[add]上传训练benchmark by z00560161
This commit is contained in:
@@ -0,0 +1,28 @@
|
||||
tensorflow_config:
|
||||
# 基本参数
|
||||
max_steps: 1000
|
||||
data_url: /home/imagenet_TF/
|
||||
epoches: 1
|
||||
epochs_between_evals: 1
|
||||
batch_size: 32
|
||||
|
||||
# 仅多机执行需要配置: ip1:卡数量1,ip2:卡数量2
|
||||
mpirun_ip: 90.90.176.152:8,90.90.176.154:8
|
||||
|
||||
# docker 镜像名称:版本号
|
||||
docker_image: mpirun3:latest
|
||||
|
||||
|
||||
# 1. 指定 device id, 多个 id 使用空格分隔, 数量需与 rank_size 相同
|
||||
# 2. 仅在小于 8p 时生效
|
||||
# 3. 若不使用该配置, 请使用在行首添加'#'注释的方法将其关闭
|
||||
# device_group: 0 1 2 3
|
||||
device_group_1p: 0
|
||||
device_group_2p: 0 1
|
||||
device_group_4p: 0 1 2 3
|
||||
|
||||
profiling_mode: false
|
||||
profiling_options: training_trace
|
||||
fp_point: fp32_vars/conv2d/Conv2Dfp32_vars/BatchNorm/FusedBatchNormV3_Reduce
|
||||
bp_point: loss_scale/gradients/AddN_70
|
||||
aicpu_profiling_mode: false
|
||||
Reference in New Issue
Block a user