AI research papers.

Looped Transformers reuse one block of layers several times: by spending extra computation they push a model of fixed size further, and so use its parameters more fully; while sparse mixture-of-experts (MoE) models activ…

When a large language model (LLM) agent executes the same task, token consumption can vary by over an order of magnitude across runs.

Unified multimodal models can both look at and render images, so in principle they can repair their own generations: diagnose what an image gets wrong, revise it, observe the result, and diagnose again.

A voice agent can handle almost every call on the words alone and still fail the few its sector's rules were written for.

Multi-agent LLM systems increasingly mix models from several providers, yet exposing each agent's underlying model identity to its peers significantly impairs cooperation.

Prompt injection against LLM agents becomes much stronger when the injected instruction is wrapped in the model's own chat template.

Large language model post-training generates self-generated rollouts through reinforcement learning and on-policy distillation, yet this experience is often treated as stale once the policy advances.

Pixel-space diffusion Transformers (DiTs) directly operate on high-dimensional visual data, yet their hidden representations typically undergo uniform refinement across depth.

We present LongCat-DeepResearch, a deep research system that combines an enhanced LongCat model with a multi-agent workflow for producing comprehensive, evidence-grounded reports.

Multi-step matching models, including flow and diffusion models, produce high-quality outputs but incur substantial inference costs and may reproduce unwanted components of their training datasets.

Long-running LLM agents compress past interactions into persistent memories that may be reused as premises for later tasks.

Before invoking external tools, an agentic LLM must select among a K-way action space: executing a call, seeking clarification, answering directly, or declining.

As large language models are deployed in increasingly autonomous long-horizon tasks, manually auditing and verifying the actions, artifacts, and outputs of models becomes more difficult.

Reinforcement learning has helped AI agents solve increasingly difficult tasks, but high rewards do not always reflect the work users intended.

State Space Models (SSMs) compress sequence history into a bounded recurrent state, making the resulting memory law a central architectural choice for long-context performance.

Chunked KV-cache compression reduces the memory and attention costs of long-context inference by compressing windows of consecutive tokens into fewer cache entries at a fixed stride.

Vision-language-action (VLA) models perform well on shorter-horizon manipulation tasks but still struggle with long-horizon tasks that require multiple dependent manipulations from a single command.

A 3D scene reconstructed from a single image is most useful when represented not as a rendering or a fixed 3D output, but as an explicit scene program whose execution yields a scene that can be inspected, edited, and que…

An activation-steering coefficient sets intervention strength, but requesting a particular degree of persona expression requires a behavioral scale.

We propose RegLLM, a diagnostic harness for bounded autonomy in regulated agentic workflows.

When generating text, a Transformer produces representations of past tokens at every layer, but each layer can normally use only representations from the same depth.

High-performance GPU kernels are essential to modern machine learning systems, yet automatically generating kernels that are both correct and efficient remains challenging.

Generating responsive listener facial motion is an important task for embodied conversational AI.

World-action models give visual navigation agents a way to anticipate how candidate actions will change future observations and to act from the predicted consequences.

Diffusion Transformers (DiTs) have established themselves as a scalable backbone for high-fidelity image synthesis.

We present VGGT-Diff, a geometry-routed multi-view diffusion model for sparse-view novel view synthesis.

An agent can complete a task while exhibiting undesirable behavior during execution.

Personal Graphical User Interface (GUI) agents rely on interaction history to infer what a user wants from ambiguous instructions and to anticipate recurring routines.

Moving ML-mediated decision making onto privacy-preserving clients decentralises the economic decision along with the inference.

Visual backbones have evolved from Convolutional Neural Networks (CNNs) with local aggregation to Vision Transformers (ViTs) with global interactions, State-Space Models (SSMs) with input-dependent state transitions, and…

Information-seeking agents increasingly operate over information spaces that are too large to process exhaustively.

Masked diffusion language models (MDMs) admit flexible generation orders, making the unmasking strategy an inference decision.

A central goal in robotics is to enable manipulation across changing tasks and environments.

Coding-agent evaluation has progressed from resolving individual issues to carrying out long-horizon development, yet task completion is still largely assessed within a single codebase.

Reinforcement learning with verifiable rewards (RLVR) often relies on sparse outcome rewards, providing coarse supervision for long-horizon agents.

In realistic education, a solution is often expressed not only in words but in a drawing--a circuit, a geometric construction, a function plot--and a teacher must grade the drawing as carefully as the text.

Data videos communicate data insights through dynamic charts, voice narration, and synchronized animations, and have become a widely adopted form of data storytelling.

Comparisons in video self-supervised learning often evaluate complete training recipes rather than isolating the method itself: architecture, objective, data exposure, schedule, scale, and decoder capacity can all vary a…

Model merging integrates the capabilities of multiple experts without joint retraining, but standard expert training optimizes task loss alone and does not guarantee good performance after merging.

Supervised fine-tuning (SFT) learns most aggressively from tokens that the model deems least likely.

While Large Language Models (LLMs) advertise million-token context windows, reasoning quality often collapses as inputs grow -- a phenomenon termed context rot.

EEG foundation models hold promise for scalable brain-signal decoding across clinical and cognitive neuroscience applications, yet their pre-training pipelines remain poorly understood.

Low-precision Transformer systems increasingly quantize attention matrix multiplications, while softmax often remains at higher precision.

Vision-language models (VLMs) can benefit from geometric priors for multi-view spatial reasoning, yet answer-only training does not directly supervise the intermediate geometric estimates and their use in deriving quanti…

Modern language models are likely to be updated throughout their lifetime rather than trained once and frozen.

Despite recent progress in agentic multi-shot video generation, producing coherent and consistent long-form stories remains challenging.

Real-world tasks often require large language models (LLMs) to learn from complex task-specific context rather than pretrained parametric knowledge.

Connectionist temporal classification (CTC) naturally supports offline and streaming speech recognition with utterance-level supervision, but conventional implementations materialize frame-by-vocabulary activations in me…

Multimodal large language models can expose task-relevant evidence through generation, but producing useful evidence does not by itself determine how it enters a retrieval embedding.

Automatically generated environments provide a scalable source of interaction data for training general agents.