llama.cpp
Backend
- will install to /usr/local/ by default.
- llama.cpp for CANN
- Setting Up CUDA on Fedora
- llama.cpp for ET
- llama.cpp for OpenCL
- OpenVINO Backend for llama.cpp
- llama.cpp for SYCL
- GGML-VirtGPU Backend Configuration
- Development and Testing
- GGML-VirtGPU Backend
- llama.cpp for AMD ZenDNN
- Snapdragon-based devices
- Hexagon backend developer details
- Snapdragon-based Linux devices
- Snapdragon-based Windows devices
- llama.cpp for IBM zDNN Accelerator
Development
Multimodal
Llama.Cpp
- Android
- Auto-Parser Architecture
- Build profiling
- Build riscv64 spacemit
- Build llama.cpp locally (for s390x)
- Build llama.cpp locally
- Completions
- Docker
- Function Calling
- Install pre-built version of llama.cpp
- LLGuidance Support in llama.cpp
- Obtaining and quantizing models
- Using multiple GPUs with llama.cpp
- Multimodal
- GGML Operations
- llama.cpp INI Presets
- Release process
- Speculative Decoding
- XCFramework