awslabs / Lancet-Accelerating-MoE-Training-via-Whole-Graph-Computation-Communication-OverlappingLinks
Official implementation for the paper Lancet: Accelerating Mixture-of-Experts Training via Whole Graph Computation-Communication Overlapping, published in MLSys'24.
☆14Updated 11 months ago
Alternatives and similar repositories for Lancet-Accelerating-MoE-Training-via-Whole-Graph-Computation-Communication-Overlapping
Users that are interested in Lancet-Accelerating-MoE-Training-via-Whole-Graph-Computation-Communication-Overlapping are comparing it to the libraries listed below
Sorting:
- ☆81Updated 2 years ago
- LLM serving cluster simulator☆108Updated last year
- Compiler for Dynamic Neural Networks☆46Updated last year
- ☆55Updated 3 months ago
- nnScaler: Compiling DNN models for Parallel Training☆118Updated last week
- Magicube is a high-performance library for quantized sparse matrix operations (SpMM and SDDMM) of deep learning on Tensor Cores.☆89Updated 2 years ago
- MAGIS: Memory Optimization via Coordinated Graph Transformation and Scheduling for DNN (ASPLOS'24)☆53Updated last year
- A lightweight design for computation-communication overlap.☆161Updated this week
- ☆150Updated last year
- Tile-based language built for AI computation across all scales☆48Updated this week
- High performance Transformer implementation in C++.☆132Updated 7 months ago
- Artifacts of EVT ASPLOS'24☆26Updated last year
- ☆106Updated last year
- REEF is a GPU-accelerated DNN inference serving system that enables instant kernel preemption and biased concurrent execution in GPU sche…☆100Updated 2 years ago
- ☆42Updated last year
- ☆84Updated 5 months ago
- Chimera: bidirectional pipeline parallelism for efficiently training large-scale models.☆66Updated 5 months ago
- Proteus: A High-Throughput Inference-Serving System with Accuracy Scaling☆13Updated last year
- ☆28Updated last year
- TACCL: Guiding Collective Algorithm Synthesis using Communication Sketches☆75Updated 2 years ago
- Synthesizer for optimal collective communication algorithms☆116Updated last year
- A GPU-optimized system for efficient long-context LLMs decoding with low-bit KV cache.☆58Updated last week
- An interference-aware scheduler for fine-grained GPU sharing☆145Updated 7 months ago
- Open-source implementation for "Helix: Serving Large Language Models over Heterogeneous GPUs and Network via Max-Flow"☆63Updated 9 months ago
- ASPLOS'24: Optimal Kernel Orchestration for Tensor Programs with Korch☆38Updated 5 months ago
- Paella: Low-latency Model Serving with Virtualized GPU Scheduling☆62Updated last year
- Thunder Research Group's Collective Communication Library☆41Updated last month
- ☆41Updated last year
- Official repository for the paper DynaPipe: Optimizing Multi-task Training through Dynamic Pipelines☆20Updated last year
- ☆51Updated 2 months ago