Researcher
Co-author of 'OpenThoughts: Data Recipes for Reasoning Models' (2025), cited for work on distillation for reasoning models
How media typically covers Sudhir Guha
Research or work cited
On-policy distillation training approaches enable smaller specialized models to match or exceed larger generalist models' performance in focused domains through efficient knowledge transfer from teacher models.
“Co-author of 'OpenThoughts: Data Recipes for Reasoning Models' (2025), cited for work on distillation for reasoning models”