# gururise/AlpacaDataCleaned

Alpaca dataset from Stanford, cleaned and curated

Repository: https://github.com/gururise/AlpacaDataCleaned
Canonical: https://ross.abutalabs.com/products/alpacadatacleaned
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-03-07T14:15:41+00:00

## Health v2 (maintenance only)
Score: 62/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 71, release rhythm 35, longevity 90
- inputs: {"age_days": 1261, "days_push": 179, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1606, forks 158 (observed 2026-08-28T04:05:10.460761+00:00)

## What it is
A cleaned and curated version of the Stanford Alpaca instruction-tuning dataset used to train the Alpaca LLM. It fixes quality issues in the GPT-3-generated original data and provides benchmark comparisons showing improved fine-tuning results.

## Use cases
- fine-tune a llama model with cleaner instruction data
- get a better alpaca dataset than the original stanford one
- improve lora fine-tuning loss curves
- download cleaned instruction tuning data for 7b models
- compare fine-tuning datasets by benchmark results
- train an alpaca-style chatbot on curated data

## When to choose
- you are fine-tuning LLaMA-family models with LoRA and want higher-quality instruction data
- you saw poor results or loss curves with the original Alpaca dataset
- you want benchmarked evidence that a dataset improves model performance

## When to avoid
- you need a large-scale modern instruction dataset rather than a cleaned 2023-era one
- you are not doing LLM fine-tuning
- you need multilingual instruction data

## Facets
- artifact type: dataset
- maturity: maintenance
- function: machine-learning, data-science, etl
- domain: large-language-models, machine-learning
- platform: python
- tags: alpaca, llm-training-data, instruction-tuning, data-cleaning, lora, fine-tuning, natural-language-processing

## Member repositories
- gururise/AlpacaDataCleaned (main) score 62

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:10.460761+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 2026-08-30T03:52:13.846408+00:00, confidence not recorded.
  - readme: https://github.com/gururise/AlpacaDataCleaned (fetched 2026-08-28T04:05:10.460761+00:00, sha 7142eb60353c)
- Data as of 2026-08-30T08:39:29.467469+00:00.