Merge branch 'main' of hf.co:spaces/muryshev/llama-cpp-server

Dockerfile

@@ -1,42 +1,31 @@
 ARG UBUNTU_VERSION=22.04
-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 ${BASE_CUDA_DEV_CONTAINER} as build
-
-ARG CUDA_DOCKER_ARCH=all
+FROM ubuntu:$UBUNTU_VERSION as build
 
 RUN apt-get update && \
-    apt-get install -y build-essential git cmake wget
-
-WORKDIR /build
+    apt-get install -y build-essential git wget
 
+WORKDIR /app
 RUN git clone https://github.com/ggerganov/llama.cpp.git
 
-WORKDIR /build/llama.cpp
-
-ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
-ENV LLAMA_CUBLAS=1
-
-RUN mkdir build && \
-    cd build && \
-#    cmake .. #-DLLAMA_CUBLAS=ON && \
-    cmake .. && \
-    cmake --build . --config Release
+WORKDIR /app/llama.cpp
+RUN make
 
 WORKDIR /data
 RUN wget https://huggingface.co/IlyaGusev/saiga2_7b_gguf/resolve/main/model-q8_0.gguf -nv -O model.gguf
 
-FROM ${BASE_CUDA_RUN_CONTAINER} as runtime
-
+FROM ubuntu:$UBUNTU_VERSION as runtime
 
 WORKDIR /app
 
-# Copy the executable from the build stage
-COPY --from=build /build/llama.cpp/build/bin/server /app
+COPY --from=build /app/llama.cpp/server /app
 COPY --from=build /data/model.gguf /data/model.gguf
-WORKDIR /app
+COPY ./run.sh /app/run.sh
+
+RUN chmod +x run.sh
+
+ENV LC_ALL=C.utf8
+
 EXPOSE 7860
 
-CMD ./server -m /data/model.gguf -c 2048 --port 7860 --host 0.0.0.0
+CMD ./run.sh

run.sh

@@ -0,0 +1,2 @@
+#!/bin/bash
+/app/server -m /data/model.gguf -c 4096 --port 7860 --host 0.0.0.0