Skip to content

O
OneFlow-Benchmark

Oneflow-Inc / OneFlow-Benchmark
上一次同步 2 年多

通知 1
Star 92
Fork 0
O
OneFlow-Benchmark

体验新版 GitCode，发现更多精彩内容 >>

提交 04a16945 编写于 作者: S ShawnXuan
浏览文件
操作

modify for local test

上级 730ca874
隐藏空白更改
内联 并排
Showing with 4 addition and 4 deletion
of_e2e.sh
浏览文件 @ 04a16945
rm -rf core.* rm -rf core.*
rm -rf ./output/snapshots/* rm -rf ./output/snapshots/*
#DATA_ROOT=/DATA/disk1/of_imagenet_example #DATA_ROOT=/DATA/disk1/of_imagenet_example
DATA_ROOT=/DATA/disk1/ImageNet/ofrecord #DATA_ROOT=/DATA/disk1/ImageNet/ofrecord
#DATA_ROOT=/dataset/ImageNet/ofrecord DATA_ROOT=/dataset/ImageNet/ofrecord
#DATA_ROOT=/dataset/imagenet-mxnet #DATA_ROOT=/dataset/imagenet-mxnet
#python3 cnn_benchmark/of_cnn_train_val.py \ #python3 cnn_benchmark/of_cnn_train_val.py \
#nvprof -f -o resnet.nvvp \ #nvprof -f -o resnet.nvvp \
...@@ -14,11 +14,11 @@ DATA_ROOT=/DATA/disk1/ImageNet/ofrecord ...@@ -14,11 +14,11 @@ DATA_ROOT=/DATA/disk1/ImageNet/ofrecord
--val_data_part_num=256 \ --val_data_part_num=256 \
--num_nodes=1 \ --num_nodes=1 \
--node_ips='11.11.1.12,11.11.1.14' \ --node_ips='11.11.1.12,11.11.1.14' \
--gpu_num_per_node=8 \ --gpu_num_per_node=4 \
--optimizer="momentum-cosine-decay" \ --optimizer="momentum-cosine-decay" \
--learning_rate=0.256 \ --learning_rate=0.256 \
--loss_print_every_n_iter=20 \ --loss_print_every_n_iter=20 \
--batch_size_per_device=64 \ --batch_size_per_device=56 \
--val_batch_size_per_device=125 \ --val_batch_size_per_device=125 \
--use_boxing_v2=True \ --use_boxing_v2=True \
--use_new_dataloader=True \ --use_new_dataloader=True \
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或