Update QUICKSTART.md

QUICKSTART.md

@@ -103,37 +103,6 @@ print(f"Top relations: {results['summary']['top_relations']}")
   - Checkpoints: ~811MB
   - Code repositories: ~1GB
 
-## What the Setup Script Does
-
-1. **Creates Conda Environment** (`vine_demo`)
-   - Python 3.10
-   - PyTorch 2.7.1 with CUDA 12.6
-
-2. **Installs Dependencies**
-   - transformers, huggingface-hub, safetensors
-   - opencv-python, pillow, matplotlib
-   - All required packages for video processing
-
-3. **Clones Repositories**
-   - `video-fm/video-sam2` (SAM2 segmentation)
-   - `video-fm/GroundingDINO` (object detection)
-   - `kevinxuez/LASER` (video processing utilities)
-   - `kevinxuez/vine_hf` (VINE HuggingFace interface)
-
-4. **Installs Packages**
-   - Installs all repositories in editable mode (`-e`)
-   - Builds GroundingDINO native extensions
-
-5. **Downloads Checkpoints**
-   - SAM2 model and config from Facebook Research
-   - GroundingDINO model and config from IDEA Research
-
-6. **Tests Installation**
-   - Verifies PyTorch and CUDA
-   - Loads VINE model from HuggingFace
-   - Creates pipeline
-   - Confirms all components work
-
 ## Troubleshooting
 
 ### CUDA Not Available