Name		Name	Last commit message	Last commit date
.idea		.idea
Model		Model
__pycache__		__pycache__
evaluate		evaluate
DataflowGraph.txt		DataflowGraph.txt
GetDensityMap.py		GetDensityMap.py
README.md		README.md
benchmark_grpc_recv.py		benchmark_grpc_recv.py
distributed_train.py		distributed_train.py
distributed_train_horovod.py		distributed_train_horovod.py
distributed_train_tweaked.py		distributed_train_tweaked.py
gpu_profile		gpu_profile
horovod_timeline_1worker.json		horovod_timeline_1worker.json
kill_process.py		kill_process.py
refined_test.py		refined_test.py
refined_train.py		refined_train.py
run_horovod.sh		run_horovod.sh
script_move_result.sh		script_move_result.sh
scripts		scripts
show_density_map.py		show_density_map.py
show_predicted_map.py		show_predicted_map.py
stats.py		stats.py
testrun.py		testrun.py
training_naive.py		training_naive.py

Repository files navigation

README

DistributedCNN

This project aims at finding the better distributed training architecture between Parameter Server (Standard distributed TensorFlow) and Ring Allreduce (Horovod).