Lattice AI Research

Research focus

Computer Vision (2)Inference & Quantization (1)Training Efficiency & Optimization (1)Architecture Design (Transformers, SSMs, MoE) (1)Multimodal Models (1)

Frequent co-authors

Aram Davtyan (2)Leello Tadesse Dadi (1)Volkan Cevher (1)Aram Davtyan (1)

Papers (2)

Mar 16, 2026

Faster Inference of Flow-Based Generative Models via Improved Data-Noise Coupling

Squeeze 2x more speed from your conditional flow matching models by optimizing data-noise coupling across minibatches.

Aram Davtyan, Leello Tadesse Dadi, Volkan Cevher +1

Computer Vision Inference & Quantization Training Efficiency & Optimization

Feb 24, 2026

Aram Davtyan +11Feb 24, 2026

Communication-Inspired Tokenization for Structured Image Representations

Object-centric image tokenization is now possible by modeling the sequential refinement process of human communication, leading to better compositional generalization and relational reasoning.

Aram Davtyan, Aram Davtyan, Y. Şahin +9

Architecture Design (Transformers, SSMs, MoE)Computer Vision Multimodal Models

Search

Paolo Favaro

Research focus

Frequent co-authors

Papers (2)