docs: add a doc for framework lib versions

FRAMEWORK_VERSIONS.md

@@ -0,0 +1,95 @@
+<!--
+SPDX-FileCopyrightText: Copyright (c) 2024-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+SPDX-License-Identifier: Apache-2.0
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+-->
+
+# Dynamo Framework Versions
+
+This document tracks the major dependencies and versions used in the NVIDIA Dynamo project.
+
+## Core Framework Dependencies
+
+### vLLM
+- **Version**: v0.11.0
+
+### TensorRT-LLM
+- **Version**: 1.1.0rc5
+
+### FlashInfer
+- **Version**: v0.3.1
+- **Description**: High-performance attention kernel library
+
+### NIXL
+- **Version**: 0.6.0
+- **Description**: NVIDIA's high-performance networking library for distributed LLM inference
+
+### UCX (Unified Communication X)
+- **Version**: v1.19.0
+- **Description**: Communication framework used by NIXL
+
+## Base Images
+
+### CUDA Development Image
+- **Version**: CUDA 12.8
+- **Base Image Tag**: `25.01-cuda12.8-devel-ubuntu24.04`
+- **Description**: NVIDIA CUDA development environment
+
+### CUDA Runtime Image
+- **Version**: CUDA 12.8.1
+- **Runtime Image Tag**: `12.8.1-runtime-ubuntu24.04`
+- **Description**: NVIDIA CUDA runtime environment for production deployments
+
+## Framework-Specific Configurations
+
+### vLLM Configuration
+- **CUDA Version**: 12.8
+- **FlashInfer Integration**: Enabled for source builds or ARM64 builds
+- **Build Location**: `container/deps/vllm/install_vllm.sh`
+
+### TensorRT-LLM Configuration
+- **Runtime Image**: `12.9.1-runtime-ubuntu24.04` (CUDA 12.9.1)
+- **Build Location**: `container/Dockerfile.trtllm`
+
+### SGLang Configuration
+- **CUDA Version**: 12.8
+- **Base Image**: Same as vLLM (`25.01-cuda12.8-devel-ubuntu24.04`)
+- **Build Location**: `container/Dockerfile.sglang`
+
+## Dependency Management
+
+### Build Scripts
+- **Main Build Script**: `container/build.sh`
+- **vLLM Installation**: `container/deps/vllm/install_vllm.sh`
+- **TensorRT-LLM Wheel**: `container/build_trtllm_wheel.sh`
+- **NIXL Installation**: `container/deps/trtllm/install_nixl.sh`
+
+### Python Dependencies
+- **Requirements File**: `container/deps/requirements.txt`
+- **Standard Requirements**: `container/deps/requirements.standard.txt`
+- **Test Requirements**: `container/deps/requirements.test.txt`
+
+## Notes
+
+- FlashInfer is only used when building vLLM from source or for ARM64 builds
+- Different frameworks may use slightly different CUDA versions for runtime images
+- NIXL and UCX are primarily used for distributed inference scenarios
+- The dependency versions are centrally managed through Docker build arguments and shell script variables
+
+## Container Documentation
+
+For detailed information about container builds and usage, see:
+- [Container README](container/README.md)
+- [Container Build Script](container/build.sh)
+- [Container Run Script](container/run.sh)