Pinned Loading

1. mcrl/spipe mcrl/spipe Public

   Hybrid GPU and CPU Pipeline for Training LLMs under Memory Pressure (PACT 2025)

   Python

2. mcrl/tccl mcrl/tccl Public

   Thunder Research Group's Collective Communication Library

   C++ 47 7

3. SCEC2023-TeamH SCEC2023-TeamH Public

   Forked from mcrl/SCEC2023-TeamH

   Fastest inference of HellaSwag with LLaMA 30B on single machine with four NVIDIA V100 GPUs (Samsung SCEC'23 1st place)

   Python

4. deepshare deepshare Public

   Network Contention-Aware Cluster Scheduling with Reinforcement Learning (IEEE ICPADS'23)

   C 19 3

5. cuda-advanced cuda-advanced Public

   Advanced implementations of CUDA GEMM

   Cuda 1

6. fastgen fastgen Public

   A Fast and Scalable Generative Model Inference on Distributed Multi-GPU Environment (KCC 2023)

   Cuda 1 1