Colour gradient
Microsoft Research Cambridge

Machine Intelligence

Advanced machine learning research, grounded in trust, efficiency, capability.

ニュース&特集

diagram
記事

Given a language model, can we tell whether it is truly reasoning, or if its performance owes only to pattern recognition and memorization?

KBLaM blog | A flowchart illustrating the process of handling a prompt using a language model. The process begins with documents being used to construct and summarize a knowledge base (KB) offline. The summarized KB is then encoded and fed into the main process. A prompt goes through a tokenizer, followed by rectangular attention, and then into the large language model (LLM). The LLM retrieves information from the encoded KB to generate an answer.
Microsoft Research ブログ

| Taketomo Isazawa, Xi Wang, Liana Mikaelyan, Mathew Salvaris, と James Hensman

Introducing KBLaM, an approach that encodes and stores structured knowledge within an LLM itself. By integrating knowledge without retraining, it offers a scalable alternative to traditional methods.

ニュース | Windows Experience Blog

Phi Silica, small but mighty on-device SLM 

Today we will share how the Applied Sciences team used a multi-interdisciplinary approach to achieve a breakthrough in power efficiency, inference speed and memory efficiency for a state-of-the-art small language model (SLM), Phi Silica.