Subject
6 entries
Microsoft
Bookmarks
AgentRC: context engineering and AI-readiness for repos
AgentRC (formerly Primer) measures and generates AI-readiness for repos — scoring across 9 pillars, generating instruction files, and running CI drift detection so agent context stays current as code evolves.
monitors4codegen — Monitor-Guided Decoding
Monitor-Guided Decoding uses LSP (Language Server Protocol) static analysis as a 'monitor' during code LM generation to enforce semantic validity — identifiers must exist, types must match. NeurIPS 2023 paper from Microsoft with the multispy Python library for building LSP-backed code gen applications.
JARVIS / HuggingGPT: LLM as AI Model Orchestrator
Microsoft JARVIS (also published as HuggingGPT) uses ChatGPT as a task planner that routes subtasks to specialized Hugging Face models — an early demonstration that LLMs could orchestrate other AI models as tools. A prototype of the multi-model agent pattern.
Image as a Foreign Language: BEIT-3 Pretraining for All Vision and Vision-Language Tasks
Wang, Bao, Dong et al. at Microsoft introduce BEIT-3, a general-purpose multimodal foundation model that treats images as a 'foreign language' and applies masked language modeling uniformly across images, text, and image-text pairs. BEIT-3 achieves state-of-the-art across seven vision and vision-language benchmarks including COCO, ImageNet, VQA, and NLVR2.
ZeRO: Memory Optimizations Toward Training Trillion Parameter Models
ZeRO (Zero Redundancy Optimizer) eliminates memory redundancy in distributed training by partitioning optimizer states, gradients, and parameters across data-parallel processes rather than replicating them. It enables training models 8x larger than prior methods on the same hardware and forms the foundation of Microsoft's DeepSpeed library.
MLOps Maturity Models: Google and Microsoft Frameworks
ZenML's overview of MLOps maturity models from Google and Microsoft — frameworks for thinking about how ML organizations can systematically improve how they develop and deploy models. Useful if you're trying to level up a team's ML practices from ad-hoc to automated.
