tdpeuter/open-webui
tdpeuter/open-webui
1
Fork 0
forked from open-webui/open-webui
Code Issues Pull requests 1 Projects 1 Releases 2 Packages 1 Activity Actions

docker improvements & changed universal device type env for different models used

Browse source
This commit is contained in:
Jannik Streidl 2024-03-20 08:44:09 +01:00
parent 132d741c55
commit 1f6739337b
4 changed files with 36 additions and 19 deletions
Download patch file Download diff file Expand all files Collapse all files

37
Dockerfile
View file

@ -1,4 +1,7 @@
# syntax=docker/dockerfile:1
# Initialize device type args
ARG USE_CUDA=false
ARG USE_MPS=false
######## WebUI frontend ########
FROM node:21-alpine3.19 as build
@ -23,6 +26,10 @@ RUN npm run build
######## WebUI backend ########
FROM python:3.11-slim-bookworm as base
# Use args
ARG USE_CUDA
ARG USE_MPS
## Basis ##
ENV ENV=prod \
PORT=8080
@ -54,7 +61,8 @@ ENV RAG_EMBEDDING_MODEL="all-MiniLM-L6-v2" \
# Important:
# If you want to use CUDA you need to install the nvidia-container-toolkit (https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html)
# you can set this to "cuda" but its recomended to use --build-arg CUDA_ENABLED=true flag when building the image
RAG_EMBEDDING_MODEL_DEVICE_TYPE="cpu"
RAG_EMBEDDING_MODEL_DEVICE_TYPE="cpu" \
DEVICE_COMPUTE_TYPE="int8"
# device type for whisper tts and embbeding models - "cpu" (default), "cuda" (nvidia gpu and CUDA required) or "mps" (apple silicon) - choosing this right can lead to better performance
#### Preloaded models ##########################################################
@ -62,19 +70,24 @@ WORKDIR /app/backend
# install python dependencies
COPY ./backend/requirements.txt ./requirements.txt
RUN pip3 install -r requirements.txt --no-cache-dir
RUN if [ "$RAG_EMBEDDING_MODEL_DEVICE_TYPE" = "cuda" ]; then \
echo "CUDA enabled" && \
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117 --no-cache-dir; \
else \
RUN if [ "$USE_CUDA" = "true" ]; then \
export DEVICE_TYPE="cuda" && \
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117 --no-cache-dir && \
pip3 install -r requirements.txt --no-cache-dir; \
elif [ "$USE_MPS" = "true" ]; then \
export DEVICE_TYPE="mps" && \
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu --no-cache-dir && \
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'])"; \
pip3 install -r requirements.txt --no-cache-dir && \
python -c "import os; from faster_whisper import WhisperModel; WhisperModel(os.environ['WHISPER_MODEL'], device='cpu', compute_type='int8', download_root=os.environ['WHISPER_MODEL_DIR'])" && \
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['DEVICE_TYPE'])"; \
else \
export DEVICE_TYPE="cpu" && \
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu --no-cache-dir && \
pip3 install -r requirements.txt --no-cache-dir && \
python -c "import os; from faster_whisper import WhisperModel; WhisperModel(os.environ['WHISPER_MODEL'], device='cpu', compute_type='int8', download_root=os.environ['WHISPER_MODEL_DIR'])" && \
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['DEVICE_TYPE'])"; \
fi
# 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'])"
# install required packages
RUN apt-get update \
# Install pandoc and netcat
@ -100,4 +113,4 @@ COPY ./backend .
EXPOSE 8080
CMD [ "bash", "start.sh"]
CMD [ "bash", "start.sh"]