Configurable Preference Tuning ⚙️📝
Collection
CPT uses rubric-guided synthetic data and DPO to enable LLMs to dynamically adjust behavior (e.g., writing style) at inference with system prompts
•
7 items
•
Updated
•
1