AI research papers.

How much of a running workflow should a language model agent revise?

Video-language models are increasingly used as judges of video understanding, both for evaluating model outputs and for training reward models.

Scaling test-time computation is a powerful way to improve language-model reasoning, and is particularly appealing for small reasoning models (sRMs) that are cheap to serve.

Video and audio are perceived together, yet most generative models treat them in isolation.

Language model agents serving organizations must coordinate requests from multiple users while using knowledge distributed across their interactions.

We present EditWorld, a video world model for precise editing and flexible referencing in interactable worlds.

Vision-language (VL) pretraining using paired chest X-ray (CXR) images and radiology reports has shown strong potential for medical image understanding.

Language-model agents increasingly use tools to act on external systems.

Open-vocabulary semantic segmentation (OVSS) enables pixel-level prediction over arbitrary text-specified vocabularies and has shown strong generalization on common benchmarks.

Latent visual reasoning (LVR) enables multimodal large language models (MLLMs) to perform intermediate computation in continuous latent tokens rather than expressing every reasoning step in words.

Reinforcement learning (RL) is increasingly used to post-train vision-language-action (VLA) models, yet how RL reshapes these policies remains poorly understood.

Multi-teacher on-policy distillation (MOPD) has emerged as a popular post-training paradigm for integrating specialized capabilities in frontier language models.

Native 3D texture generation synthesizes colors directly in 3D space for a given geometry, conditioned on multi-view reference images.

Reinforcement learning (RL) post-training for large language models (LLMs) coordinates multiple models across generation, inference, and training on GPU clusters.

World models predict future observations from current experience and actions, yet prediction can depend on observations seen far in the past.

Physical trajectories contain more than snapshots of a system: they also reveal how its states evolve under governing conditions.

Long-horizon camera-controlled video generation requires recovering previously observed content from an ever-growing visual history.

Large language model (LLM) routing aims to select the most suitable model for each incoming query.

Latent communication passes internal states between language models instead of decoded text, but higher receiver accuracy does not show that the receiver used the message content.

Recent vision-language models (VLMs) have advanced vision-and-language navigation (VLN), enabling models to predict navigation actions from visual observations and language instructions.

Reliable AI safeguards require both control mechanisms that reduce unsafe behavior and monitoring mechanisms that detect safety risks during model interactions.

Humans often solve spatial problems by mentally simulating visual transformations.

RLVR provides reliable trajectory-level credit, while OPSD offers dense supervision for token-level credit.

Multi-vector retrievers built on vision-language models lead visual document retrieval (VDR), but they run a multi-billion-parameter query encoder on every search.

Safety-aligned LLMs can exhibit emergent misalignment (EM): narrow domain adaptation unexpectedly triggers catastrophic safety failures across unrelated domains.

Long visual token sequences often account for a substantial fraction of the computational overhead in multimodal large language models~(MLLMs).

Recent gains in language model capability have come more from data than from architecture.

Rendering document text as images allows vision-language models to encode documents as visual tokens, which can reduce input sequence length compared with text input.

On-policy distillation (OPD) is a widely adopted post-training technique for LLM reasoning.

Visual-token compression is effective for improving the efficiency of vision-language models, but under extreme compression budgets, token pruning can break visual grounding while learned resamplers increase parameter co…

When a VLM answers a visual query, current interpretability tools rely on text rationales, which use a mismatched modality, or on internal read-outs, which originate too early to reflect the final output and require whit…

On-policy learning has been argued to reduce catastrophic forgetting, produce sparser parameter updates, and improve generalisation.

As language models take a growing role in AI development, a natural aspiration is for them to reflect on their own learning process, as humans do, and use that reflection to improve themselves.

Multi-agent large language model (LLM) debate is often evaluated by whether final answers improve, but movement is not necessarily improvement: the same discussion can rescue an initially wrong majority or destroy an ini…

Modern AI models are aligned through post-training to adapt them to downstream tasks.

The appearance of Jev marked the era of System One Models, foundation models that return structured decisions with probability distributions rather than text.

Reinforcement learning can turn one language model into several specialists, each excellent at a single skill such as mathematics, coding or following instructions, but users need one model with all of these skills.

In response to the growing demand for long sequences in agentic and reasoning use cases, many state-of-the-art LLMs combine multiple variants of attention to mitigate the quadratic complexity of traditional softmax atten…

Modern LLMs are increasingly capable as autonomous agents, but they follow sequential interaction cycles: read, think, reply or call tools, repeat.

Vision-language-action (VLA) and world-action (WAM) models map observations and instructions directly to robot actions.

Reinforcement learning from verifiable rewards (RLVR) frequently reuses rollouts across multiple policy updates, increasing the mismatch between the current policy and the data-generating policy.

Most modern multimodal large language models (MLLMs) build on a pretrained visual encoder that provides a strong visual prior.

Discrete diffusion language models (dLLMs) expose a denoised solution at every step, which makes process reward model (PRM) guidance look like a way to spend compute at test time.

Machine intelligence is often evaluated through abstract reasoning problems, yet many real-world problems are visual, such as arranging objects, repairing layouts, or tracing routes.

We study on-policy distillation (OPD) through the lens of reinforcement learning, establishing a connection between the reverse-KL objective in OPD and KL-regularized policy optimization.

In ecology, psychometrics, and the analysis of social and financial networks, binary matrices are often analyzed conditional on their observed row and column sums, which restricts the problem to a finite sample space of…

Recent image generation models can take multiple reference images as input and combine them into a new image.

In multi-agent systems, reliable consultation is challenging because advisor capabilities vary across tasks, and misleading information can make consultation worse than autonomous reasoning.

Interactive world models require responding in real time to versatile controls and maintaining long-horizon consistency.

Lookup-based memory has been a promising way to scale the parameters of large language models (LLMs).