InvokeAI
InvokeAI copied to clipboard
[bug]: Emergency model reconfigure shouldn't kick in after an unrelated exception
Is there an existing issue for this?
- [X] I have searched the existing issues
OS
Windows
GPU
cuda
VRAM
4GB
What happened?
Found in one of the Discord threads. The user experienced a crash due to CUDA OOM (low-VRAM GPU). But because the crash happened on model load, the program kicks them into the emergency-model-reconfigure flow, even though it's an unrelated issue.

Screenshots
No response
Additional context
No response
Contact Details
No response
I agree. This has already been fixed in the main branch and will appear in 2.2.5
@lstein thank you! just for my understanding - is this the fix, or something else?
ah, no, i think it's this PR https://github.com/invoke-ai/InvokeAI/pull/2066/. Great!