#Is there an existing command line option to run the same prompt across all models registered?

4 messages · Page 1 of 1 (latest)

opal crest
#

I have been reviewing the documentation for the commannd line interface (https://invoke-ai.github.io/InvokeAI/features/CLI/#the-invokeai-command-line-interface) and I do not think this option exists. I can write my own script, but I was curious if anyone has already done this first.

Sometimes I want to see how all of my models perform on a new type of prompt.

At a minimum, I would like to be able to configure a batch run where I define a prompt and settings which are run against all the models currently added/registered with invokeai. Maybe 4 iterations per model (configurable).

Ideally, I could "save" a prompt configuration from the UI and then tell the UI to run that saved prompt/settings against all the models and watch it in the UI.

Is there anything like this already available?

undone parrot
#

Did you go through with your script for this? It sounds like it'd be a real boon for the user community (if not bespoke to your particular setup)

chilly kettle
#

i have something like this in a workflow, i test my prompt against ~5 models

chilly kettle
#

i will share it after some cleanup.