MMLU ProMMLU-Pro is an open-source benchmark for evaluating large language models on challenging reasoning tasks across 14+ academic domains. Presented at NeurIPS 2024.(0)0
M Marian NMTMarian NMT is an efficient, open-source neural machine translation framework written in pure C++. Supports Transformer and RNN architectures with multi-GPU training, developed by Microsoft Translator.(0)0
FairseqFairseq is Facebook AI Research's open-source Python toolkit for training custom sequence models for machine translation, summarization, language modeling, and more.(0)0
Meta ESMOpen-source pretrained language models for protein sequences. Enables protein structure prediction, mutational effect analysis, and generative protein design using transformer-based architectures.(0)0
EleutherAI lm-evalA flexible, open-source framework for few-shot benchmarking and evaluation of large language models across hundreds of tasks and multiple inference backends.(0)0
FastChatFastChat is an open-source platform for training, serving, and evaluating large language models. Powers Chatbot Arena with OpenAI-compatible APIs and a distributed multi-model serving system.(0)0
ONNX RuntimeONNX Runtime is Microsoft's open-source AI engine for accelerated machine learning inference and training across cloud, edge, mobile, and web platforms.(0)0
MT BenchMT Bench is an open-source multi-turn benchmark for evaluating large language models using GPT-4 as an automated judge. Part of the FastChat ecosystem by lm-sys.(0)0
MatterSimMatterSim is Microsoft's open-source deep learning atomistic model for simulating material properties across elements, temperatures, and pressures using M3GNet architecture.(0)0
Ray AI FrameworkRay is an open source Python-native framework for scaling and orchestrating distributed AI, ML, and GenAI workloads across CPUs and GPUs at any scale.(0)0
