An index of datasets, SDKs, APIs and other open source code created by Microsoft researchers and shared with the broader academic community. We also maintain a collection highlighting some of the tools you’ll find here.
RadFact is a framework for the evaluation of model-generated radiology reports given a ground-truth report, with or without grounding. Leveraging the logical inference capabilities of large language models, RadFact is not a single number but…
vAttention is a memory manager for KV-cache in LLM serving systems. It decouples the allocation of virtual memory and physical memory using the CUDA virtual memory APIs. This approach enables allocating physical memory on demand…
Trace is a new AutoDiff-like tool for training AI systems end-to-end with general feedback (like numerical rewards or losses, natural language text, compiler errors, etc.). Trace generalizes the back-propagation algorithm by capturing and propagating an…
LongRoPE is a novel method that extends the context window of pre-trained LLMs to an impressive 2048k tokens by non-uniformly rescaling RoPE positional embeddings. LongRoPE has been integrated into Microsoft Phi-3.
A public framework for time-series forecasting with spiking neural networks (SNNs).
The Intelligence Toolkit is a suite of interactive workflows for creating AI intelligence reports from real-world data sources. The toolkit is designed to help users identify patterns, answers, relationships, and risks within complex datasets, with…
BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment.
MetaOpt is the first general-purpose and scalable tool that enables users to analyze a broad class of heuristics through easy-to-use abstractions that apply to a broad range of practical heuristics. For more information, checkout MetaOpt’s project webpage and…
VisEval: A NL2VIS Benchmark. VisEval is a benchmark designed to evaluate visualization generation methods. In this repository, we provide both the toolkit to support the benchmarking, as well as the data used for benchmarks.
