Researchers investigate whether training policies to predict both actions and environmental changes using World Action Models improves robot manipulation learning from egocentric human videos.
Peer-Predictive Self-Training is a collaborative, label-free fine-tuning framework where multiple language models use aggregated responses to mutually improve their reasoning capabilities.
AWS has launched the Claude apps gateway, a self-hosted control plane allowing organizations to manage access, costs, and policies for Claude Code and Claude Desktop.
NVIDIA announced that its Nemotron 3 Ultra model achieved high accuracy and throughput on agent tasks when optimized with LangChain's Deep Agents harness.
Jet-Long introduces a dynamic bifocal rotary position embedding method to efficiently extend the context window of large language models in a zero-shot manner.
A new auditing technique called overthinking uses task vectors to amplify a model's reasoning process, helping to expose hidden information or misaligned behaviors.
This paper addresses feature entanglement in language models by proposing a method to encourage almost orthogonal features, enabling more precise and isolated mechanistic interventions.
A new study reveals that training models on data following a power-law distribution, rather than a curated uniform distribution, actually improves their ability to perform compositional reasoning tasks.
The paper presents MultiSim, an ensemble approach that utilizes multiple driving simulators to identify and mitigate testing discrepancies in automated driving assistance systems.
Chinese memory chip manufacturer ChangXin Memory Technologies has initiated its STAR Market IPO process to raise 29.5 billion RMB.
Qatari IT service provider Meeza has completed a data center expansion project for an unnamed major hyperscale cloud provider ahead of schedule.
Researchers have demonstrated that hackers can exploit security vulnerabilities in popular large language models to construct large-scale botnets through a technique called HalluSquatting.
Alibaba's stock price rose significantly in Hong Kong following optimistic analyst projections for revenue growth driven by rising demand for artificial intelligence and its proprietary T-Head chips.
This paper compares softmax attention with four recurrent linear-attention architectures to analyze their mechanisms and trade-offs for long-context processing.
Omni-Sleep is a new foundation model designed to analyze sleep physiology by modeling the coordinated dynamics of the central and autonomic nervous systems.
Researchers have found that persuasion-based jailbreak attacks can undermine the effectiveness of chain-of-thought monitoring in detecting misaligned AI behavior.
The paper introduces TAG, a lightweight framework designed to improve the reliability of LLM-generated structured artifacts through rigorous, test-driven validation.
The V-VLAPS framework integrates value-guided planning with vision-language-action models to improve robotic manipulation performance under distribution shifts and long-horizon tasks.
Researchers have introduced ParamMute, a method that improves the faithfulness of retrieval-augmented generation by suppressing knowledge-critical feed-forward networks that cause models to ignore retrieved context.
The Echoes dataset provides over 130 hours of audio across multiple genres to benchmark and train robust music deepfake detection systems.
This position paper argues that current retrieval-augmented generation systems are overly focused on factual grounding and fail to adequately represent diverse opinions.
The ABot-C0 technical report details a new framework for quadrupedal robot control designed to overcome the scarcity of animal motion-capture data.
VocaDet is a new open-vocabulary object detection and segmentation framework that scales to large object categories using visual tokenization and vector database retrieval.
Researchers developed TRACE, a robust watermarking method designed to protect and verify the attribution of LLM agent trajectory logs against unauthorized rebranding or model substitution.