Initial commit.

Author: wangshangsam

multimodal/vl2l/README.md

@@ -0,0 +1,4 @@
+# MLPerf Inference Vision-language-to-language (VL2L) Reference Implementation
+
+[Qwen3-VL-235B-A22B](https://github.com/QwenLM/Qwen3-VL/tree/4aae93c9fcca19d7cbca6b095d64f578d25ed75f)
+is currently selected as the model for the VL2L benchmark.

multimodal/vl2l/pyproject.toml

@@ -0,0 +1,13 @@
+[tool.ruff]
+lint.select = ["ALL"]
+lint.ignore = [
+  "ANN002",  # Missing type annotation for `*args`"
+  "ANN003",  # Missing type annotation for `**kwargs`
+]
+
+[tool.ruff.lint.pydocstyle]
+convention = "google"
+
+[tool.mypy]
+check_untyped_defs = true
+plugins = ['pydantic.mypy']