# DestinyLinker/MingLi-Bench

A benchmark for evaluating LLMs on Chinese traditional fortune telling — Bazi (八字) and Ziwei Doushu (紫微斗数).

Repository: https://github.com/DestinyLinker/MingLi-Bench
Canonical: https://ross.abutalabs.com/products/mingli-bench
Homepage: https://destinyLinker.github.io/MingLi-Bench/
Language: Python
License: MIT
License Family: permissive
Last push: 2026-05-09T17:49:34+00:00

## Health v2 (maintenance only)
Score: 50/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 81, release rhythm 35, longevity 9
- inputs: {"age_days": 133, "days_push": 116, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2338, forks 344 (observed 2026-08-28T04:06:38.495323+00:00)

## What it is
A Python benchmark dataset and CLI for evaluating large language models on Chinese traditional fortune telling, covering Bazi (八字) and Ziwei Doushu (紫微斗数). It provides 160 normalized multiple-choice questions sourced from the 2022–2025 Global Fortune Teller Competition, with pre-computed astrological charts and multi-provider LLM evaluation support.

## Use cases
- evaluate llms on chinese fortune telling questions
- benchmark models on bazi reasoning
- test llm performance on ziwei doushu charts
- compare chain-of-thought vs direct answering on chinese metaphysics
- measure llm accuracy against human fortune teller competition results
- run model evaluations across openai anthropic deepseek and openrouter providers

## When to choose
- you need a standardized benchmark for LLM reasoning in Chinese traditional metaphysics
- you want to compare LLM performance against human expert baselines from fortune telling competitions
- you need pre-computed Bazi/Ziwei charts to isolate reasoning from chart derivation

## When to avoid
- you need a general-purpose Chinese NLP or reasoning benchmark
- you want a production fortune telling application rather than an evaluation harness
- your evaluation domain is unrelated to Chinese astrology or destiny analysis

## Facets
- artifact type: dataset
- maturity: active
- function: benchmarking, llm-inference, cli
- domain: artificial-intelligence, large-language-models
- platform: python, cli, cross-platform
- tags: llm-benchmark, chinese-fortune-telling, bazi, ziwei-doushu, multiple-choice-qa, astrology, chain-of-thought, model-evaluation, evaluation, natural-language-processing

## Member repositories
- DestinyLinker/MingLi-Bench (main) score 50

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:38.495323+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 2026-08-30T02:37:30.207127+00:00, confidence not recorded.
  - readme: https://github.com/DestinyLinker/MingLi-Bench (fetched 2026-08-28T04:06:38.495323+00:00, sha c28e8ddda07e)
  - homepage: https://destinyLinker.github.io/MingLi-Bench/ (fetched 2026-08-29T10:18:08.604464+00:00, sha 737169bfb166)
- Data as of 2026-08-30T08:39:29.467469+00:00.