add resnet_bs64_fp16 for model_benchmark_ci (#1572)

xiegegege · web-flow · commit 75cfd4f8fb5d · 2023-01-17T15:43:19.000+08:00
* add resnet_bs64_fp16 for model_benchmark_ci,test=document_fix

* add resnet_bs64_fp16 for model_benchmark_ci,test=document_fix
diff --git a/scripts/benchmark_ci/model_ci.sh b/scripts/benchmark_ci/model_ci.sh
@@ -15,7 +15,7 @@ export FLAGS_call_stack_level=2
 #run models
 cd ${BENCHMARK_ROOT}/scripts/benchmark_ci
 #model_list='ResNet50_bs32_dygraph ResNet50_bs32 bert_base_seqlen128_fp32_bs32 transformer_base_bs4096_amp_fp16 yolov3_bs8 TSM_bs16 deeplabv3_bs4_fp32 CycleGAN_bs1 mask_rcnn_bs1 PPOCR_mobile_2_bs8 seq2seq_bs128'
-model_list='ResNet50_bs32_dygraph deeplabv3_bs4_fp32 bert_base_seqlen128_fp32_bs32 yolov3_bs8 ppyolov2_bs6'
+model_list='ResNet50_bs32_dygraph ResNet50_bs64_fp16_dygraph deeplabv3_bs4_fp32 bert_base_seqlen128_fp32_bs32 yolov3_bs8 ppyolov2_bs6'
 source run_models.sh
 for model in ${model_list}
 do
diff --git a/scripts/benchmark_ci/run_models.sh b/scripts/benchmark_ci/run_models.sh
@@ -22,7 +22,7 @@ ResNet50_bs64_fp16_dygraph(){
     cur_model_path=${BENCHMARK_ROOT}/PaddleClas
     cd ${cur_model_path}
     pip install -r requirements.txt
-    #install dali
+    # install dali
     pip install --extra-index-url https://developer.download.nvidia.com/compute/redist nvidia-dali-cuda110
     # Prepare data
     ln -s ${data_path}/imagenet100_data/ ${cur_model_path}/dataset
