Next: Allow TIs to be either a key or a name in the prompt during our transition to using keys

[brandonrising]

What type of PR is this? (check all applicable)

• Refactor
• Feature
• Bug Fix
• Optimization
• Documentation Update
• Community Node Submission

Have you discussed this change with the InvokeAI team?

• Yes
• No, because:

Have you updated all relevant documentation?

• Yes
• No

Description

While we decide the best avenue for passing TI/Embeddings to the backend, this PRs allows the TI trigger word to be either the key of the TI model or the name of it.

Related Tickets & Documents

• Related Issue 5804

[maryhipp]

This worked well for me locally using both name and key in prompt - the invocation throws if an incompatible base model is used, not sure if that is new behavior or not. I would expect it to just the user embedding as plaintext I think, not sure.

[brandonrising]

This worked well for me locally using both name and key in prompt - the invocation throws if an incompatible base model is used, not sure if that is new behavior or not. I would expect it to just the user embedding as plaintext I think, not sure.

It probably makes sense to add that exception to the catch block and let it pass through

[hipsterusername]

This worked well for me locally using both name and key in prompt - the invocation throws if an incompatible base model is used, not sure if that is new behavior or not. I would expect it to just the user embedding as plaintext I think, not sure.

It probably makes sense to add that exception to the catch block and let it pass through

Agree - If it's invalid, should just pass through as text.

[brandonrising]

Should never fail a graph for any exception while trying to load TIs now @hipsterusername @maryhipp