Victor Gallego

CPT uses rubric-guided synthetic data and DPO to enable LLMs to dynamically adjust behavior (e.g., writing style) at inference with system prompts

Merges of models of different architectures and sizes that end up working surprisingly well