Added mixtral, gpu support

Dockerfile

@@ -1,31 +1,41 @@
-ARG UBUNTU_VERSION=22.04
+ARG UBUNTU_VERSION=20.04
+ARG CUDA_VERSION=12.3.1
+ARG BASE_CUDA_DEV_CONTAINER=nvidia/cuda:${CUDA_VERSION}-devel-ubuntu${UBUNTU_VERSION}
+ARG BASE_CUDA_RUN_CONTAINER=nvidia/cuda:${CUDA_VERSION}-runtime-ubuntu${UBUNTU_VERSION}
 
-FROM ubuntu:$UBUNTU_VERSION as build
+FROM ${BASE_CUDA_DEV_CONTAINER} as build
 
-RUN apt-get update && \
-    apt-get install -y build-essential git wget
+ARG CUDA_DOCKER_ARCH=all
+
+RUN apt-get update && apt-get upgrade -y && \
+    apt-get install -y git build-essential gcc wget
+
+WORKDIR /build
 
-WORKDIR /app
 RUN git clone https://github.com/ggerganov/llama.cpp.git
 
-WORKDIR /app/llama.cpp
-RUN make
+WORKDIR /build/llama.cpp
+
+ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
+ENV LLAMA_CUBLAS=1
 
-WORKDIR /data
-RUN wget https://huggingface.co/IlyaGusev/saiga2_7b_gguf/resolve/main/model-q8_0.gguf -nv -O model.gguf
+RUN mkdir build && \
+    cd build && \
+    cmake .. -DLLAMA_CUBLAS=ON && \
+    cmake --build . --config Release
 
-FROM ubuntu:$UBUNTU_VERSION as runtime
+FROM ${BASE_CUDA_RUN_CONTAINER} as runtime
 
 WORKDIR /app
 
-COPY --from=build /app/llama.cpp/server /app
-COPY --from=build /data/model.gguf /data/model.gguf
+# Copy the executable from the build stage
+COPY --from=build /build/llama.cpp/build/bin/server /app
 COPY ./run.sh /app/run.sh
+WORKDIR /app
+EXPOSE 7867
 
+# Make the script executable
 RUN chmod +x run.sh
 
-ENV LC_ALL=C.utf8
-
-EXPOSE 7860
-
+# CMD to run your script
 CMD ./run.sh

build.bat

@@ -1 +1 @@
-docker build -t llama-test .
+docker build -t mixstral-api .

run.bat

@@ -1 +1 @@
-docker run --name llama-test -v "C:\Work\nornickel-llm\data\models\saiga7b-q8:/data" -p 7860:7860  llama-test 
+docker run --name mixstral-api -v "C:\Work\ai-models\mixtral-cpp:/models" -p 7867:7867  mixstral-api

run.sh

@@ -1,2 +1,14 @@
 #!/bin/bash
-/app/server -m /data/model.gguf -c 4096 --port 7860 --host 0.0.0.0
+file_path="/models/mixtral-8x7b-instruct-v0.1.Q2_K.gguf"
+url="https://huggingface.co/TheBloke/Mixtral-8x7B-Instruct-v0.1-GGUF/resolve/main/mixtral-8x7b-instruct-v0.1.Q2_K.gguf"
+# Check if the file exists
+if [ ! -e "$file_path" ]; then
+    echo "Downloading model."
+    # If the file doesn't exist, download it using wget
+    wget "$url" -O "$file_path"
+    # You can add more commands here if needed after the download
+    echo "File downloaded successfully."
+else
+    echo "File already exists."
+fi
+/app/server -m "$file_path" -c 100000 --port 7867 --host 0.0.0.0