insightface

ZF/insightface

Fork 0

mirror of https://github.com/deepinsight/insightface.git synced 2026-05-14 12:17:55 +00:00

Files

History

anxiang 834b7646f8 Update Folder.

2020-10-19 01:07:30 +08:00

evaluation

Update Folder.

2020-10-19 01:07:30 +08:00

hosts

Update Folder.

2020-10-19 01:07:30 +08:00

setup-utils

Update Folder.

2020-10-19 01:07:30 +08:00

symbol

Update Folder.

2020-10-19 01:07:30 +08:00

callbacks.py

Update Folder.

2020-10-19 01:07:30 +08:00

config.sh

Update Folder.

2020-10-19 01:07:30 +08:00

default.py

Update Folder.

2020-10-19 01:07:30 +08:00

image_iter.py

Update Folder.

2020-10-19 01:07:30 +08:00

memory_bank.py

Update Folder.

2020-10-19 01:07:30 +08:00

memory_module.py

Update Folder.

2020-10-19 01:07:30 +08:00

memory_samplers.py

Update Folder.

2020-10-19 01:07:30 +08:00

memory_scheduler.py

Update Folder.

2020-10-19 01:07:30 +08:00

memory_softmax.py

Update Folder.

2020-10-19 01:07:30 +08:00

optimizer.py

Update Folder.

2020-10-19 01:07:30 +08:00

README.md

Update Folder.

2020-10-19 01:07:30 +08:00

run.sh

Update Folder.

2020-10-19 01:07:30 +08:00

train_memory.py

Update Folder.

2020-10-19 01:07:30 +08:00

README.md

Train

Requirements

python==3.6
cuda==10.1
cudnn==765
mxnet-cu101==1.6.0.post0
pip install easydict mxboard opencv-python tqdm
nccl
openmpi==4.0.0
horovod==0.19.2

Failures due to SSH issues

The host where horovodrun is executed must be able to SSH to all other hosts without any prompts.

Run with horovodrun

Typically one GPU will be allocated per process, so if a server has 8 GPUs, you will run 8 processes. In horovodrun, the number of processes is specified with the -np flag.

To run on a machine with 8 GPUs:

horovodrun -np 8 -H localhost:8 bash config.sh

To run on two machine with 16 GPUs:

horovodrun -np 16 -H ip1:8,ip2:8 bash config.sh

Run with mpi

bash run.sh