AI & ML Tech Glossary

PEFT adapts a pretrained model by training only a small number of additional or selected parameters, which reduces the compute, memory, and storage required compared...

RLHF aligns a model with human preferences by training a reward model from human judgments and then using reinforcement learning to optimize the model to...

Agent-to-Agent Protocol (A2A) is a structured messaging pattern that lets AI agents delegate tasks, share results, and coordinate safely using shared schemas, retries, and traceable...

Prompt leakage is when an AI system unintentionally reveals hidden prompts or sensitive context—like system instructions, tool schemas, or private RAG documents—often due to prompt...

KV cache quantization stores a transformer’s attention KV cache in lower precision (e.g., INT8/INT4) to cut GPU memory use and serve more long-context requests concurrently,...

Hybrid search combines keyword (BM25) and vector (embedding) retrieval to improve relevance—capturing exact matches like IDs and error codes while still handling paraphrases and semantic...

RAG groundedness measures how well an LLM’s answer is supported by the retrieved context. It focuses on claim-level faithfulness and citation support, helping reduce hallucinations...

Reranking is a second-stage retrieval step that reorders initially retrieved RAG chunks using a stronger relevance model (often a cross-encoder) so the LLM receives higher-precision...