Tag
9 articles
Meta AI has open-sourced GCM, a GPU cluster monitoring tool designed to improve performance and hardware reliability in AI training environments.
Alibaba's Qwen team introduces the Qwen 3.5 Medium Model Series, challenging the trend of ever-larger AI models by prioritizing efficiency and practical performance in production environments.
Philosopher Michael Pollan argues that despite advances in AI, true consciousness remains beyond artificial systems' reach. His new book <em>A World Appears</em> explores the fundamental differences between human awareness and machine intelligence.
OpenAI shares proof attempts from its AI model tackling expert-level mathematical problems in the First Proof challenge, showcasing advanced reasoning capabilities.
Zurich-based Rapidata has raised €7.2M to build a real-time human feedback network for AI, addressing the challenge of scaling human insight for machine learning systems.
A new Google AI research introduces the Deep-Thinking Ratio, a method to improve LLM accuracy while cutting inference costs by half. It challenges the traditional belief that longer reasoning chains lead to better outcomes.
OpenAI announces it will no longer evaluate SWE-bench Verified due to contamination and data leakage issues. The organization recommends SWE-bench Pro as a replacement.
Google's Cloud AI lead identifies three critical frontiers in AI development: raw intelligence, response time, and extensibility. These boundaries represent fundamental shifts in how AI systems will interact with human users and enterprise applications.
A new tutorial from MarkTechPost demonstrates how to use TruLens and OpenAI models to build transparent and measurable evaluation pipelines for LLM applications.
