Lattice AI Research

Research focus

Multimodal Models (2)Data Curation & Synthetic Data (1)Eval Frameworks & Benchmarks (1)Architecture Design (Transformers, SSMs, MoE) (1)Inference & Quantization (1)

Frequent co-authors

B. Schiele (2)Muhammad Ferjad Naeem (2)Matteo Farina (1)Vishaal Udandarao (1)

Papers (2)

Jun 26, 2026

Stanford HAIJun 26, 2026·also CMU ML, University of Tubingen

DataComp-VLM: Improved Open Datasets for Vision-Language Models

Data mixing, especially with instruction-heavy data, emerges as the crucial factor for optimizing VLM training, challenging traditional filtering approaches.

Matteo Farina, Vishaal Udandarao, Thao Nguyen +33

Data Curation & Synthetic Data Eval Frameworks & Benchmarks Multimodal Models

May 28, 2026

Google ResearchMay 28, 2026·also Max Planck

PARCEL: Pool-Anchored Resampling with Conditioned Elastic Queries for Efficient Vision-Language Understanding

PARCEL redefines visual tokenization, achieving superior efficiency and performance by dynamically anchoring feature extraction to spatial pool tokens.

S. Kuzucu, Alessio Tonioni, Vasile Lup +3

Architecture Design (Transformers, SSMs, MoE)Inference & Quantization Multimodal Models

Search

S. Kuzucu

Research focus

Frequent co-authors

Papers (2)