New research on attack surfaces for persistent-state AI agents.
New research on attack surfaces for persistent-state AI agents.
source: Distributed Attacks in Persistent-State AI Control
介绍了一种新的方法，通过约束实现对编码代理的可扩展监督。
source: Reasoning LLM Improves Speaker Recognition in Long-form TV Dramas
Novo protocolo de consenso para computação distribuída, oferecendo velocidade e eficiência.
source: Cadence: Extreme Pipelining with Multiple Concurrent Proposers
Generative art pieces from Lineage series, relevant to the operator's profile.
source: 2 framed pieces from Lineage series
提出了一种新的方法，通过递归证据回放来增强LLM在长上下文推理中的能力。
source: ReContext: Recursive Evidence Replay as LLM Harness for Long-Context Reasoning
Python framework for dynamic neural networks with GPU acceleration.
source: pytorch/pytorch
Powerful tool for building and deploying AI agents.
source: langflow-ai/langflow
Documentation for developing AI-powered workflows.
source: agentskills/agentskills
New testing framework for evaluating unlearning methods in LLMs.
source: LACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearning
Important release for online safety monitoring of LLMs.
source: Online Safety Monitoring for LLMs
English edition. Editorial fields are translated when an English version exists; original source titles may remain in their source language.