Prompting PTI concept in merged LoRas

[levi]

Two simple questions:

1. If I merge two LoRa models that were trained with PTI, how do the tokens work post merge? For example, I have one model with <s1><s2> and another with <s1><s2><s3>. How do these coexist and how do I prompt properly?

2. If patch a dreambooth model with a trained concept (ie. "analog style") with a LoRa model trained with PTI, how do I prompt the style and the PTI concept? Is it as simple as "analog style, <s1><s2>"?

Cheers!