Hola! I’m Oscar Mañas. I’m a Research Scientist at Meta Superintelligence Labs on the Media Generation team in Zurich. I recently completed my PhD at Mila and Université de Montréal, advised by Prof. Aishwarya Agrawal.

My research explores the intersection of vision and language, with a focus on multimodal vision-language generative models: systems capable of generating images, videos and text from multimodal inputs. I’m especially interested in building models that reason fluidly across modalities, treating them as complementary channels of perception and thought.

Previously, I was a Visiting Researcher at Meta FAIR and a Research Intern at Element AI. See my CV for more details.

News

Selected Publications

SneakPeek: Future-Guided Instructional Streaming Video Generation
Cheeun Hong, German Barquero, Fadime Sener, Markos Georgopoulos, Edgar Schönfeld, Stefan Popov, Yuming Du, Oscar Mañas, Albert Pumarola
arXiv 2025
paper

Controlling Multimodal LLMs via Reward-guided Decoding
Oscar Mañas, Pierluca D'Oro, Koustuv Sinha, Adriana Romero-Soriano, Michal Drozdzal, Aishwarya Agrawal
ICCV 2025
first author paper

Improving Text-to-Image Consistency via Automatic Prompt Optimization
Oscar Mañas, Pietro Astolfi, Melissa Hall, Candace Ross, Jack Urbanek, Adina Williams, Aishwarya Agrawal, Adriana Romero-Soriano, Michal Drozdzal
TMLR 2024
first author paper

Improving Automatic VQA Evaluation Using Large Language Models
Oscar Mañas, Benno Krojer, Aishwarya Agrawal
AAAI 2024
first author paper

MAPL: Parameter-Efficient Adaptation of Unimodal Pre-Trained Models for Vision-Language Few-Shot Prompting
Oscar Mañas, Pau Rodríguez, Saba Ahmadi, Aida Nematzadeh, Yash Goyal, Aishwarya Agrawal
EACL 2023
first author paper code

Seasonal Contrast: Unsupervised Pre-Training from Uncurated Remote Sensing Data
Oscar Mañas, Alexandre Lacoste, Xavier Giró-i-Nieto, David Vázquez, Pau Rodríguez
ICCV 2021
first author paper

View all publications →