refine resnet50v1.5

Classification/resnet50v1.5/train.sh

+export ENABLE_USER_OP=True
+rm -rf core.* 
+rm -rf ./output/snapshots/*
+
+DATA_ROOT=/dataset/ImageNet/ofrecord
+
+python3 of_cnn_train_val.py \
+    --train_data_dir=$DATA_ROOT/train \
+    --train_data_part_num=256 \
+    --val_data_dir=$DATA_ROOT/validation \
+    --val_data_part_num=256 \
+    --num_nodes=1 \
+    --gpu_num_per_node=4 \
+    --model_update="momentum" \
+    --learning_rate=0.256 \
+    --loss_print_every_n_iter=20 \
+    --batch_size_per_device=64 \
+    --val_batch_size_per_device=125 \
+    --num_epoch=90 \
+    --model="resnet50" 
\ No newline at end of file

run_cnn_train.sh

-export ENABLE_USER_OP=True
-rm -rf core.* 
-rm -rf ./output/snapshots/*
-
-# DATA_ROOT=/dataset/ImageNet/ofrecord
-
-python3 cnn_benchmark/of_cnn_train_val.py \
-    --train_data_part_num=256 \
-    --val_data_part_num=256 \
-    --num_nodes=1 \
-    --gpu_num_per_node=1 \
-    --model_update="momentum" \
-    --mom=0.875 \
-    --learning_rate=0.256 \
-    --loss_print_every_n_iter=1 \
-    --batch_size_per_device=8 \
-    --val_batch_size_per_device=8 \
-    --num_epoch=90 \
-    --use_fp16=false \
-    --use_boxing_v2=true \
-    --model="resnet50" 
-
-    # --node_ips='11.11.1.12,11.11.1.14' \
-    # --model_load_dir=/DATA/disk1/mx_imagenet/of_mx_mixed_init_model \
-    # --train_data_dir=$DATA_ROOT/train \
-    # --train_data_part_num=256 \
-    # --val_data_dir=$DATA_ROOT/validation \
-    # --val_data_part_num=256 \
-    #--num_examples=1024 \