Skip to content

Latest commit

 

History

History
65 lines (44 loc) · 2.47 KB

README_tflite.md

File metadata and controls

65 lines (44 loc) · 2.47 KB
Raw

[ Back to index ]

Running a mobilenet model using TFLite backend

Please follow this guide to run the entire set of MobileNet and EfficientNet models trained using ImageNet dataset (total 81 models) using TFLite backend.

If you want to try an individual model run, you can proceed as follows:

Run Command

cm run script --tags=run,mlperf,inference,run-mlperf,_submission,_short  \
   --adr.python.version_min=3.8 \
   --implementation=tflite-cpp \
   --model=efficientnet \
   --backend=tflite \
   --device=cpu \
   --scenario=SingleStream \
   --test_query_count=100 \
   --adr.tflite-model.tags=_lite0 \
   --adr.mlperf-inference-implementation.tags=_armnn,_use-neon \
   --clean

Run ResNet50 TFLite via CM

Do a test run to detect and record the system performance

cm run script --tags=run-mlperf,inference,_find-performance,_all-scenarios \
--model=resnet50 --implementation=tflite-cpp --device=cpu --backend=tflite \
--category=edge --division=open --quiet
  • Use --division=closed to run all scenarios for the closed division (compliance tests are skipped for _find-performance mode)
  • Since only singlestream is implemented for tflite-cpp, datacenter submission is not possible
  • Use --adr.mlperf-inference-implementation.tags=_armnn,_use-neon to use ARMNN backend

Do full accuracy and performance runs for all the scenarios

cm run script --tags=run-mlperf,inference,_submission,_all-scenarios --model=resnet50 \
--device=cpu --implementation=tflite-cpp --backend=tflite \
--execution-mode=valid --category=edge --division=open --quiet
  • Use --power=yes for measuring power. It is ignored for accuracy and compliance runs
  • Use --division=closed to run all scenarios for the closed division including the compliance tests
  • --offline_target_qps, --server_target_qps, --singlestream_target_latency and multistream_target_latency can be used to override the determined performance numbers
  • Use --adr.mlperf-inference-implementation.tags=_armnn,_use-neon to use ARMNN backend

Generate and upload MLPerf submission

Follow this guide to generate the submission tree and upload your results.

Questions? Suggestions?

Check the MLCommons Task Force on Automation and Reproducibility and get in touch via public Discord server.