`torch` wasn't seeing the environment variable. I suspect this is
because it was imported before the variable was set, so was running with
a different environment.
Many `torch` ops are supported on MPS so this wasn't noticed
immediately, but some samplers like k_dpm_2 still use unsupported
operations and need this fallback.
This PR forces the installer to install the official torch-cu117 wheel
from download.torch.org, rather than relying on PyPi.org to return the
correct version. It ought to correct the problems that some people have
experienced with cuda support not being installed.
1. The convert module was converting ckpt models into
StableDiffusionGeneratorPipeline objects for use in-memory, but then
when saved to disk created files that could not be merged with
StableDiffusionPipeline models. I have added a flag that selects which
pipeline class to return, so that both in-memory and disk conversions
work properly.
2. This PR also fixes an issue with `invoke.sh` not using the correct
path for the textual inversion and merge scripts.
3. Quench nags during the merge process about the safety checker being
turned off.
`torch` wasn't seeing the environment variable. I suspect this is because it was imported before the variable was set, so was running with a different environment.
Many `torch` ops are supported on MPS so this wasn't noticed immediately, but some samplers like k_dpm_2 still use unsupported operations and need this fallback.
* remove non maintained Dockerfile
* adapt Docker related files to latest changes
- also build the frontend when building the image
- skip user response if INVOKE_MODEL_RECONFIGURE is set
- split INVOKE_MODEL_RECONFIGURE to support more than one argument
* rename `docker-build` dir to `docker`
* update build-container.yml
- rename image to invokeai
- add cpu flavor
- add metadata to build summary
- enable caching
- remove build-cloud-img.yml
* fix yarn cache path, link copyjob
Crashes would occur in the invokeai-configure script if no HF token
was found in cache and the user declines to provide one when prompted.
The reason appears to be that on Linux systems getpass_asterisk()
raises an EOFError when no input is provided
On windows10, getpass_asterisk() does not raise the EOFError, but
returns an empty string instead. This patch detects this and raises
the exception so that the control logic is preserved.
if reinstalling over an existing installation where the .venv was
created with symlinks to system python instead of copies of the python
executable, the installer would raise a `SameFileError`, because it
would attempt to copy Python over itself. This fixes the issue.
Copying the executable is still preferred for new environments, because
this guarantees the stable Python version.
- fixes bug in finding the source of the configs dir;
- updates the docs for manual install to clarify the preference to
keeping the `.venv` inside the runtime dir, and the caveat/extra steps
required if done otherwise
