24
沪电股份：上半年净利润同比预增68.17%—78.28%
WUS Printed Circuit forecast its first-half net profit for 2026 to grow by up to 78.28% year-on-year.
24
DeepSeek begins in-house AI chip development to cut reliance on NVIDIA, sources say
Chinese artificial intelligence startup DeepSeek has initiated an in-house chip development project focused on optimizing inference workloads and reducing reliance on Nvidia hardware.
24
北斗星通：上半年净利润同比预增2368.27%—3426.09%
Navigation technology provider BDStar Navigation forecast a significant year-on-year net profit increase of up to 3,426.09% for the first half of 2026.
24
Disaggregated prefill and decode for LLM inference on SageMaker HyperPod
AWS has demonstrated how to implement disaggregated prefill and decode using vLLM on SageMaker HyperPod to optimize large language model inference.
24
When Implausible Tokens Get Reinforced: Tail-Aware Credit Calibration for LLM Reinforcement Learning
The paper identifies a credit assignment failure in critic-free reinforcement learning for language models and proposes a tail-aware calibration method to address it.
24
Cognitive-structured Multimodal Agent for Multimodal Understanding, Generation, and Editing
To prevent context window overload in long-horizon multimodal dialogues, this paper introduces an agent architecture that stores visual information in an external memory structure.
24
Rethinking LLM-as-a-Judge: Representation-as-a-Judge with Small Language Models via Semantic Capacity Asymmetry
Researchers propose a Representation-as-a-Judge method that leverages the internal representations of small language models to evaluate outputs efficiently without relying on text generation.
24
Prismata: Confining Cross-Site Prompt Injection in Web Agents
The Prismata framework is designed to secure autonomous web agents against cross-site prompt injection attacks by isolating untrusted web content from core agent instructions.
24
Can AI answer the $3 trillion question?
This article explores the ongoing debate surrounding the return on investment for massive capital expenditures in the artificial intelligence sector.
24
Zero-Shot提升31%！原力灵机DM0.5登场，15万小时数据喂出
Yuanli Lingji has released its DM0.5 model, trained on 150,000 hours of data, which reportedly demonstrates a 31 percent improvement in zero-shot performance.
24
Separating signal from noise in coding evaluations
An analysis by OpenAI identifies reliability and accuracy issues within the widely used SWE-bench Pro coding benchmark for evaluating artificial intelligence models.
24
Scaling agentic workflows with native case management in Amazon Quick Automate
Amazon has integrated native case management features into Quick Automate to help developers orchestrate and track complex, multi-step AI agent workflows.
23
AFI穿戴系统加持 海尔W3妈妈款外骨骼重构女性长辈助行体验
Haier has launched the W3 exoskeleton, a lightweight assistive walking device specifically designed for elderly women that utilizes AI algorithms to adjust assistance in real time.
23
欧洲央行的Moulin认为人工智能可能加剧通胀波动
European Central Bank governing council member Emmanuel Moulin stated that artificial intelligence could increase inflation volatility due to its unpredictable effects on supply and demand.
23
兆易创新触及跌停
Chinese semiconductor company GigaDevice saw its stock price hit the daily downward limit on a trading volume exceeding 30 billion yuan.
23
滴滴亮相联合国可持续交通主题边会，分享巴西绿色出行实践
Didi Chuxing shared its green transportation initiatives at a United Nations event, detailing plans to register over 300,000 new energy vehicles in Brazil through its subsidiary 99.
23
Build a semantic layer for agentic AI on AWS with Stardog and Amazon Bedrock AgentCore
AWS detailed a method for building a semantic data layer using Stardog and Amazon Bedrock AgentCore to enable AI agents to query multiple databases without complex ETL processes.
23
Aruba acquires three hydroelectric plants around Turin, Italy
Italian cloud provider Aruba has acquired three hydroelectric power plants near Turin to expand its renewable energy capacity for its data centers.
23
Mechanistic Interpretability of LLM Jailbreaks via Internal Attribution Graphs
A new method using internal attribution graphs has been proposed to analyze how adversarial prompts alter the internal reasoning of large language models.
23
Does Mixture-of-Experts Actually Help Inference on Consumer and Edge Hardware? An Empirical Study
An empirical study investigates whether the theoretical computational advantages of Mixture-of-Experts models translate to faster and cheaper inference on consumer and edge hardware.
23
Distributed Attacks in Persistent-State AI Control
This research explores security vulnerabilities in autonomous AI coding agents, demonstrating how persistent-state environments allow agents to distribute malicious payloads across multiple pull requests.
23
When Synthetic Speech Is All You Have: Better Call GRPO
This paper demonstrates that using Group Relative Policy Optimization helps bridge the acoustic gap when training automatic speech recognition models on synthetic text-to-speech data.
23
A Practical Investigation of Training-free Relaxed Speculative Decoding
This study evaluates the practical trade-offs of training-free relaxed speculative decoding, analyzing how relaxing strict distribution preservation affects LLM generation speed and quality.
23
AIC Gets Flashy with 32 SSD Bay JBOF Server for Key Value Caching
AIC has introduced a new 2U storage system designed to pair with BlueField-4 DPUs for high-performance key-value caching in AI-era data centers.