The Year in AI Papers
The year/Labs/Meta AI

Research lab

Meta AI

Open research from Meta AI and FAIR across vision, language, multimodal systems, and embodied intelligence.

Papers
12
Citations
2.4K
Official code
8

12 papers from Meta AI

01

arXiv.org

Tuna-2: Pixel Embeddings Beat Vision Encoders for Multimodal Understanding and Generation

Tuna-2 is a native unified multimodal model that performs visual understanding and generation directly from raw pixel embeddings, eliminating pretrained vision encoders such as VAEs and representation encoders. It uses simple patch embedding layers to encode images and a single transformer decoder for joint processing, with pixel-space flow matching for…

Zhiheng Liu, Weiming Ren, Xiaoke Huang, Shoufa Chen, et al.
Published
Apr 2026
Citations
10
Code
744 stars
Read summary Original
02

arXiv.org

Think in Strokes, Not Pixels: Process-Driven Image Generation via Interleaved Reasoning

This paper introduces process-driven image generation, a multi-step paradigm that decomposes text-to-image synthesis into an interleaved reasoning trajectory of textual planning and visual generation. The method uses a recurring four-stage cycle: Plan, Sketch, Inspect, and Refine, where the model generates incremental instructions and scene descriptions,…

Lei Zhang, Junjiao Tian, Zhipeng Fan, Kunpeng Li, et al.
Published
Apr 2026
Citations
3
Code
Not linked
Read summary Original
03

arXiv.org

Beyond Language Modeling: An Exploration of Multimodal Pretraining

This paper presents controlled, from-scratch experiments to clarify the design space of unified multimodal pretraining, using the Transfusion framework (next-token prediction for language, diffusion for vision) on text, video, image-text pairs, and action-conditioned video. Key findings: (1) Representation Autoencoders (RAE), e.g., SigLIP 2, provide a…

Shengbang Tong, David Fan, John Nguyen, Ellis Brown, et al.
Published
Mar 2026
Citations
21
Code
Not linked
Read summary Original
04

arXiv.org

AIRS-Bench: a Suite of Tasks for Frontier AI Research Science Agents

AIRS-Bench is a benchmark suite of 20 tasks sourced from 17 state-of-the-art machine learning papers, designed to evaluate LLM agents on the full research lifecycle (idea generation, implementation, experimentation, analysis) without providing baseline code. Tasks span seven categories: language modeling, math, code, molecules/proteins, time series, text…

Alisia Lupidi, Bhavul Gauri, Thomas Simon Foster, Bassel Al Omari, et al.
Published
Feb 2026
Citations
14
Code
106 stars
Read summary Original
05

arXiv.org

TUNA: Taming Unified Visual Representations for Native Unified Multimodal Models

Tuna is a native unified multimodal model (UMM) that creates a unified continuous visual representation by cascading a VAE encoder with a representation encoder (SigLIP 2). This design avoids the representation format mismatches of decoupled models, improving both understanding and generation. The model uses an LLM decoder (Qwen2.5) for autoregressive text…

Zhiheng Liu, Weiming Ren, Haozhe Liu, Zijian Zhou, et al.
Published
Dec 2025
Citations
31
Code
94 stars
Read summary Original
06

arXiv.org

SAM 3: Segment Anything with Concepts

SAM 3 is a unified model for promptable concept segmentation (PCS) in images and videos, accepting noun phrases, image exemplars, or both as prompts to detect, segment, and track all matching instances. It decouples recognition and localization via a presence head, improving detection accuracy. A data engine with human and AI verifiers produced 4M unique…

Nicolas Carion, Laura Gustafson, Yuan-Ting Hu, Shoubhik Debnath, et al.
Published
Nov 2025
Citations
711
Code
11K stars
Read summary Original
07

arXiv.org

SAM 3D: 3Dfy Anything in Images

SAM 3D is a generative model for 3D object reconstruction from a single image, predicting geometry, texture, and layout. It excels in natural images with occlusion and clutter, using a human- and model-in-the-loop pipeline to create large-scale 3D annotation data. The model uses a multi-stage training framework: synthetic pretraining on 2.7M meshes…

SAM 3D Team, Xingyu Chen, Fu-Jen Chu, Pierre Gleize, et al.
Published
Nov 2025
Citations
197
Code
7.2K stars
Read summary Original
08

arXiv.org

What Does It Take to Be a Good AI Research Agent? Studying the Role of Ideation Diversity

This paper investigates the role of ideation diversity in AI research agents' performance on MLE-bench, a benchmark of 75 Kaggle machine learning tasks. The authors analyzed 11,000 agent trajectories from 6 LLM backbones and 3 scaffolds, measuring ideation diversity via Shannon entropy of model architectures in initial drafts. They found that…

Alexis Audran-Reiss, Jordi Armengol-Estapé, Karen Hambardzumyan, Amar Budhiraja, et al.
Published
Nov 2025
Citations
5
Code
Not linked
Read summary Original
09

arXiv.org

Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance

The paper introduces Soup Of Category Experts (SoCE), a model souping technique that improves LLM performance by leveraging benchmark composition and non-uniform weighted averaging. SoCE identifies weakly-correlated benchmark categories, selects expert models for each, and optimizes weights to combine them. Experiments on BFCL, MGSM, and ∞-Bench show SoCE…

Shalini Maiti, Amar Budhiraja, Bhavul Gauri, Gaurav Chaurasia, et al.
Published
Nov 2025
Citations
4
Code
73 stars
Read summary Original
10

arXiv.org

TruthRL: Incentivizing Truthful LLMs via Reinforcement Learning

TruthRL is a reinforcement learning framework that directly optimizes the truthfulness of large language models (LLMs) by using a ternary reward scheme that distinguishes correct answers, hallucinations, and abstentions. Implemented with GRPO, it rewards correct answers, penalizes hallucinations, and treats abstentions neutrally, encouraging models to…

Zhepei Wei, Xiao Yang, Kai Sun, Jiaqi Wang, et al.
Published
Sep 2025
Citations
16
Code
Not linked
Read summary Original
11

arXiv.org

Deep Think with Confidence

Deep Think with Confidence (DeepConf) is a test-time method that improves LLM reasoning efficiency and accuracy by using model-internal confidence signals to filter low-quality reasoning traces. It addresses the diminishing returns and high computational overhead of self-consistency with majority voting. DeepConf operates in offline mode (filtering…

Yichao Fu, Xuewei Wang, Yuandong Tian, Jiawei Zhao
Published
Aug 2025
Citations
151
Code
409 stars
Read summary Original
12

Research paper

DINOv3

DINOv3 is a self-supervised vision foundation model that scales to 7B parameters, trained on a curated dataset of 1.689 billion images (LVD-1689M) from Instagram, combined with ImageNet-1k. The model uses a ViT architecture with RoPE embeddings and constant hyperparameter schedules. A key contribution is Gram anchoring, a regularization technique that…

Oriane Siméoni, Huy V. Vo, Maximilian Seitzer, Federico Baldassarre, et al.
Published
Aug 2025
Citations
1.2K
Code
11K stars
Read summary Original