L

llm-cpp

Single-header C++17: run a prompt N times, measure consistency, compare models or prompts and score responses. Part of llm-cpp.

Single-header C++17: OpenAI embeddings, cosine/dot/euclidean similarity and a small on-disk vector store. Part of llm-cpp.

Single-header C++17: shrink conversation history with head/tail/smart truncation, a sliding window or an LLM summary. Part of llm-cpp.

Single-header C++17: multi-turn conversation with token-budget trimming, a pinned system prompt, save and restore. Part of llm-cpp.

Single-header C++17: LRU response cache with TTL and hit/miss stats, so identical prompts skip the API. Part of llm-cpp.

Single-header C++17: run a JSONL file of prompts through a thread pool with rate limiting and resumable checkpoints. Part of llm-cpp.

Single-header C++17: tool-calling agent loop; register C++ lambdas as tools and let the model call them. Part of llm-cpp.