$ -weight: 500;">git clone https://github.com/ggerganov/llama.cpp cd llama.cpp make -weight: 500;">git clone https://github.com/ggerganov/llama.cpp cd llama.cpp make -weight: 500;">git clone https://github.com/ggerganov/llama.cpp cd llama.cpp make -weight: 500;">curl -fsSL https://ollama.com/-weight: 500;">install.sh | sh -weight: 500;">curl -fsSL https://ollama.com/-weight: 500;">install.sh | sh -weight: 500;">curl -fsSL https://ollama.com/-weight: 500;">install.sh | sh # 모델 다운로드 mkdir -p models -weight: 500;">wget https://huggingface.co/TheBloke/Qwen-1.5B-Chat-GGUF/resolve/main/qwen-1.5b-chat.Q4_K_M.gguf -P models/ # 서버 실행 ./server -m models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 # 모델 다운로드 mkdir -p models -weight: 500;">wget https://huggingface.co/TheBloke/Qwen-1.5B-Chat-GGUF/resolve/main/qwen-1.5b-chat.Q4_K_M.gguf -P models/ # 서버 실행 ./server -m models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 # 모델 다운로드 mkdir -p models -weight: 500;">wget https://huggingface.co/TheBloke/Qwen-1.5B-Chat-GGUF/resolve/main/qwen-1.5b-chat.Q4_K_M.gguf -P models/ # 서버 실행 ./server -m models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 # 모델 푸시 ollama push qwen:1.5b # API 호출 -weight: 500;">curl http://localhost:11434/api/generate \ -H "Content-Type: application/json" \ -d '{ "model": "qwen:1.5b", "prompt": "Hello, how are you?", "stream": false }' # 모델 푸시 ollama push qwen:1.5b # API 호출 -weight: 500;">curl http://localhost:11434/api/generate \ -H "Content-Type: application/json" \ -d '{ "model": "qwen:1.5b", "prompt": "Hello, how are you?", "stream": false }' # 모델 푸시 ollama push qwen:1.5b # API 호출 -weight: 500;">curl http://localhost:11434/api/generate \ -H "Content-Type: application/json" \ -d '{ "model": "qwen:1.5b", "prompt": "Hello, how are you?", "stream": false }' -weight: 600;">sudo nano /etc/systemd/system/llama.-weight: 500;">service -weight: 600;">sudo nano /etc/systemd/system/llama.-weight: 500;">service -weight: 600;">sudo nano /etc/systemd/system/llama.-weight: 500;">service [Unit] Description=Local LLM Server After=network.target [Service] Type=simple User=your_user WorkingDirectory=/home/your_user/llama.cpp ExecStart=/home/your_user/llama.cpp/server -m /home/your_user/models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 Restart=always RestartSec=10 [Install] WantedBy=multi-user.target [Unit] Description=Local LLM Server After=network.target [Service] Type=simple User=your_user WorkingDirectory=/home/your_user/llama.cpp ExecStart=/home/your_user/llama.cpp/server -m /home/your_user/models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 Restart=always RestartSec=10 [Install] WantedBy=multi-user.target [Unit] Description=Local LLM Server After=network.target [Service] Type=simple User=your_user WorkingDirectory=/home/your_user/llama.cpp ExecStart=/home/your_user/llama.cpp/server -m /home/your_user/models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 Restart=always RestartSec=10 [Install] WantedBy=multi-user.target -weight: 600;">sudo -weight: 500;">systemctl daemon-reload -weight: 600;">sudo -weight: 500;">systemctl -weight: 500;">enable llama.-weight: 500;">service -weight: 600;">sudo -weight: 500;">systemctl -weight: 500;">start llama.-weight: 500;">service -weight: 600;">sudo -weight: 500;">systemctl daemon-reload -weight: 600;">sudo -weight: 500;">systemctl -weight: 500;">enable llama.-weight: 500;">service -weight: 600;">sudo -weight: 500;">systemctl -weight: 500;">start llama.-weight: 500;">service -weight: 600;">sudo -weight: 500;">systemctl daemon-reload -weight: 600;">sudo -weight: 500;">systemctl -weight: 500;">enable llama.-weight: 500;">service -weight: 600;">sudo -weight: 500;">systemctl -weight: 500;">start llama.-weight: 500;">service # 추론 시간 측정 time -weight: 500;">curl -X POST http://localhost:8080/completion \ -H "Content-Type: application/json" \ -d '{"prompt":"Hello world", "n_predict":50}' # 추론 시간 측정 time -weight: 500;">curl -X POST http://localhost:8080/completion \ -H "Content-Type: application/json" \ -d '{"prompt":"Hello world", "n_predict":50}' # 추론 시간 측정 time -weight: 500;">curl -X POST http://localhost:8080/completion \ -H "Content-Type: application/json" \ -d '{"prompt":"Hello world", "n_predict":50}' # 실시간 CPU/MEM 사용량 htop # 로그 확인 journalctl -u llama.-weight: 500;">service -f # 실시간 CPU/MEM 사용량 htop # 로그 확인 journalctl -u llama.-weight: 500;">service -f # 실시간 CPU/MEM 사용량 htop # 로그 확인 journalctl -u llama.-weight: 500;">service -f # 100MB 모델로 간단한 테스트 ollama run qwen:1.5b # 100MB 모델로 간단한 테스트 ollama run qwen:1.5b # 100MB 모델로 간단한 테스트 ollama run qwen:1.5b import requests def call_local_llm(prompt): response = requests.post('http://localhost:8080/completion', json={'prompt': prompt, 'n_predict': 100}) return response.json()['content'] # 사용 예제 result = call_local_llm("Python으로 간단한 웹 서버를 만들어줘") print(result) import requests def call_local_llm(prompt): response = requests.post('http://localhost:8080/completion', json={'prompt': prompt, 'n_predict': 100}) return response.json()['content'] # 사용 예제 result = call_local_llm("Python으로 간단한 웹 서버를 만들어줘") print(result) import requests def call_local_llm(prompt): response = requests.post('http://localhost:8080/completion', json={'prompt': prompt, 'n_predict': 100}) return response.json()['content'] # 사용 예제 result = call_local_llm("Python으로 간단한 웹 서버를 만들어줘") print(result) # CPU 코어 제한 taskset -c 0-3 ./server -m models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 # 메모리 최적화 export CUDA_LAUNCH_BLOCKING=1 export CUDA_DEVICE_MAX_CONNECTIONS=1 # CPU 코어 제한 taskset -c 0-3 ./server -m models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 # 메모리 최적화 export CUDA_LAUNCH_BLOCKING=1 export CUDA_DEVICE_MAX_CONNECTIONS=1 # CPU 코어 제한 taskset -c 0-3 ./server -m models/qwen-1.5b-chat.Q4_K_M.gguf -c 2048 --port 8080 # 메모리 최적화 export CUDA_LAUNCH_BLOCKING=1 export CUDA_DEVICE_MAX_CONNECTIONS=1 # 메모리 부족 error: failed to allocate memory # 해결: 메모리 제한 해제 ulimit -v unlimited # 포트 충돌 bind: Address already in use # 해결: 포트 변경 ./server --port 8081 # 메모리 부족 error: failed to allocate memory # 해결: 메모리 제한 해제 ulimit -v unlimited # 포트 충돌 bind: Address already in use # 해결: 포트 변경 ./server --port 8081 # 메모리 부족 error: failed to allocate memory # 해결: 메모리 제한 해제 ulimit -v unlimited # 포트 충돌 bind: Address already in use # 해결: 포트 변경 ./server --port 8081 - OS: Ubuntu 20.04 이상 - GPU: NVIDIA GTX 1060 이상 (CUDA 11.8 필요) - RAM: 최소 16GB (24GB 이상 권장) - 저장공간: 최소 50GB (모델별로 2-10GB 필요) - 모델 선택: 정확도가 중요한 작업은 Q5_K_M 이상 사용 - GPU 메모리: 16GB RAM 기준 8GB 모델 추천 - API 설정: 8080 포트로 기본 설정 유지 - 로그 관리: Systemd 로그 자동 회전 설정