[VLM] Offline scenario, performance-only mode of the reference implementation

.gitignore

@@ -3,3 +3,6 @@ libmlperf_loadgen.a
 __pycache__/
 generated/
 *.swp
+*.egg-info/
+*.so
+.vscode/

multimodal/vl2l/.gitignore

@@ -0,0 +1,170 @@
+# 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/latest/usage/project/#working-with-version-control
+.pdm.toml
+.pdm-python
+.pdm-build/
+
+# 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/
+
+# IDE
+.vscode
+
+# 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/
+
+# NPM
+node_modules
+package-lock.json
+/package.json

multimodal/vl2l/README.md

@@ -0,0 +1,130 @@
+# Reference Implementation for the Vision-language-to-language (VL2L) Benchmark 
+
+## Quick Start
+
+### Get the source code 
+
+Clone the MLPerf Inference repo via:
+
+```bash
+git clone --recurse-submodules https://github.com/mlcommons/inference.git mlperf-inference
+```
+
+Then enter the repo: 
+
+```bash
+cd mlperf-inference/
+```
+
+### Create a Conda environment
+
+Follow [this link](https://www.anaconda.com/docs/getting-started/miniconda/install#quickstart-install-instructions)
+on how to install Miniconda on your host machine. Then, you can create a new conda 
+environment via:
+
+```bash
+conda create -n mlperf-inf-mm-vl2l python=3.12
+```
+
+### Install LoadGen
+
+Update `libstdc++` in the conda environment:
+
+```bash
+conda install -c conda-forge libstdcxx-ng
+```
+
+Install `absl-py` and `numpy`:
+
+```bash
+conda install absl-py numpy
+```
+
+Build and install LoadGen from source:
+
+```bash
+cd loadgen/
+CFLAGS="-std=c++14 -O3" python -m pip install .
+cd ../
+```
+
+Run a quick test to validate that LoadGen was installed correctly:
+
+```bash
+python loadgen/demos/token_metrics/py_demo_server.py
+```
+
+### Install the VL2L benchmarking CLI
+
+For users, install `mlperf-inf-mm-vl2l` with:
+
+```bash
+pip install multimodal/vl2l/
+```
+
+For developers, install `mlperf-inf-mm-vl2l` and the development tools with:
+
+```bash
+pip install multimodal/vl2l/[dev]
+```
+
+After installation, you can check the CLI flags that `mlperf-inf-mm-vl2l` can take with:
+
+```bash
+mlperf-inf-mm-vl2l --help
+```
+
+You can enable shell autocompletion for `mlperf-inf-mm-vl2l` with:
+
+```bash
+mlperf-inf-mm-vl2l --install-completion
+```
+
+> NOTE: Shell auto-completion will take effect once you restart the terminal.
+
+### Start an inference endpoint on your local host machine with vLLM
+
+Please refer to [this guide on how to launch vLLM for various Qwen3 VL MoE models](https://docs.vllm.ai/projects/recipes/en/latest/Qwen/Qwen3-VL.html).
+
+```bash
+docker run --gpus all \                                 # Use all the GPUs on this host machine.
+    -v ~/.cache/huggingface:/root/.cache/huggingface \  # Use the HuggingFace cache from your host machine.
+    -p 8000:8000 \                                      # This assumes the endpoint will use port 8000.
+    --ipc=host \                                        # The container can access and utilize the host's IPC mechanisms (e.g., shared memory).
+    vllm/vllm-openai:nightly \                          # You can also use the `:latest` container or a specific release.
+        --model Qwen/Qwen3-VL-235B-A22B-Instruct \      # Specifies the model for vLLM to deploy.
+        --tensor-parallel-size 8 \                      # 8-way tensor-parallel inference across 8 GPUs.
+        --limit-mm-per-prompt.video 0                   # The input requests will contain images only (i.e., no videos).
+```
+
+### Run the benchmark for the Offline scenario
+
+Performance only mode:
+
+```bash
+mlperf-inf-mm-vl2l --settings.senario offline --settings.mode performance_only
+```
+
+Accuracy only mode:
+
+TBD
+
+### Run the benchmark for the Server scenario
+
+Performance only mode:
+
+TBD
+
+Accuracy only mode:
+
+TBD
+
+## Developer Guide
+
+### Linting
+
+You can lint the VL2L benchmark source code by running the following script:
+
+```bash
+bash multimodal/vl2l/scripts/linters.sh
+```