@besSveta , because of some technical aspects, in order to make nested scripts interesting (e.g. to support choices among sub-scripts in hierarchical templates), I moved interpret from Agent and messages sections on top to be applied to all expressions in the script. This should add a lot of flexibility to prompt templates, but while it doesn't affect unit tests, I didn't check if it works well for examples. Please, check that the examples and SNet assistant scripts are not broken by this change.
@besSveta , because of some technical aspects, in order to make nested scripts interesting (e.g. to support choices among sub-scripts in hierarchical templates), I moved
interpretfromAgentand messages sections on top to be applied to all expressions in the script. This should add a lot of flexibility to prompt templates, but while it doesn't affect unit tests, I didn't check if it works well for examples. Please, check that the examples and SNet assistant scripts are not broken by this change.