI asked to convert the currently loaded model, the model would crash
on the first rendering. CLI will now refuse to convert a model loaded
in memory (probably a good idea in any case).
- CLI will offer the `v1-inpainting-inference.yaml` as the configuration
file when importing an inpainting a .ckpt or .safetensors file that
has "inpainting" in the name. Otherwise it offers `v1-inference.yaml`
as the default.
- The following were supposed to be equivalent, but the latter crashes:
```
invoke> banana sushi
invoke> --prompt="banana sushi"
```
This PR fixes the problem.
- Fixes#2548
Previously conversions of .ckpt and .safetensors files to diffusers
models were failing with channel mismatch errors. This is corrected
with this PR.
- The model_manager convert_and_import() method now accepts the path
to the checkpoint file's configuration file, using the parameter
`original_config_file`. For inpainting files this should be set to
the full path to `v1-inpainting-inference.yaml`.
- If no configuration file is provided in the call, then the presence
of an inpainting file will be inferred at the
`ldm.ckpt_to_diffuser.convert_ckpt_to_diffUser()` level by looking
for the string "inpaint" in the path. AUTO1111 does something
similar to this, but it is brittle and not recommended.
- This PR also changes the model manager model_names() method to return
the model names in case folded sort order.
- Issue is that if insufficient diffusers models are defined in
models.yaml the frontend would ungraciously crash.
- Now it emits appropriate error messages telling user what the problem
is.
