실시간
국내외 뉴스 · 기업 공식 발표 · 논문 · 영상 · 커뮤니티 소식을 한국시간 기준으로 제공합니다. 같은 소식은 하나로 묶고, 기본은 핵심만 보여 드립니다.
모든 수집 소식(구글 뉴스로 들어온 단건 보도 포함).
Transferring LLMs into System One decision models typically replaces the language modeling head with a decision head while retaining the full backbone, making…
Video generation models produce visually realistic videos but still struggle with physical dynamics and causal consistency.
Multimodal coding agents decouple 4D world construction from visual generation, repurposing video models from world generators toward generative renderers.
We present UniSpark-9B, a unified multimodal model that investigates hybrid attention for visual understanding and generation.
Second-order optimizers (e.g., Shampoo, SOAP) have emerged as compelling alternatives to Adam/AdamW and promise faster per-step convergence than AdamW in…
Decision models answer a question with a typed output, such as a choice among options with a probability for each, so one model can choose an agent's actions…
다른 보도 Latent Space · AI Insider
Large foundation models such as GPT-6 Astra and Claude Fable 5.1 have recently been shown to control robots directly from visual observations.
In this report, we introduce the Reciprocal World Model (RWM) and present Duo v1, a full-duplex world model that continuously perceives audiovisual inputs…
System One models promise fast, general-purpose decision-making without lengthy token generation.
Few-step sampling reduces the cost of streaming video generation. Distribution Matching Distillation (DMD) supervises few-step generators with a separate…
Predicting long action chunks, often spanning dozens of timesteps, has become a key ingredient in increasingly capable robot policies.
Under a fixed frame budget, video understanding with vision-language models (VLMs) depends on whether the sampled frames provide sufficient evidence for the…
Dexterous manipulation requires rich spatial foresight over both the surrounding scene and the robot hand.
Memory-augmented vision-language-action (VLA) policies are meant to remember what happened.
Weight-space programming, introduced in recent MIT work by Takehi et al., enables a frontier LLM agent to read a network's weights and write a repair.
System-1 Decision Models directly make structured probabilistic decisions over query-specified options given a state in a single forward pass, without…
RLVR trains language models to reason by rewarding correct final answers.
Much of the variation in financial time series is driven by market-wide shocks that cannot be predicted from an asset's own history.
Flow-matching models can produce high-quality image restorations, but their learned dynamics may amplify small input perturbations and make reconstructions…
External memory enables LLM agents to improve from experience by distilling trajectories of past tasks into concise lessons for new tasks.
We present Sable, an argument system and single-GPU prover for the statement that a committed, integer-quantized transformer produced a given token sequence…
Interpretability demonstrations often show that intervening on a group of neurons changes a model's answer.
We present TorchCurves, a PyTorch library that treats vector-valued parametric curves as learnable primitives.
As text-to-video (T2V) models are increasingly positioned as world simulators, it is essential to evaluate whether their videos conform to physical commonsense.
Foundation models such as GPT and Claude have shown strong capabilities in spatial understanding and embodied intelligence.
Automated segmentation in multi-view echocardiography is essential for comprehensive cardiac assessment, yet remains challenging due to substantial anatomical…
When two objects share a name, how much does their visual similarity explain, and what do the teaching examples add?
Weight-space programming allows language models to read a trained neural network's weights and edit them.
Low-bit quantization reduces the inference cost of LLM agents but can compromise task success.
Local decoding usually fixes one weight precision for the whole generation.
Typed-decision encoders answer a question about a text in one forward pass, but their behaviour outside English is rarely measured on a fixed yardstick.
We present FAME-QP, a GPU-accelerated solver for dense quadratic programs arising in linear model predictive control.
AI agents deployed on the open web are becoming more capable of performing actions that may breach contractual provisions and violate computer security laws.
Driving world models aim to simulate how a scene unfolds under different ego actions.