Distill Locally, Schedule Globally: Flow Maps for Few-Step Text-to-Speech
Flow-map distillation of large-scale flow-matching text-to-speech models for few-step inference.
Machine learning research
Yentl Collin
AI Scientist Intern · IFM / MBZUAI
I work on generative models at the Institute of Foundation Models (IFM) at MBZUAI, with the Speech / LLM team. My interests lie in flow matching, discrete diffusion, and efficient generation for speech and language. I am also interested in recurrent and subquadratic sequence models, including SSMs, xLSTMs, linear attention, and hybrid architectures.
I hold a master’s degree in Mathematics, Vision and Learning (MVA) from ENS Paris-Saclay and an engineering degree from École des Ponts et Chaussées (ENPC). Previously, I worked on recurrent models and state-space models (SSMs) for interpretable time-series learning in Chirag J. Patel’s lab at Harvard University.
Research interests
Generative modelingEfficient foundation modelsTime series & tabularDiscrete diffusionLLMs & speech
Selected publications
All publications
SSM-CGM: Interpretable State-Space Forecasting Model of Continuous Glucose Monitoring for Personalized Diabetes Management
Interpretable state-space modeling for continuous glucose forecasting and personalized diabetes management.
Notes & writing
A notebook for ideas worth keeping.
Research notes, readings, and occasional detours.
First entries coming soon.
Always happy to talk research.
yentl.collin@eleves.enpc.fr ↗