01
vanguarda · score 9
AI Penetration Testing
A fully autonomous AI system for complex penetration testing tasks.
fonte: vxcontrol/pentagi
Caderno público · Scout
AI Penetration Testing
A fully autonomous AI system for complex penetration testing tasks.
itens
10
vanguarda
5
interessante
5
data
07-10
02
vanguarda · score 9
Video Reasoning Models
Video generation models that offer a reasoning path distinct from previous Chain-of-Thought.
fonte: OpenCoF: Learning to Reason Through Video Generation
03
vanguarda · score 9
Watermark para LLMs
Propõe um sistema de atribuição robusta para LLMs, usando logs de trajetória.
fonte: TRACE: A Two-Channel Robust Attribution Watermark via Complementary Embeddings for LLM-Agent Trajectories
04
vanguarda · score 9
Computação em Redes Dinâmicas
Deterministic computation in anonymous dynamic networks with one-bit communications.
fonte: Computing in Anonymous Dynamic Networks with One-Bit Communications
05
vanguarda · score 9
Agente proativo de memória para tarefas a longo prazo
Desenvolvimento de um agente de memória que auxilia em tarefas a longo prazo, uma abordagem inovadora.
fonte: Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
06
interessante · score 8
LLM Workflows Semantics
A conceptual model for semantic persistence in LLM-mediated workflows.
fonte: AUTOPILOT VQA: Benchmarking Vision-Language Models for Incident-Centric Dashcam Understanding
07
interessante · score 8
Workflow Conceptual Model
A conceptual model for semantic persistence in LLM-mediated workflows.
fonte: AUTOPILOT VQA: Benchmarking Vision-Language Models for Incident-Centric Dashcam Understanding
08
interessante · score 7
VQA for Autonomous Driving
Benchmarking Vision-Language Models for understanding dashcam footage.
fonte: Using AI-based Learning Assistants in Higher Education: A Large-Scale Descriptive Analysis
09
interessante · score 7
Low-Rank Regularization
A simple and efficient method for in-training low-rank regularization in neural networks.
fonte: SLORR: Simple and Efficient In-Training Low-Rank Regularization
10
interessante · score 7
Seleção Adaptativa de Linhas
Uma técnica adaptativa de seleção de linhas para o método Kaczmarz, que pode melhorar a eficiência em máquinas compartilhadas.
fonte: Adaptive Row Selection Meets Asynchrony in Randomized Kaczmarz