StreamingLLM: tiny tweak to KV LRU improves long conversations (news.mit.edu)

Screenshot of news.mit.edu: StreamingLLM: tiny tweak to KV LRU improves long conversations

First screen, captured 2 years ago.

Looks similar

Screenshot of arxiv.org: A Multimodal Automated Interpretability Agent

A Multimodal Automated Interpretability Agent (arxiv.org)

64 points by el_duderino 2 years ago | 7 comments

Screenshot of news.mit.edu: Engineering household robots to have a little common sense

Engineering household robots to have a little common sense (news.mit.edu)

24 points by CharlesW 2 years ago | 8 comments

Screenshot of news.mit.edu: Arvind has died

Arvind has died (news.mit.edu)

231 points by bookofjoe 2 years ago | 51 comments

Screenshot of news.mit.edu: Engineers find a new way to convert carbon dioxide into useful products

Engineers find a new way to convert carbon dioxide into useful products (news.mit.edu)

94 points by gmays 2 years ago | 75 comments

Screenshot of news.mit.edu: LLMs use a surprisingly simple mechanism to retrieve some stored knowledge

LLMs use a surprisingly simple mechanism to retrieve some stored knowledge (news.mit.edu)

336 points by CharlesW 2 years ago | 121 comments

MIT News article on research showing LLMs often retrieve stored facts via simple linear functions, and a technique to probe what a model knows about subjects.

large-language-models interpretability knowledge-retrieval mit

Screenshot of news.mit.edu: How an MIT research project became the Julia programming language

How an MIT research project became the Julia programming language (news.mit.edu)

200 points by theanonymousone 1 month ago | 101 comments

Looks similar

A Multimodal Automated Interpretability Agent(arxiv.org)

Engineering household robots to have a little common sense(news.mit.edu)

Arvind has died(news.mit.edu)

Engineers find a new way to convert carbon dioxide into useful products(news.mit.edu)

LLMs use a surprisingly simple mechanism to retrieve some stored knowledge(news.mit.edu)

How an MIT research project became the Julia programming language(news.mit.edu)