Our approach to data and AI
Browse
AI News
30934 items — filtered, classified, deduplicated
Understanding the source of what we see and hear online
DeepSeek-V2 beats Mixtral 8x22B with >160 experts at HALF the cost
API Partnership with Stack Overflow
Introducing the Open Leaderboard for Hebrew LLMs!
$100k to predict LMSYS human preferences in a Kaggle contest
Bringing the Artificial Analysis LLM Performance Leaderboard to Hugging Face
Evals: The Next Generation
Not much happened today
LLMs-as-Juries
Operation "Doppelganger": Russian influence activity targeting Ukraine
Operation "Zero Zeno": Israel-linked influence activity
IUVM: Iran-linked influence content network
"Bad Grammar": Russian-linked Telegram comment activity
Powerful ASR + diarization + speculative decoding with Hugging Face Inference Endpoints
Operation "Spamouflage": China-linked influence activity
Improving Prompt Consistency with Structured Generations
A quiet weekend
We’re bringing the Financial Times’ world-class journalism to ChatGPT
StarCoder2-Instruct: Fully Transparent and Permissive Self-Alignment for Code Generation
Apple's OpenELM beats OLMo with 50% of its dataset, using DeLighT
Snowflake Arctic: Fully Open 10B+128x4B Dense-MoE Hybrid LLM
OpenAI's Instruction Hierarchy for the LLM OS
Introducing ChatGPT and Whisper APIs
GPT-4 API general availability and deprecation of older models in the Completions API
Perplexity, the newest AI unicorn
FineWeb: 15T Tokens, 12 years of CommonCrawl (deduped and filtered, you're welcome)
OpenAI’s commitment to child safety: adopting safety by design principles
Introducing the Open Chain of Thought Leaderboard
Introducing more enterprise-grade features for API customers