# Docta-ai/docta

A Doctor for your data

Repository: https://github.com/Docta-ai/docta
Canonical: https://ross.abutalabs.com/products/docta
Language: Python
License: NOASSERTION
License Family: other
Topics: data, data-centric-ai, data-centric-machine-learning, data-curation, data-diagnosis, language-model, rlhf
Last push: 2026-06-16T06:03:43+00:00

## Health v2 (maintenance only)
Score: 69/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 87, release rhythm 35, longevity 87
- inputs: {"age_days": 1219, "days_push": 78, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3486, forks 257 (observed 2026-08-28T04:08:07.002776+00:00)

## What it is
Docta is a Python library for data-centric AI that diagnoses and fixes issues in datasets such as label errors, noise, and misannotations. It works training-free on tabular data, text, images, and pre-trained model embeddings, including LLM alignment/RLHF data.

## Use cases
- find label errors in my dataset
- clean noisy labels in RLHF data
- detect human annotation mistakes in LLM alignment data
- curate and diagnose image datasets
- fix mislabeled tabular data before training
- audit dataset quality without training a model

## When to choose
- you suspect label noise or annotation errors are hurting model performance
- you need training-free data diagnosis on tabular, text, or image data
- you are preparing RLHF or LLM alignment datasets and want to audit label quality

## When to avoid
- you need a fully automated production data pipeline rather than diagnostic tooling
- your use is commercial and you cannot accept the non-commercial CC BY-NC 4.0 license
- you need active data labeling or annotation workflow management

## Facets
- artifact type: library
- maturity: active
- function: data-science, machine-learning, nlp, etl
- domain: data-science, machine-learning, large-language-models, developer-tools
- platform: python, cross-platform
- tags: data-curation, data-diagnosis, label-errors, data-quality, rlhf, training-free, data-centric-ai

## Member repositories
- Docta-ai/docta (main) score 69

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:07.002776+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:36:23.008233+00:00, confidence not recorded.
  - readme: https://github.com/Docta-ai/docta (fetched 2026-08-28T04:08:07.002776+00:00, sha 7d6db27ae82c)
- Data as of 2026-08-30T08:39:29.467469+00:00.