Investigate how to pass the "flow" (sequence or user-defined structure) into prompts to guide LLM output. Explore the relationship between flow and relevance in generated content.

• Scope:

  • Test methods to encode and pass "flow" into LLM prompts.
  • Experiment with different flow strategies (e.g., using markers or hierarchical structures).
  • Evaluate the impact of "flow" on output quality and consistency.

• Outcome:

  • Summary of findings on flow integration.
  • Defined method for "flow" integration in prompts.