Shadowgraph
About
Sign in

MaverickJune/pMoE

Sub token level load balancing of MoE models

0

stars

2

commits

Python

primary language

Feb 20, 2025

updated

README

pMoE

Predictable MoE that fully balanced the expert workloads

Table of Contents

Installation

Provide instructions on how to install the project. For example:

# Clone the repository
git https://github.com/MaverickJune/pMoE.git

# Navigate to the project directory
cd fastmoe

# Install dependencies
python setup.py install

Usage

Replace your ffn layers in the mode to pMoE layers

License

This project is licensed under the MIT License.

Contributors

MaverickJune

1 commits

sanghyun-snu

1 commits

Similar repos

UNITES-Lab/MoE-Quantization

Official code for the paper "Examining Post-Training Quantization for Mixture-of-Experts: A…

31

inclusionAI/MoBE

Mixture-of-Basis-Experts for Compressing MoE-based LLMs

38

futureke/MTFSLaMM

0

CASE-Lab-UMD/Unified-MoE-Compression

The official implementation of the paper "Towards Efficient Mixture of Experts: A Holistic Study of…

91

tyukei/llm-mojo

0

Modalities/modalities

Modalities, a PyTorch-native framework for distributed and reproducible foundation model training.

111

EfficientMoE/MoE-Infinity

PyTorch library for cost-effective, fast and easy serving of MoE models.

356

VITA-Group/READ-ME

[NeurIPS2024] "Read-ME: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System…

16

Languages

Python

81.3%

C++

10.0%

Cuda

3.9%

Shell

3.6%

C

1.1%

MaverickJune/pMoE

Sub token level load balancing of MoE models

0

stars

2

commits

Python

primary language

Feb 20, 2025

updated

README

pMoE

Predictable MoE that fully balanced the expert workloads

Table of Contents

Installation

Provide instructions on how to install the project. For example:

# Clone the repository
git https://github.com/MaverickJune/pMoE.git

# Navigate to the project directory
cd fastmoe

# Install dependencies
python setup.py install

Usage

Replace your ffn layers in the mode to pMoE layers

License

This project is licensed under the MIT License.

Similar repos

UNITES-Lab/MoE-Quantization

Official code for the paper "Examining Post-Training Quantization for Mixture-of-Experts: A…

31

inclusionAI/MoBE

Mixture-of-Basis-Experts for Compressing MoE-based LLMs

38

futureke/MTFSLaMM

0

CASE-Lab-UMD/Unified-MoE-Compression

The official implementation of the paper "Towards Efficient Mixture of Experts: A Holistic Study of…

91

tyukei/llm-mojo

0

Modalities/modalities

Modalities, a PyTorch-native framework for distributed and reproducible foundation model training.

111

EfficientMoE/MoE-Infinity

PyTorch library for cost-effective, fast and easy serving of MoE models.

356

VITA-Group/READ-ME

[NeurIPS2024] "Read-ME: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System…

16

Contributors

MaverickJune

1 commits

sanghyun-snu

1 commits

Languages

Python

81.3%

C++

10.0%

Cuda

3.9%

Shell

3.6%

C

1.1%