Update start.sh

start.sh

@@ -5,13 +5,17 @@ if [ ! -d "llama.cpp" ]; then
   git clone https://github.com/ggerganov/llama.cpp
 fi
 
+export GGML_CUDA=OFF
 if [[ -z "${RUN_LOCALLY}" ]]; then
   # enable CUDA if NOT running locally
-  export LLAMA_CUDA=1
+  export GGML_CUDA=ON
 fi
 
 cd llama.cpp
-make -j llama-quantize llama-gguf-split llama-imatrix
+cmake -B build -DBUILD_SHARED_LIBS=OFF -DGGML_CUDA=${GGML_CUDA} -DLLAMA_CURL=ON
+cmake --build build --config Release -j --target llama-quantize llama-gguf-split llama-imatrix
+cp ./build/bin/llama-* .
+rm -rf build
 
 cd ..
 python app.py