Develop a framework for generating synthetic reasoning data using model-in-the-loop validation. This will help smaller teams build high-quality fine-tuning sets.
Suggested repo: synapseData
"Generate high-quality reasoning datasets automatically."
Estimated effort: 150h