open-webui/Dockerfile

80 lines
2.9 KiB
Docker
Raw Normal View History

2023-10-09 00:38:42 +02:00
# syntax=docker/dockerfile:1
FROM node:alpine as build
2023-11-15 01:28:51 +01:00
WORKDIR /app
2024-01-08 05:50:09 +01:00
# wget embedding model weight from alpine (does not exist from slim-buster)
RUN wget "https://chroma-onnx-models.s3.amazonaws.com/all-MiniLM-L6-v2/onnx.tar.gz" -O - | \
tar -xzf - -C /app
2024-01-08 05:50:09 +01:00
COPY package.json package-lock.json ./
RUN npm ci
COPY . .
RUN npm run build
2023-10-09 00:38:42 +02:00
2024-01-08 05:50:09 +01:00
2024-01-07 17:28:35 +01:00
FROM python:3.11-slim-bookworm as base
2023-11-15 01:28:51 +01:00
ENV ENV=prod
2024-01-13 04:38:30 +01:00
ENV PORT ""
2024-01-06 02:16:35 +01:00
ENV OLLAMA_API_BASE_URL "/ollama/api"
2024-01-05 03:55:15 +01:00
ENV OPENAI_API_BASE_URL ""
ENV OPENAI_API_KEY ""
ENV WEBUI_SECRET_KEY ""
2023-11-15 01:28:51 +01:00
ENV SCARF_NO_ANALYTICS true
ENV DO_NOT_TRACK true
######## Preloaded models ########
# whisper TTS Settings
ENV WHISPER_MODEL="base"
2024-02-15 08:32:54 +01:00
ENV 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 persormance 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.
2024-02-18 20:16:10 +01:00
ENV RAG_EMBEDDING_MODEL="all-MiniLM-L6-v2"
# device type for whisper tts and ebbeding 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="cpu"
2024-02-19 19:56:50 +01:00
ENV RAG_EMBEDDING_MODEL_DIR="/app/backend/data/cache/embedding/models"
ENV SENTENCE_TRANSFORMERS_HOME $RAG_EMBEDDING_MODEL_DIR
######## Preloaded models ########
2023-11-15 01:28:51 +01:00
WORKDIR /app/backend
# install python dependencies
2023-11-15 01:28:51 +01:00
COPY ./backend/requirements.txt ./requirements.txt
2024-01-08 06:22:37 +01:00
RUN pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu --no-cache-dir
RUN pip3 install -r requirements.txt --no-cache-dir
2024-01-26 20:57:35 +01:00
# Install pandoc and netcat
2024-01-23 08:11:50 +01:00
# RUN python -c "import pypandoc; pypandoc.download_pandoc()"
RUN apt-get update \
2024-01-26 20:57:35 +01:00
&& apt-get install -y pandoc netcat-openbsd \
2024-01-23 08:11:50 +01:00
&& rm -rf /var/lib/apt/lists/*
# preload embedding model
RUN python -c "import os; from chromadb.utils import embedding_functions; sentence_transformer_ef = embedding_functions.SentenceTransformerEmbeddingFunction(model_name=os.environ['RAG_EMBEDDING_MODEL'], device=os.environ['RAG_EMBEDDING_MODEL_DEVICE_TYPE'])"
# preload tts model
RUN python -c "import os; from faster_whisper import WhisperModel; WhisperModel(os.environ['WHISPER_MODEL'], device='auto', compute_type='int8', download_root=os.environ['WHISPER_MODEL_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
# copy built frontend files
COPY --from=build /app/build /app/build
# copy backend files
2023-11-15 01:28:51 +01:00
COPY ./backend .
2024-01-30 01:24:16 +01:00
CMD [ "bash", "start.sh"]