Skip to main content
The first configuration of a strategy comparison. Every other configuration is compared with it: metric differences, and the questions it fixed or broke.
A classic keyword-ranking method: chunks containing the question’s rarer words, more often, rank higher. Used by hybrid search.
A passage of a document, typically a few paragraphs. Documents are split into chunks, and retrieval returns chunks. See Chunking.
The method used to split documents into chunks (Recursive, Character, Token, Markdown, Code or JSON), with its settings such as size and overlap.
An embedding model with its retrieval pipeline, for example text-embedding-3-small - hybrid - rerank cohere/rerank-v3.5. Evaluations and comparisons run configurations.
All the documents (and their chunks) in a project.
The measure of how close two embeddings are. Higher means more similar. Scores are only comparable within one model.
The length of the vectors a model produces. More dimensions means more storage per chunk.
A retrieved chunk that isn’t one of the expected chunks for the question.
A list of numbers (a vector) that represents the meaning of a text. Texts with similar meanings have close embeddings.
The model that turns text into embeddings. It’s the main thing Truvec helps you choose.
The chunks that contain the answer to a test question. The answer key used to grade a model.
The chunks the first search (dense or hybrid) returns for a reranker to re-score: 30 by default, from 1 to 100, and at least top k.
A set of test questions with their expected chunks. See Golden datasets.
The share of questions for which at least one expected chunk is in the top k.
Vector search and BM25 keyword search run together, their rankings merged with reciprocal rank fusion. Helps on exact terms such as product names and error codes. See Hybrid search and reranking.
Large language model, such as GPT or Claude. In RAG, it writes the answer from the retrieved chunks.
The average of 1 ÷ the rank of the first expected chunk. 1.0 means the right chunk is always first.
A 0-to-1 score of how well all expected chunks are ranked, giving more credit to higher positions.
Text repeated between consecutive chunks, so ideas aren’t cut in half at chunk boundaries.
The share of the top k retrieved chunks that are expected chunks.
A technique where a system retrieves relevant passages from your documents and gives them to an LLM to answer a question. See How RAG retrieval works.
How hybrid search merges the dense and keyword rankings: each chunk scores by its rank in each list, weighted by the Keyword / Dense setting.
The share of a question’s expected chunks that appear in the top k.
A model (a cross-encoder) that reads the question with each first-stage candidate and re-scores them, keeping the top k. Useful when the right chunk is retrieved but not ranked first. Truvec offers Cohere Rerank 3.5 and Voyage rerank-2.5 and rerank-2.5-lite. See Hybrid search and reranking.
Finding the chunks most relevant to a question. It’s the step Truvec evaluates.
What runs around the embedding model in a configuration: dense or hybrid search, and an optional reranker. Chunking isn’t part of it, because it changes the chunks themselves.
The similarity of the best expected chunk minus the similarity of the best distractor. A larger gap means the model separates right from wrong chunks more clearly.
The unit models read and providers bill: roughly three quarters of an English word.
The number of chunks retrieved for each question, and the cut-off at which metrics are computed.
A database that stores embeddings and finds the ones closest to a query. Truvec uses Qdrant.