AI research papers.

Realistic online handwriting depends not only on individual character shapes, but also on how a writer connects, spaces, and aligns adjacent characters.

Repository-level software engineering (SWE) comprises heterogeneous task categories, whose progress under pooled agentic reinforcement learning can be uneven: gains in some categories coincide with regressions in others,…

Humans can effortlessly localize the direction of a sound source and integrate it with visual cues for reasoning, yet this remains challenging for embodied agents.

Standard positional encodings represent position as a one-dimensional reading-order coordinate, but reading order alone does not determine hierarchical textual structure.

Despite impressive visual quality, state-of-the-art video diffusion models often generate content that violates real-world physical laws.

Robotic bin packing requires long-horizon sequential decision-making, as each object placement affects the available space for subsequent packing.

Single-image 3D object generation can now produce high-fidelity assets, yet accurately placing them into a coherent scene layout remains an open challenge.

Manipulation policies must know which objects matter and where they are, yet the pretrained backbones that current robot foundation models build on, from language in vision-language-action models (VLAs) to video generati…

Collective intelligence depends not only on what team members know, but also on how they organize their work.

Reinforcement learning (RL) has become a key technique for improving the reasoning and agentic abilities of large language models (LLMs).

Reinforcement learning (RL) is vital for optimizing video generation models, with a robust reward model (RM) serving as the cornerstone.

Humans can seamlessly adapt to both physical and digital worlds, suggesting that while a digital-to-real gap exists in embodiment, environment and task, human intelligence itself may transfer across this gap.

Spatial reasoning is essential for vision-language models (VLMs) to understand and act in the physical world.

Modern Transformer design and compression both reduce to allocating capacity under a budget.

Educational foundation models must solve problems, understand curriculum structure, diagnose learner difficulties, and provide appropriate instructional support.

High-quality texture generation is essential for creating realistic and production-ready 3D assets.

arXiv:2603.26556v3 Announce Type: replace-cross Abstract: Converting a pretrained Transformer into a more efficient hybrid model through distillation offers a promising approach to reducing inference costs.

arXiv:2609.20779v1 Announce Type: cross Abstract: Safety evaluations for large language models rely on surface-form classifiers that report declining harm scores across model generations.

Pretrained vision-language-action (VLA) models enable broad manipulation but remain unreliable in tasks demanding precision and repeatability.

Human video datasets offer an abundant and diverse source of interaction data that can complement expensive real-robot data.

Standard subword tokenizers either treat every orthographic variation of a word (such as hello, Hello, HELLO, and Héllo) as unrelated vocabulary entries, which fragments the embedding space, or discard this variation thr…

Mixture-of-Experts (MoE) scales capacity, but existing designs cannot set three quantities independently.

We define OmniVChat (Omni Video Chat) as the task of native audio-visual dialogue between a user and an omni model.

On-policy distillation (OPD) improves reasoning models by learning the token-level discrepancy between a stronger teacher and an on-policy student.

Skills can improve the performance of Large Language Model (LLM) agents by providing task-specific procedural guidance, while skill optimization further improves their effectiveness through iterative refinement.

A pretrained robot foundation policy may execute most of a long-horizon task yet repeatedly fail at a few critical subtasks.

Large language models can hold knowledge they do not report.

Computer-use agents (CUAs) have advanced along two separate lines: graphical interaction and software development through code and the command line.

We need studies on conversational AI (CAI) at scale to understand human behavior and shape CAI design.

Training capable coding agents via reinforcement learning (RL) requires diverse tasks with reliable verifiers.

Reference-to-video (R2V) generation is evolving toward increasingly general and versatile reference control, giving rise to the emerging paradigm of omni R2V generation.

APort Vault is a benchmark for payment authorization in tool-using AI agents.

We present MintAct, a family of vision-language models that unifies UI grounding, multi-step navigation across mobile, desktop, and web, and visual tool use, trained at 2B, 4B, and 8B scales.

Professional graphic design is a long-horizon agentic task in which structured, editable artifacts emerge from many interdependent actions, yet outcomes admit no reliable programmatic oracle.

Multi-modal image generation, particularly subject-driven customization, has garnered growing attention in recent years.

Creating deformable assets for robot manipulation requires jointly specifying their geometry, appearance, and physical properties.

Telecom fraud scripts evolve rapidly and are often designed to resemble routine service conversations, creating two key requirements for audio-based telecom-fraud evaluation.

Visual signals require compact yet sufficient representations for robust downstream prediction.

Spatial intelligence requires more than describing object locations.

Information retrieval is increasingly important as LLM agents tackle complex tasks involving diverse information needs.

Large Reasoning Models (LRMs) achieve strong performance on complex tasks but exhibit systematic inefficiency: they often overthink easy problems and underthink hard ones.

Fine-grained visual perception in MLLMs is commonly improved by raising the resolution, but the added visual tokens inflate vision-encoding and language-model prefilling costs.

Question answering has advanced rapidly with large language models, but predominantly for high-resource languages, in both text and spoken settings.

The widespread adoption of long-horizon agents has made model workloads increasingly input-heavy.

Document parsing converts document images into structured content and requires reliable performance across diverse layouts and acquisition conditions.

We study length inflation in on-policy distillation (OPD), where student responses can become excessively long and even exhaust the generation budget.

As coding agents move from supervised code completion to unattended, around-the-clock exploration, their work expands from isolated predictions into long trajectories of reasoning, tool use, and feedback.

On-policy self-distillation (OPSD) lets a language model learn from a frozen copy of itself that sees an answer or a worked solution.

Text-guided image editing must introduce the requested changes while preserving unrelated source content.