AI at Meta · 17 records

dateidtitleauthorsvenue↑
2026-11-192611.19200LogT: Logically Think with Images for Visual SearchYanjun Fu et al.arXiv13
2026-10-082610.09484MIMESIS: Learning User Simulators as Training Environments for Interactive AgentsHoang Phan et al.arXiv23
2026-10-012609.40316Scaling Laws for Looped Mixture of ExpertsYanbei Chen et al.arXiv18
2026-08-192608.17402MoE-ViE: Mixture of Experts Vision Encoder for Efficient Image and Video UnderstandingBonan Zhang et al.arXiv20
2026-08-102608.07222Skaling: Chinchilla's Exponents Meet Kaplan's CouplingMathurin Videau et al.arXiv15
2026-08-062608.05000Towards Physics of Multimodal Pretraining: Knowledge Flow, Modality Synergy, Early Unification, and RecipesJunlin Han et al.arXiv60
2026-07-292607.25970Reinforcement Learning for Code OptimizationPierre Chambon et al.arXiv12
2026-07-222607.19322Two-Level Meta-Rubrics for Evaluating Open-Ended Generation: GAMUT, a Benchmark for Factual CompletenessXilun Chen et al.arXiv10
2026-07-092607.07386Sparse Delta Memory: Scaling the State of Linear RNNs through SparsityLoïc Cabannes et al.arXiv12
2026-06-302606.28480TUA-Bench: A Benchmark for General-Purpose Terminal-Use AgentsShoufa Chen et al.arXiv48
2026-06-262606.21795Discretizing Reward ModelsVijay Viswanathan et al.arXiv16
2026-06-252606.25996Autodata: An agentic data scientist to create high quality synthetic dataIlia Kulikov et al.arXiv21
2026-06-182606.19162The Reward Was in Your Data All Along: Correcting Flow Matching with Discriminator-Guided RLNicolas Beltran-Velez et al.arXiv21
2026-06-152606.14700RepFusion: Leveraging Multimodal Priors for Denoising in Representation SpaceXichen Pan et al.arXiv17
2026-06-012605.30561VLM3: Vision Language Models Are Native 3D LearnersZhipeng Cai et al.arXiv24
2026-05-272605.27358MobileMoE: Scaling On-Device Mixture of ExpertsYanbei Chen et al.arXiv15
2026-05-152605.13852Realiz3D: 3D Generation Made Photorealistic via Domain-Aware LearningIdo Sobol et al.arXiv24