Major update with CM automation for the latest ABTF model, Cognata dataset and loadgen

.gitignore

@@ -0,0 +1,160 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/

COPYRIGHT.txt

@@ -0,0 +1 @@
+Copyright (c) 2024 MLCommons

README.md

@@ -0,0 +1,8 @@
+# Collective Mind interface and automation for ABTF
+
+This repository contains [CM scripts (cross-platform automation recipes)](https://github.com/mlcommons/ck)
+to make it easier to prepare and benchmark different versions of ABTF models 
+(public or private) with MLPerf loadgen across different software and hardware.
+
+* Run and benchmark reference ABTF model via CM (CPU and CUDA): [README](docs/test-abtf-model/README.md)
+* Knowledge base: [README](docs/test-abtf-model/README-kb.md)

cmr.yaml

@@ -0,0 +1,13 @@
+alias: mlcommons@cm4abtf
+uid: 566d31eda11948a9
+
+git: true
+
+deps:
+  - alias: mlcommons@cm4mlops
+    uid: 9e97bb72b0474657
+
+  - alias: mlcommons@ck
+    uid: a4705959af8e447a
+    conflict: True
+

docs/test-abtf-model/README-cuda.md

@@ -0,0 +1,88 @@
+[ [Back to the main page](README.md) ]
+
+
+## Prepare workflow to benchmark ABTF model on CUDA-based device
+
+### Prerequisites
+
+* We expect that you already have CUDA driver installed
+* Tested with PyTorch 2.2.2 and CUDA 11.8 and 12.1
+
+
+
+### Detect or install CUDA toolkit and libraries
+
+```bash
+cmr "get cuda _toolkit _cudnn"
+cmr "get cuda-devices"
+```
+
+### Build MLPerf loadgen
+
+```bash
+cmr "get mlperf inference loadgen _copy" --version=main
+```
+
+
+### Install or detect PyTorch and PyTorchVision
+
+#### CUDA 11.8
+
+```bash
+cmr "get generic-python-lib _torch_cuda" --extra-index-url=https://download.pytorch.org/whl/cu118 --force-install
+cmr "get generic-python-lib _torchvision_cuda" --extra-index-url=https://download.pytorch.org/whl/cu118 --force-install
+```
+
+#### CUDA 12.1
+
+```bash
+cmr "get generic-python-lib _torch_cuda" --extra-index-url=https://download.pytorch.org/whl/cu121 --force-install
+cmr "get generic-python-lib _torchvision_cuda" --extra-index-url=https://download.pytorch.org/whl/cu121 --force-install
+```
+
+
+
+
+
+## Test Model with a test image
+
+```bash
+cmr "test abtf ssd-resnet50 cognata pytorch _cuda" --input=0000008766.png --output=0000008766_prediction_test.jpg --config=baseline_8MP_ss_scales --num-classes=13
+```
+
+## Benchmark model with MLPerf loadgen
+
+```bash
+cmr "generic loadgen python _pytorch _cuda _custom _cmc" --samples=5 --modelsamplepath=0000008766.png.cuda.pickle --modelpath=baseline_8mp_ss_scales_ep15.pth --modelcfg.num_classes=13 --modelcfg.config=baseline_8MP_ss_scales
+```
+
+
+## Benchmarking other models
+
+Other ways to download public or private model code and weights:
+```bash
+cmr "get ml-model abtf-ssd-pytorch _skip_weights" --adr.abtf-ml-model-code-git-repo.env.CM_ABTF_MODEL_CODE_GIT_URL=https://github.com/mlcommons/abtf-ssd-pytorch
+cmr "get ml-model abtf-ssd-pytorch _skip_weights" --model_code_git_url=https://github.com/mlcommons/abtf-ssd-pytorch --model_code_git_branch=cognata-cm
+cmr "get ml-model abtf-ssd-pytorch _skip_weights _skip_code"
+```
+
+Other ways to run local (private) model:
+
+You can first copy ABTF model code from GitHub to your local directory `my-model-code`.
+
+```
+cmr "generic loadgen python _pytorch _cuda _custom _cmc" --samples=5 --modelsamplepath=0000008766.png.cpu.pickle \
+  --modelpath=baseline_8mp_ss_scales_ep15.pth \
+  --modelcfg.num_classes=13 \
+  --modelcodepath="my-model-code" \
+  --modelcfg.config=baseline_8MP_ss_scales
+```
+
+
+
+
+
+## Feedback
+
+Join MLCommons discord or get in touch with developer: gfursin@cknowledge.org
+

docs/test-abtf-model/README-kb.md

@@ -0,0 +1,17 @@
+# Issues
+
+## Weird case on Windows
+
+### 20240410: Grigori
+
+If I download baseline_8mp_ss_scales_ep15.pth to the ROOT directory with the virtual environment,
+pip stops working since it considers this file as a broken package ...
+
+
+# Misc commands
+
+Register local ABTF model in CM cache to be the default
+
+```bash
+cmr "get ml-model abtf-ssd-pytorch _local.baseline_8mp_ss_scales_ep15.pth"
+```