微软研究院博客

Advances in run-time strategies for next-generation foundation models

2024年11月27日 | Eric Horvitz, Harsha Nori, 和 Naoto Usuyama

Discover the most effective run-time strategies on the OpenAI o1-preview model, improving accuracy in medical language tasks.

A visual illustration of Medprompt performance on the MedQA benchmark. Moving from left to right on a horizontal line, the illustration shows how different Medprompt components and additive contributions improve accuracy starting with zero-shot at 81.7 accuracy, to random few-shot at 83.9 accuracy, to random few-shot, chain-of-thought at 87.3 accuracy, to kNN, few-shot, chain-of-thought at 88.4 accuracy, to ensemble with choice shuffle at 90.2 accuracy.

最新文章

按研究领域筛选

Advances in run-time strategies for next-generation foundation models

2024年11月27日 | Eric Horvitz, Harsha Nori, 和 Naoto Usuyama

Discover the most effective run-time strategies on the OpenAI o1-preview model, improving accuracy in medical language tasks.
Accelerating drug discovery with TamGen: A generative AI approach to target-aware molecule generation

2024年11月25日

TamGen uses generative AI to design new drug candidate compounds to treat TB, going beyond traditional methods by generating novel chemical structures. Learn how a collaboration with the Global Health Drug Discovery Institute is making this possible.
LazyGraphRAG: Setting a new standard for quality and cost

2024年11月25日 | Darren Edge, Ha Trinh, 和 Jonathan Larson

Introducing a new approach to graph-enabled RAG. LazyGraphRAG needs no prior summarization of source data, avoiding prohibitive up-front indexing costs. It’s inherently scalable in cost and quality across multiple methods and search mechanisms.
Introducing Yasuyuki Matsushita: Tackling societal challenges with AI at Microsoft Research Asia – Tokyo

2024年11月18日

Yasuyuki Matsushita rejoins Microsoft to lead the new Microsoft Research Asia - Tokyo lab. Learn more about his journey and his perspective on the Tokyo lab's role in evolution of AI.
$A stylized illustration of a green line-drawn hand holding a transparent prism with colorful bands of light being refracted through it against a black background.$

BiomedParse: A foundation model for smarter, all-in-one biomedical image analysis

2024年11月18日

BiomedParse reimagines medical image analysis, integrating advanced AI to capture complex insights across imaging types—a step forward for diagnostics and precision medicine.
GraphRAG: Improving global search via dynamic community selection

2024年11月15日 | Bryan Li, Ha Trinh, Darren Edge, 和 Jonathan Larson

Retrieval-augmented generation (RAG) helps AI systems provide more information to a large language model (LLM) when generating responses to user queries. A new method for conducting “global” queries can optimize the performance of global search in GraphRAG.
Orca-AgentInstruct: Agentic flows can be effective synthetic-data generators

2024年11月14日 | Arindam Mitra, Ahmed Awadallah, 和 Yash Lara

Orca-AgentInstruct, from Microsoft Research, can generate diverse, high-quality synthetic data at scale to post-train and fine-tune base LLMs for expanded capabilities, continual learning, and increased performance.
Toward modular models: Collaborative AI development enables model accountability and continuous learning

2024年11月13日 | Lucas Caccia, Darya Moldavskaya, 和 Alessandro Sordoni

Modular models can democratize AI development while unlocking new benefits and use cases. Modularized AI can be more flexible, more compliant, and cheaper to develop—requiring less data and fewer compute resources to train expert models.
Research Focus: Week of November 11, 2024

2024年11月13日

Holistic motion-capture calibration technique without calibration, manual intervention or custom hardware; Research on AI agents for autonomous clouds; Automating proof-oriented program construction; One-to-many testing for natural language code generation.
Preventing side-channels in the cloud

2024年11月12日 | Stavros Volos 和 Boris Köpf

Sophisticated side-channel attacks present new security challenges for cloud providers. Learn how Microsoft is exploring defenses against emerging attacks with principled microarchitectural isolation:
From static prediction to dynamic characterization: AI2BMD advances protein dynamics with ab initio accuracy

2024年11月6日

Microsoft Research’s AI2BMD, an AI-based system that efficiently simulates a wide range of proteins in all-atom resolution, can advance drug discovery and biomolecular research.
Microsoft at SOSP 2024: Innovations in systems research

2024年11月4日

Building resilient systems, scaling deep learning computation, and reproducing failures in production are just some of the ways Microsoft researchers are advancing the state of the art in computer systems research at SOSP 2024.