…AI-generated summary Model merging combines task experts into one model and avoids joint training, retraining, or deploying many expert models, but the merged model often still underperforms task experts . We study…
…Adaptive Coordination of Understanding and Generation for Unified Multimodal Reasoning Published on May 12 Submitted by Jindong Wang on May 13 The College of William & Mary Authors: , , , , Abstract Unified multimodal models can…
…Training LLMs for Long-Lifecycle Agents with Cross-Domain Generalization Via Reinforcement Learning Published on Jun 18 Submitted by Yanxi Chen on Jun 23 TongyiLab Authors: Yanxi Chen , , , , , , Abstract Large language models…
…Benchmarking AI Agents on Workspace Tasks with Large-Scale File Dependencies Published on May 5 Submitted by taesiri on May 6 Authors: , , , , , , , , , , , , , , , , , , , Abstract Workspace-Bench is a benchmark for evaluating AI agents…
…hf papers read 2605.12497 Don't have the latest CLI? curl -LsSf https://hf.co/cli/install.sh | bash No model linking this paper Cite arxiv.org/abs/2605.12497 in…
…CausalCine generates causally across shot changes, accepts dynamic prompts on the fly, and reuses context without regenerating previous shots. To achieve this, we first train a causal base model on native multi…
Papers arxiv:2606.16281 Who Should Lead Decoding Now? Tracking Reliable Trajectories for Ensembling Masked Diffusion Language Models Published on Jun 15 Submitted by Heecheol Yun on Jun 16 KAIST AI Authors…
…A Model-Agnostic Framework for Efficient Feed-Forward 3D Reconstruction Published on May 12 Submitted by HuggingFace Zhang on May 13 Peking University Authors: , Zeyu Zhang , , , Abstract Lite3R addresses efficiency challenges in…
…AI-generated summary Recent generative models can produce images that appear highly realistic, raising challenges in distinguishing real and AI-generated images. Yet existing detectors based on pre-trained feature extractors tend…
…AI-generated summary Speculative decoding speeds up autoregressive generation in Large Language Models (LLMs) through a two-step procedure, where a lightweight draft model proposes tokens which the target model then verifies…