What are the two techniques discussed for strongly influencing Claude's output besides changing the prompt?	Pre-filling assistant messages and stop sequences.
How is the technique of pre-filling an assistant message implemented?	By manually adding an assistant message to the list of messages before sending the prompt to the model.
What is the primary effect of pre-filling an assistant message?	It steers the model's response, causing it to continue generating text from the end of the pre-filled content.
What is a critical limitation of using pre-filled assistant messages?	The final output is often a partial response that requires manual stitching to complete.
What is the function of a stop sequence?	To force the LLM to immediately cease generating output upon producing a specific string.
How are stop sequences provided to the model?	As an optional parameter (a list of strings) within the API call or function configuration.
If a stop sequence is provided, what happens to the sequence itself when the model generates it?	The generation stops immediately, and the sequence is not included in the final output.