This repository has been archived by the owner on Jan 26, 2021. It is now read-only.
Multiverso Caffe Benchmark
Junjie Li edited this page Aug 1, 2016
·
5 revisions
Note: This benchamrk is a work in progress, we will add more experiments and documents soon.
Perform CIFAR-10 classification with Alex full model.
Distributed Training For Caffe With Multiverso
Please follow this guide to setup your environment.
- Hosts : 1
- GPU : GeForce GTX TITAN X * 4
- CPU : Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
- Memory : 128GB
- batchSize per GPU: 100
- reduce learning rate after 120 epochs by factor 0f 10 then another factor of 10 after 10 more epochs
| Code Name | #Process(es) | #GPU(s) per Process | Use multiverso | Seconds per epoch | Best Model |
|---|---|---|---|---|---|
| 1P1G0M | 1 | 1 | 0 | -- | 81.37 % |
| 1P4G0M | 1 | 4 | 0 | -- | 81.21 % |
| 4P1G1M | 4 | 1 | 1 | -- | 80.93 % |
DMTK
Multiverso
- Overview
- Multiverso setup
- Multiverso document
- Multiverso API document
- Multiverso applications
- Logistic Regression
- Word Embedding
- LightLDA
- Deep Learning
- Multiverso binding
- Run in docker
LightGBM