Beyond a Single Extractor: Re-thinking HTML-to-Text Extraction for LLM Pretraining
One of the first pre-processing steps for constructing web-scale LLM pretraining datasets involves extracting text…
Recent Articles
View AllThe Potential of CoT for Reasoning: A Closer Look at Trace Dynamics
Chain-of-thought (CoT) prompting is a de-facto standard technique to elicit reasoning-like responses from large language…
AMUSE: Audio-Visual Benchmark and Alignment Framework for Agentic Multi-Speaker Understanding
Recent multimodal large language models (MLLMs) such as GPT-4o and Qwen3-Omni show strong perception but…
Agenda Reflection: Testing Automated Alignment
Published on February 23, 2026 9:53 PM GMTSharing a mostly inactive empirical agenda from mid…
With AI, investor loyalty is (almost) dead: At least a dozen OpenAI VCs now also back Anthropic
While some dual investors are understandable, others were more shocking, and signal the disregard of…
Metal Detector for Life
Published on February 23, 2026 9:18 PM GMTYou can't care for something you can't see.To…
Claude Sonnet 4.6 Gives You Flexibility
Published on February 23, 2026 8:30 PM GMTAnthropic first gave us Claude Opus 4.6, then…
Secrets of the LessWrong RSS Feed
Published on February 23, 2026 8:12 PM GMTLessWrong's RSS feed includes all recently published articles…
Abstract/Concrete Axis: Robustness Testing Semantic Feature Separation in Gemma 3 270M
Published on February 23, 2026 8:07 PM GMTIn my previous post, I found that abstract…
Anthropic accuses Chinese AI labs of mining Claude as US debates AI chip exports
Anthropic accuses DeepSeek, Moonshot, and MiniMax of using 24,000 fake accounts to distill Claude’s AI…