DachengLi1 / AMP
(NeurIPS 2022) Automatically finding good model-parallel strategies, especially for complex models and clusters.
☆38Updated 2 years ago
Alternatives and similar repositories for AMP:
Users that are interested in AMP are comparing it to the libraries listed below
- ☆72Updated 3 years ago
- Automated Parallelization System and Infrastructure for Multiple Ecosystems☆78Updated 4 months ago
- nnScaler: Compiling DNN models for Parallel Training☆103Updated last month
- ☆77Updated 2 years ago
- [ICML 2024] Serving LLMs on heterogeneous decentralized clusters.☆23Updated 10 months ago
- A resilient distributed training framework☆91Updated 11 months ago
- An experimental parallel training platform☆54Updated last year
- FTPipe and related pipeline model parallelism research.☆41Updated last year
- Supplemental materials for The ASPLOS 2025 / EuroSys 2025 Contest on Intra-Operator Parallelism for Distributed Deep Learning☆23Updated 3 months ago
- Chimera: bidirectional pipeline parallelism for efficiently training large-scale models.☆61Updated last week
- ☆55Updated 9 months ago
- A ChatGPT(GPT-3.5) & GPT-4 Workload Trace to Optimize LLM Serving Systems☆156Updated 5 months ago
- FlexFlow Serve: Low-Latency, High-Performance LLM Serving☆32Updated this week
- ☆81Updated 3 years ago
- ☆24Updated last year
- AlpaServe: Statistical Multiplexing with Model Parallelism for Deep Learning Serving (OSDI 23)☆82Updated last year
- DISB is a new DNN inference serving benchmark with diverse workloads and models, as well as real-world traces.☆53Updated 7 months ago
- Stateful LLM Serving☆50Updated 3 weeks ago
- An Efficient Pipelined Data Parallel Approach for Training Large Model☆74Updated 4 years ago
- An extention of TVMScript to write simple and high performance GPU kernels with tensorcore.☆51Updated 8 months ago
- ☆136Updated 8 months ago
- A Python library transfers PyTorch tensors between CPU and NVMe☆111Updated 4 months ago
- Ultra | Ultimate | Unified CCL☆54Updated last month
- Microsoft Collective Communication Library☆60Updated 4 months ago
- Official repository for the paper DynaPipe: Optimizing Multi-task Training through Dynamic Pipelines☆17Updated last year
- ☆102Updated 7 months ago
- TileFusion is an experimental C++ macro kernel template library that elevates the abstraction level in CUDA C for tile processing.☆74Updated this week
- [NeurIPS 2024] Efficient LLM Scheduling by Learning to Rank☆43Updated 4 months ago
- ☆79Updated 4 months ago
- [IJCAI2023] An automated parallel training system that combines the advantages from both data and model parallelism. If you have any inte…☆51Updated last year