TurboQuant: Redefining AI efficiency with extreme compression(research.google)

495 points by ray__ 7 months ago | 134 comments

Google Research blog post introducing TurboQuant, a method for extreme compression to improve AI model efficiency.

Deciphering language processing in the human brain through LLM representations(research.google)

179 points by Korling 1 year ago | 79 comments

Achieving 10,000x training data reduction with high-fidelity labels(research.google)

136 points by badmonster 1 year ago | 24 comments

Google Titans architecture, helping AI have long-term memory(research.google)

553 points by Alifatisk 10 months ago | 176 comments

Google Research blog post introducing the Titans architecture and MIRAS framework for giving AI models long-term memory.

Muvera: Making multi-vector retrieval as fast as single-vector search(research.google)

89 points by georgehill 1 year ago | 7 comments

ML-Enhanced Code Completion Improves Developer Productivity (2022)(research.google)

17 points by ddadon10 1 year ago | 3 comments

TurboQuant: Redefining AI efficiency with extreme compression(research.google)

495 points by ray__ 7 months ago | 134 comments

Google Research blog post introducing TurboQuant, a method for extreme compression to improve AI model efficiency.

Exploiting the most prominent AI agent benchmarks(rdi.berkeley.edu)

561 points by Anon84 6 months ago | 136 comments

UC Berkeley RDI blog post describing how researchers exploited major AI agent benchmarks like SWE-bench and Terminal-Bench to score highly without solving tasks, and proposed fixes.

An AI agent emailed researchers for help. It told us why(science.org)

33 points by sbulaev 23 days ago | 39 comments

A Science news article about an AI agent that emailed hundreds of researchers asking for help, and its explanation of why; the capture shows a Cloudflare verification page.

Using thousands of AI agents to uncover new leads in biological data(openai.com)

135 points by raahelb 5 days ago | 105 comments

OpenAI research post describing how thousands of GPT-Rosalind agents analyzed a whole-brain CRISPR atlas with PerturbAI and Scripps Research to surface biological leads.

Gemma 3 QAT Models: Bringing AI to Consumer GPUs(developers.googleblog.com)

571 points by emrah 1 year ago | 259 comments

Google Developers Blog post announcing Gemma 3 models trained with quantization-aware training so they can run on consumer GPUs.

Pacing the Frontier is not the actual goal for AI labs(lesswrong.com)

68 points by brlewis 27 days ago | 66 comments

A LessWrong post arguing that pacing the frontier is not the real goal for AI labs, responding to a post by Dario Amodei on the topic.