LiFT: Does Instruction Fine-Tuning Improve In-Context Learning for Longitudinal Modelling by Large Language Models?

Build a specialized fine-tuning framework that allows models to better handle long-term state tracking. This is critical for medical or behavioral research apps where context is historically dependent.

Suggested repo: lift-train

"Make your LLM remember long-term temporal relationships."

Estimated effort: 50h