Cleanup for v0.1.0

Signed-off-by: Adityan Jothi <ajothi@nvidia.com>

Author: adityanjothi

.github/CODEOWNERS

@@ -1,17 +1,14 @@
-#cpp code owners
-cpp/               @<org>/<repo>-cpp-codeowners
-
 #python code owners
-python/            @<org>/<repo>-python-codeowners
+modules @metropolis-sdg-team/metropolis-sdg-codeowners
+
+#notebooks code owners
+notebooks @metropolis-sdg-team/metropolis-sdg-codeowners
 
-#cmake code owners
-**/CMakeLists.txt  @<org>/<repo>-cmake-codeowners
-**/cmake/          @<org>/<repo>-cmake-codeowners
+#data code owners
+data @metropolis-sdg-team/metropolis-sdg-codeowners
 
 #build/ops code owners
-.github/           @<org>/ops-codeowners 
-ci/                @<org>/ops-codeowners
-conda/             @<org>/ops-codeowners
-**/Dockerfile      @<org>/ops-codeowners
-**/.dockerignore   @<org>/ops-codeowners
-docker/            @<org>/ops-codeowners
+deploy @metropolis-sdg-team/metropolis-sdg-codeowners
+.github @metropolis-sdg-team/metropolis-sdg-codeowners
+pyproject.toml @metropolis-sdg-team/metropolis-sdg-codeowners
+

.github/ISSUE_TEMPLATE/bug_report.md

@@ -1,13 +1,41 @@
-# ___PROJECT___ 0.0.0 (DD Mon YYYY)
+# Metropolis SDG for Smart Cities 0.1.0 (18 Nov 2025)
 
 ## New Features
 
-- ...
+- End-to-end SDG workflow:
+  - Stage 1: CARLA 0.9.16 ground-truth generation (RGB, Depth, Semantic/Instance Segmentation, Normals, Edges, masks, ODVG JSONs)
+  - Stage 2: Cosmos-Transfer2.5 video augmentation with Gradio client
+  - Stage 3: SoM-aligned post-processing and VLM-ready artifacts
+- Docker Compose deployment:
+  - Homogeneous mode: single-host stack for NIM services, Cosmos-Transfer, CARLA, and Jupyter Workbench
+  - Heterogeneous mode: split NIM stack and Workbench across hosts with `NIM_HOST`
+- One-command deploy script (`deploy/compose/deploy.sh`):
+  - `./deploy.sh` (homogeneous), `./deploy.sh nim`, `./deploy.sh workbench`
+  - Per-service GPU pinning via `*_GPU_ID` and `*_GPU_COUNT`
+  - Health checks guidance and URLs printed
+- Notebook (`notebooks/carla_synthetic_data_generation.ipynb`):
+  - Self-guided flow: ground truth generation → captioning → template/prompt generation → augmentation → post-processing
+  - NIM endpoints and ports derived from env
+- Config assets and samples:
+  - CARLA scenario JSON and camera YAML examples
+  - Augmentation YAML (`modules/augmentation/configs/config_carla.yaml`)
+  - Class filter YAML for masked edges
 
 ## Improvements
 
-- ...
+- README updates:
+  - Quickstart with clear homogeneous/heterogeneous steps
+  - Troubleshooting covers image access errors, health checks, GPU runtime setup, Docker permissions, port issues, and CARLA GPU usage.
+- Advanced configuration guide (`data/docs/advanced_configuration.md`):
+  - Field-by-field documentation for env, CARLA JSON/YAML, augmentation YAML
+  - Validation rules and type constraints, recommendations, and examples
+  - Notes on external reference configs (Inverted AI examples)
+- Deploy UX:
+  - `workbench` mode prompts to confirm `NIM_HOST` loaded from `deploy/compose/env`
+  - README clarifies `NIM_HOST` in `env` overrides previously exported values
 
 ## Bug Fixes
 
-- ...
+- Robust config validation for augmentation pipeline (required sections and fields)
+- Safer handling of missing seeds and prompt files
+- Graceful errors for missing controls per selected modalities in Cosmos execution

.github/ISSUE_TEMPLATE/documentation-request.md

@@ -2,16 +2,6 @@
 
 If you use this project or its components in your research, please cite the project and the third-party systems it builds upon.
 
-## Cite This Project
-```tex
-@misc{smartcity_sdg_2025,
-  title        = {Synthetic Data Generation for Smart City Applications},
-  author       = {Project Contributors},
-  year         = {2025},
-  note         = {Software repository}
-}
-```
-
 ## NIM Components
 
 ### Cosmos-Reason1-7B (VLM)

.github/ISSUE_TEMPLATE/feature_request.md

@@ -140,8 +140,8 @@ Notes:
 1) Clone the repository 
 
 ```bash 
-git clone <repository.git>
-cd its-github 
+git clone https://github.com/NVIDIA/metropolis-sdg-smart-cities.git
+cd metropolis-sdg-smart-cities 
 ```
 
 2) Download sample CARLA logs
@@ -191,7 +191,7 @@ Once the NIM stack is up, launch the CARLA server and notebook/workbench stack,
 
 ```bash
 # On the second machine, ensure steps 1-3 are complete to have the repository and configuration ready before this step.
-export NIM_HOST=<ip_of_nim_node> # Set NIM host to IP of machine 1 printed in the previous step.
+# The deployment script sources `deploy/compose/env`, where `NIM_HOST` defaults to `localhost`. This will override any previously exported `NIM_HOST`. Before running `./deploy.sh workbench`, edit `deploy/compose/env` and set `NIM_HOST=<ip_of_nim_node>`. The script will prompt you to confirm the detected value.
 cd deploy/compose
 ./deploy.sh workbench
 ```
@@ -220,34 +220,6 @@ curl http://$HOST:8002/v1/health/ready  # LLM should return "Service is live."
     - Stage 3: SoM-aligned post-processing for VLM training
 ---
 
-## Configuration
-
-Environment file (`deploy/compose/env.example`)
-- Set credentials: `NGC_API_KEY`, `HF_TOKEN`
-- Select GPUs per service: `*_GPU_ID`, `*_GPU_COUNT`
-- Ports: `NOTEBOOK_PORT`, `VLM_PORT`, `LLM_PORT`, `TRANSFER_GRADIO_PORT`, `CARLA_PORT`
-- Endpoints auto-derived from host and ports
-
-CARLA ground-truth generation:
-- JSON workflow/config and per-camera YAMLs (see `modules/carla-ground-truth-generation/config/`)
-- Typical invocation from the repository root directory:
-```bash
-cd modules/carla-ground-truth-generation
-python main.py --config config/wrong_way.json
-# Or with workflow config overrides
-python main.py --wf-config wf-config.json --config config/wrong_way.json
-```
-
-COSMOS augmentation (Cosmos-Transfer2.5):
-- YAML config drives cameras, template/prompt generation, and COSMOS execution (see `modules/augmentation/configs/`)
-- Typical invocation from workbench shell:
-```bash
-uv run modules/augmentation/modules/cli.py --config modules/augmentation/configs/config_carla.yaml
-```
-
-Notebook workflow:
-- `notebooks/carla_synthetic_data_generation.ipynb`
-
 ## Advanced Configuration
 - For a complete list of configuration options, field definitions, and recommendations, see `data/docs/advanced_configuration.md`.
 
@@ -291,6 +263,27 @@ Example configs and logs can be found [here](https://github.com/inverted-ai/metr
 ```bash
 docker run --rm --gpus all nvidia/cuda:12.2-runtime-ubuntu22.04 nvidia-smi
 ```
+- Set NVIDIA runtime as default so containers can access GPU:
+  - Edit `/etc/docker/daemon.json`:
+    ```json
+    {
+      "runtimes": {
+        "nvidia": {
+          "path": "nvidia-container-runtime",
+          "runtimeArgs": []
+        }
+      },
+      "default-runtime": "nvidia"
+    }
+    ```
+  - Then restart Docker and relaunch containers:
+    ```bash
+    sudo systemctl daemon-reload && sudo systemctl restart docker
+    # If already running, bring down and redeploy
+    docker compose -f deploy/compose/docker-compose.nim.yml --env-file deploy/compose/env down || true
+    docker compose -f deploy/compose/docker-compose.workbench.yml --env-file deploy/compose/env down || true
+    cd deploy/compose && ./deploy.sh
+    ```
 
 **NGC login required:**
 - `docker login nvcr.io` with `$oauthtoken` and your `NGC_API_KEY`
@@ -375,6 +368,10 @@ docker logs transfer2_5-gradio
 # [11-10 19:25:53|INFO|cosmos_transfer2/inference.py:108:__init__] Saved config to outputs/config.yaml
 ```
 
+**Slow CARLA simulation:**
+- If you encounter the message `WARNING: lavapipe is not a conformant vulkan implementation, testing use only.`, this usually means your container is not using the NVIDIA container runtime, or your GPU is not supported by CARLA. 
+  - To resolve this, ensure the NVIDIA runtime is enabled for Docker (see the setup section for details on configuring it as default).
+  - If your GPU is unsupported, try to match your setup to the hardware recommendations in this README or use a supported environment.
 
 ---
 

.github/ISSUE_TEMPLATE/submit-question.md

@@ -1,4 +1,4 @@
-## Advanced Configuration Guide
+## Advanced Configuration Guide — Metropolis SDG Smart Cities
 
 This guide documents all configuration knobs used by the SDG workflow across CARLA ground-truth generation, augmentation/orchestration, and deployment/runtime. It explains each field, types, defaults, constraints, and recommendations for first-time success and advanced tuning.
 

CHANGELOG.md

@@ -126,7 +126,7 @@ case "$MODE" in
     echo "[deploy] Mode: heterogeneous (Workbench-only on this node)"
     if ! check_nvcr_login; then
       echo "[deploy][warn] Not logged into nvcr.io. If your workbench Dockerfile uses NVCR base images, please login:"
-      echo "  docker login nvcr.io (username: oauthtoken; password: your NGC API key)"
+      echo "  docker login nvcr.io (username: $oauthtoken; password: your NGC API key)"
     fi
     if [ -z "${NIM_HOST:-}" ]; then
       if [ -f env ]; then
@@ -136,10 +136,22 @@ case "$MODE" in
     fi
     if [ -z "${NIM_HOST:-}" ]; then
       echo "[deploy][error] NIM_HOST not set. Set NIM_HOST in the environment or 'env' file before deploying Workbench."
-      echo "Example: export NIM_HOST=<NIM_NODE_IP> && ./deploy.sh het-workbench"
+      echo "Example: export NIM_HOST=<NIM_NODE_IP> && ./deploy.sh workbench"
       exit 1
     fi
     echo "[deploy] Using NIM_HOST: $NIM_HOST"
+    # Confirm NIM_HOST looks correct to the user before proceeding
+    read -r -p "[deploy] Is this the correct NIM host for your NIM node? [y/N]: " _ans
+    case "${_ans:-N}" in
+      y|Y|yes|YES)
+        ;;
+      *)
+        echo "[deploy][info] Please edit 'deploy/compose/env' and set NIM_HOST to your NIM node IP, then re-run:"
+        echo "  cd deploy/compose && vi env    # or your preferred editor"
+        echo "  ./deploy.sh workbench"
+        exit 1
+        ;;
+    esac
     docker compose -f docker-compose.workbench.yml --env-file env up -d
     ;;