Merge pull request #2 from coder3101/add-dockerfile

feat: Add Dockerfile and compose with usage instructions

README.md

@@ -46,6 +46,8 @@ Requirements:
 - Python 3.11+
 - Node.js 18+
 
+### Native
+
 ```bash
 git clone https://github.com/extraltodeus/j-wash.git
 cd j-wash
@@ -63,6 +65,51 @@ python -X utf8 run.py
 
 Then open **http://localhost:8381**.
 
+### Docker
+
+Requirements:
+- NVIDIA GPU with CUDA 12.4+ drivers
+- [Docker](https://docs.docker.com/engine/install/) with the [NVIDIA Container Toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html)
+
+```bash
+# Build the image
+docker build -t j-wash .
+
+# Run (data and HF cache persist on the host)
+docker run --gpus all \
+  -p 8381:8381 \
+  -v ./data:/app/data \
+  -v ./hf_cache:/app/hf_cache \
+  -v ./lenses:/app/lenses \
+  j-wash
+```
+
+Or use Docker Compose:
+
+```bash
+docker compose up -d
+```
+
+The container binds to `0.0.0.0` by default so the UI is reachable from outside
+the container. Pass additional arguments (e.g. `--port`, `--host`) after the image
+name:
+
+```bash
+docker run --gpus all -p 8381:8381 j-wash --port 8381 --host 0.0.0.0
+```
+
+Set `HF_TOKEN` for gated/private models:
+
+```bash
+docker run --gpus all -p 8381:8381 -e HF_TOKEN=hf_your_token j-wash
+# or with docker-compose:
+#   HF_TOKEN=hf_your_token docker compose up -d
+```
+
+> **Note**: The Docker image includes the `jacobian-lens` library and pre-builds the
+> React front-end at build time. Model weights are downloaded at runtime into the
+> mounted `hf_cache` volume (shared with the host).
+
 
 ## Running