Reference the llama.cpp C API for model loading and inference workflows.
npx skills add https://github.com/datathings/marketplace --skill llamacpp
Please help me install this Agent Skill. Skill: llamacpp Source: https://github.com/datathings/marketplace/tree/main/plugins/llamacpp/skills/llamacpp Command: npx skills add https://github.com/datathings/marketplace --skill llamacpp
This Skill provides a comprehensive reference for the llama.cpp C API, covering core, context, inference, sampling, and advanced features. It serves as a centralized guide to quickly locate function usage, workflows, and best practices for local LLM deployment.
Browse the API references under references/ and try calls like llama_model_load_from_file(), llama_init_from_model(), llama_decode(), and llama_tokenize() against a local model.
High-intent search queries and answers about installing and using this skill.
To load a local model using the llama.cpp C API, call llama_model_load_from_file() followed by llama_init_from_model() to initialize the context required for inference and tokenization workflows.
Generating embeddings with llama.cpp involves loading your model and using the provided advanced API references to manage memory and extract vector representations directly through the C API workflows.
Tokenize text in llama.cpp by passing your input string to the llama_tokenize() function, which prepares the necessary token arrays before executing the llama_decode() inference call.
Yes, llama.cpp provides built-in memory management functions within its core and context APIs, allowing developers to manually control allocation and optimize resource usage during local LLM deployment.
Yes, you can apply LoRA adapters during inference using the advanced topic references in the llama.cpp API, which cover encoder-decoder usage and backend sampling modifications.
Limitations of the llama.cpp C API include the need for manual memory management and explicit context initialization, requiring developers to handle low-level resource allocation and sampling configurations directly.
kngender5Discover Hugging Face Hub models and build llama.cpp inference command lines.
xurenluDevelop RAG systems integrating vector databases, embedding models, and retrieval strategies.
Rawgrowth-ConsultingOptimize transformer attention performance and memory usage with Flash Attention.
HouseGarofaloDeploy, manage, and interact with local LLMs via Ollama.
