1.deepspeedai / DeepSpeedDeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.deep-learningpytorchgpumachine-learningbillion-parametersdata-parallelism+7 morePythonAPACHE-2.043K total5.1K forks+152 starsfrom 43,129 · score 5.03
2.bigscience-workshop / petals🌸 Run LLMs at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloadingbloomdeep-learningdistributed-systemslanguage-modelslarge-language-modelsmachine-learning+14 morePythonMIT11K total650 forks+83 starsfrom 10,596 · score 4.43
3.hpcaitech / ColossalAIMaking large AI models cheaper, faster and more accessibledeep-learninghpclarge-scaledata-parallelismpipeline-parallelismmodel-parallelism+6 morePythonAPACHE-2.041K total4.5K forks+32 starsfrom 41,401 · score 3.50