Fixes for mlperf submission (#249)

* Fix pythron version for bert deepsparse

Author: arjunsuresh

script/get-onnxruntime-prebuilt/meta.yaml

@@ -4,7 +4,7 @@ automation_uid: 5b4e0237da074764
 cache: true
 category: AI/ML frameworks
 clean_files: []
-default_version: 1.16.3
+default_version: 1.20.1
 deps:
 - tags: detect,os
 new_env_keys:

script/run-all-mlperf-models/run-cpp-implementation.sh

@@ -27,9 +27,9 @@ division="closed"
 # run "$MLC_RUN_CMD"
 
 POWER=" --power=yes --adr.mlperf-power-client.power_server=192.168.0.15 --adr.mlperf-power-client.port=4950 "
-POWER=""
+POWER=" --env.MLC_GET_PLATFORM_DETAILS=no"
 
-run "mlcr set,system,performance,mode"
+#run "mlcr set,system,performance,mode"
 
 #cpp
 run "mlcr generate-run-cmds,inference,_find-performance \
@@ -51,7 +51,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 run "mlcr generate-run-cmds,inference,_submission \
 --model=retinanet --implementation=cpp --device=cpu --backend=onnxruntime \
@@ -61,7 +61,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 run "mlcr generate-run-cmds,inference,_submission \
 --model=resnet50 --implementation=cpp --device=cpu --backend=onnxruntime \
@@ -71,7 +71,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 run "mlcr generate-run-cmds,inference,_submission \
 --model=retinanet --implementation=cpp --device=cpu --backend=onnxruntime \
@@ -81,7 +81,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 # GPU
 
@@ -106,7 +106,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 run "mlcr generate-run-cmds,inference,_submission \
 --model=retinanet --implementation=cpp --device=cuda --backend=onnxruntime \
@@ -116,7 +116,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 
 run "mlcr generate-run-cmds,inference,_submission \
@@ -128,7 +128,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 run "mlcr generate-run-cmds,inference,_submission \
 --model=retinanet --implementation=cpp --device=cuda --backend=onnxruntime \
@@ -138,7 +138,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 #multistream
 run "mlcr generate-run-cmds,inference,_submission \
@@ -150,7 +150,7 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"
 
 run "mlcr generate-run-cmds,inference,_submission \
 --model=retinanet --implementation=cpp --device=cuda --backend=onnxruntime \
@@ -160,4 +160,4 @@ run "mlcr generate-run-cmds,inference,_submission \
 --execution_mode=valid \
 --skip_submission_generation=yes \
 ${POWER} \
---results_dir=$HOME/results_dir"
+"

script/run-all-mlperf-models/run-pruned-bert.sh

@@ -46,7 +46,7 @@ scenario="Offline"
 if [[ $scenario == "Offline" ]]; then
 for stub in ${zoo_stub_list[@]}; do
 cmd="mlcr run,mlperf,inference,generate-run-cmds,_find-performance  \
-   --adr.python.version_min=3.8 \
+   --adr.python.version=3.9.12 \
    --implementation=reference \
    --model=bert-99 \
    --precision=int8 \
@@ -65,7 +65,7 @@ fi
 
 for stub in ${zoo_stub_list[@]}; do
  cmd="mlcr run,mlperf,inference,generate-run-cmds  \
-   --adr.python.version_min=3.8 \
+   --adr.python.version=3.9.12 \
    --adr.compiler.tags=gcc \
    --implementation=reference \
    --model=bert-99 \