Optimize Dockerfile for CUDA support

Refactored the Dockerfile to better organize and streamline environment variable settings, emphasizing support for a CUDA-based WebUI backend while retaining the ability to build a CPU-only image. Consolidated ENV commands to reduce layers, improving build efficiency, and set a default PORT environment to enhance container usability. Enabled exposure of the backend service on port 8080 and leveraged combined RUN directives to minimize the image footprint. These changes facilitate a more robust deployment process, catering to both CPU and CUDA environments.

Dockerfile-cuda

@@ -11,48 +11,53 @@ RUN npm ci
 COPY . .
 RUN npm run build
 
-######## WebUI backend ########
+######## CPU-only WebUI backend ########
+# To support both CPU and GPU backend, we need to keep the ability to build the CPU-only image.
+#FROM python:3.11-slim-bookworm as base
+#FROM --platform=linux/amd64 ubuntu:22.04 AS cpu-builder-amd64
+#FROM --platform=linux/amd64 cpu-builder-amd64 AS cpu-build-amd64
+#RUN OPENWEBUI_CPU_TARGET="cpu" sh gen_linux.sh
+#FROM --platform=linux/amd64 cpu-builder-amd64 AS cpu_avx-build-amd64
+#RUN OPENWEBUI_CPU_TARGET="cpu_avx" sh gen_linux.sh
+#FROM --platform=linux/amd64 cpu-builder-amd64 AS cpu_avx2-build-amd64
+#RUN OPENWEBUI_CPU_TARGET="cpu_avx2" sh gen_linux.sh
+
+######## CUDA WebUI backend ########
 ARG CUDA_VERSION=12.3.2
-#FROM nvidia/cuda:$CUDA_VERSION-devel-ubuntu22.04 as base
 FROM --platform=linux/amd64 nvidia/cuda:$CUDA_VERSION-devel-ubuntu22.04 AS cuda-build-amd64
-
 # Set environment variables for NVIDIA Container Toolkit
-ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64
-ENV NVIDIA_DRIVER_CAPABILITIES=all
-ENV NVIDIA_VISIBLE_DEVICES=all
-
-# Install NVIDIA CUDA toolkit and libraries in the container
-#RUN apt-get update && \
-#    apt-get install -y --no-install-recommends nvidia-cuda-toolkit nvidia-cuda-dev nvidia-cudnn-dev
-
-ENV ENV=prod
-ENV PORT ""
+ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64 \
+    NVIDIA_DRIVER_CAPABILITIES=all \
+    NVIDIA_VISIBLE_DEVICES=all
 
-ENV OLLAMA_BASE_URL "/ollama"
+ENV ENV=prod \
+    PORT=8080
 
-ENV OPENAI_API_BASE_URL ""
-ENV OPENAI_API_KEY ""
+## Base URL Config ##
+ENV OLLAMA_BASE_URL="/ollama" \
+    OPENAI_API_BASE_URL=""
 
-ENV WEBUI_SECRET_KEY ""
-
-ENV SCARF_NO_ANALYTICS true
-ENV DO_NOT_TRACK true
+## API Key and Security Config ##
+ENV OPENAI_API_KEY="" \
+    WEBUI_SECRET_KEY="" \
+    SCARF_NO_ANALYTICS=true \
+    DO_NOT_TRACK=true
 
 ######## Preloaded models ########
 # whisper TTS Settings
-ENV WHISPER_MODEL="base"
-ENV WHISPER_MODEL_DIR="/app/backend/data/cache/whisper/models"
+ENV WHISPER_MODEL="base" \
+    WHISPER_MODEL_DIR="/app/backend/data/cache/whisper/models"
 
 # RAG Embedding Model Settings
 # any sentence transformer model; models to use can be found at https://huggingface.co/models?library=sentence-transformers
 # Leaderboard: https://huggingface.co/spaces/mteb/leaderboard 
 # for better performance and multilangauge support use "intfloat/multilingual-e5-large" (~2.5GB) or "intfloat/multilingual-e5-base" (~1.5GB)
 # IMPORTANT: If you change the default model (all-MiniLM-L6-v2) and vice versa, you aren't able to use RAG Chat with your previous documents loaded in the WebUI! You need to re-embed them.
-ENV RAG_EMBEDDING_MODEL="all-MiniLM-L6-v2"
-# device type for whisper tts and embedding models - "cpu" (default), "cuda" (NVIDIA GPU and CUDA required), or "mps" (apple silicon) - choosing this right can lead to better performance
-ENV RAG_EMBEDDING_MODEL_DEVICE_TYPE="cuda"
-ENV RAG_EMBEDDING_MODEL_DIR="/app/backend/data/cache/embedding/models"
-ENV SENTENCE_TRANSFORMERS_HOME $RAG_EMBEDDING_MODEL_DIR
+ENV RAG_EMBEDDING_MODEL="all-MiniLM-L6-v2" \
+    # device type for whisper tts and embedding models - "cpu" (default), "cuda" (NVIDIA GPU and CUDA required), or "mps" (apple silicon) - choosing this right can lead to better performance
+    RAG_EMBEDDING_MODEL_DEVICE_TYPE="cuda" \
+    RAG_EMBEDDING_MODEL_DIR="/app/backend/data/cache/embedding/models" \
+    SENTENCE_TRANSFORMERS_HOME=$RAG_EMBEDDING_MODEL_DIR
 
 ######## Preloaded models ########
 WORKDIR /app/backend
@@ -63,12 +68,8 @@ RUN apt-get update && \
     rm -rf /var/lib/apt/lists/*
 
 COPY ./backend/requirements.txt ./requirements.txt
-RUN pip3 install torch torchvision torchaudio --no-cache-dir
-RUN pip3 install -r requirements.txt --no-cache-dir
-
-# copy embedding weight from build
-RUN mkdir -p /root/.cache/chroma/onnx_models/all-MiniLM-L6-v2
-COPY --from=build /app/onnx /root/.cache/chroma/onnx_models/all-MiniLM-L6-v2/onnx
+RUN pip3 install torch torchvision torchaudio --no-cache-dir && \
+    pip3 install -r requirements.txt --no-cache-dir
 
 # copy built frontend files
 COPY --from=build /app/build /app/build
@@ -78,4 +79,6 @@ COPY --from=build /app/package.json /app/package.json
 # copy backend files
 COPY ./backend .
 
+EXPOSE 8080
+
 CMD [ "bash", "start.sh"]