Create and curate ground-truth evaluation datasets for TruLens.
npx skills add https://github.com/truera/trulens --skill trulens-dataset-curation
Please help me install this Agent Skill. Skill: trulens-dataset-curation Source: https://github.com/truera/trulens/tree/main/skills/dataset-curation Command: npx skills add https://github.com/truera/trulens --skill trulens-dataset-curation
Creating reliable evaluation data for LLMs is time-consuming and error-prone. This Skill helps you build ground-truth datasets that enable consistent measurement of model outputs against expected responses, chunks, and retrieval results, empowering teams to track improvements across versions.
High-intent search queries and answers about installing and using this skill.
You create ground-truth evaluation datasets by structuring queries, expected responses, and expected chunks into a pandas DataFrame, then persisting it via the TruLens API for versioned assessments.
A TruLens ground-truth dataset requires a schema with fields like query, query_id, expected_response, and expected_chunks, enabling consistent measurement of model outputs against expected retrieval results.
You persist and load evaluation datasets using TruLens by calling session.add_ground_truth_to_dataset with a pandas DataFrame and metadata, then retrieving it later with session.get_ground_truth.
Yes, you can share ground-truth datasets across teams for versioned evaluations by persisting them with TruLens using dataset metadata like domain and version, standardizing collaboration.
Yes, you need pandas to manage ground-truth data for TruLens, as it handles data structuring and manipulation before you persist the evaluation datasets through the TruSession API.
The best way to organize expected responses and retrieved chunks is building a pandas DataFrame with query, expected_response, and expected_chunks fields, then persisting it via TruLens for structured evaluation.
crossxwillQuery Parquet data with DuckDB and convert results to pandas DataFrames.
trueraOrchestrates end-to-end TruLens evaluation workflows for LLM apps.
trueraOrchestrate TruLens evaluations across apps and versions with TruChain, TruGraph, and TruLlama.
Ikram-AlamEvaluate agent outputs against rubrics and ground truth with multi-dimensional scoring.
