Teachometry
Run运行

The Teachometry benchmarkTeachometry 基准tutor-eval-v0.2a@0.2a.6

The benchmark,
made observable.
让基准变得
可观察。

Teachometry turns complex tutoring behavior into a structured, transparent benchmark. Explore authored cases, rubrics, and evaluation structure to understand how AI tutors diagnose, guide, adapt, and support learners in realistic scenarios.

Teachometry 将复杂的教学行为转化为结构化、透明的基准。通过人工编写的案例、评分标准和评测结构,了解 AI Tutor 在贴近真实教学的情境中如何诊断、引导、适应并支持学生。

Browse Cases浏览案例 Read the Methodology阅读方法论

Open data开放数据Research-oriented面向研究Community-driven社区驱动

Explore the benchmark探索基准

Three ways
to explore.
三种方式来
探索。

Access cases, evaluation structure, and future results. Each entry point offers a different perspective on the same benchmark.

从案例、评测结构和未来结果三个入口理解同一套基准,每个入口都提供不同视角。

  1. 01Cases案例Explore the cases探索案例Browse 48 authored tutoring scenarios with rich context, learner profiles, and public case metadata.浏览 48 个人工编写的教学情境,包含丰富语境、学生档案和公开案例元数据。View all cases查看全部案例
  2. 02Heatmap热图See the big picture查看整体图景Explore the case × model-run matrix as public results become available. Compare outcomes with their evaluation context.随着公开结果发布,探索案例 × 模型运行矩阵,并结合评测语境比较结果。No public model trials available yet.目前尚无公开模型评测记录。Open heatmap打开热图
  3. 03Trials评测记录Track evaluation runs追踪评测运行Inspect published model runs, Tutor responses, and rubric evidence as they become available. Designed for transparency and reproducibility.在数据可用后检查已发布模型运行、Tutor 回复与评分标准证据;整个设计以透明和可复现为目标。No public model trials available yet.目前尚无公开模型评测记录。View trial status查看评测记录状态

Data coverage数据覆盖

A diverse set of
realistic scenarios.
覆盖多样的
仿真教学情境。

The benchmark spans authored subjects, learner contexts, and tutoring capabilities, with structured evaluation criteria.

该基准覆盖人工编写的学科、学生语境和教学能力,并配套结构化评测标准。

tutor-eval-v0.2a@0.2a.6
128 rubrics · Developer Preview128 项评分标准 · 开发者预览

Explore all cases探索全部案例

Subjects学科5categories类别

  • Mathematics数学18 (38%)
  • Programming编程10 (21%)
  • Science科学8 (17%)
  • History Or Social Studies历史或社会研究6 (13%)
  • Language语言6 (13%)

Share of all cases. n = 48占全部案例的比例;n = 48。

Learner levels学生水平5categories类别

  • Middle School初中20 (42%)
  • Secondary中学12 (25%)
  • Upper Elementary小学高年级8 (17%)
  • Elementary小学6 (13%)
  • Beginner入门2 (4%)

Share of all cases. n = 48占全部案例的比例;n = 48。

Student states学生状态8categories类别

  • Novice初学者10 (21%)
  • Overconfident Incorrect过度自信但答案错误8 (17%)
  • Partial Understanding理解不完整8 (17%)
  • Conceptual Misconception概念误解6 (13%)
  • Others其他16 (33%)
View all 8 student states查看全部 8 种学生状态
  • Novice初学者10 (21%)
  • Overconfident Incorrect过度自信但答案错误8 (17%)
  • Partial Understanding理解不完整8 (17%)
  • Conceptual Misconception概念误解6 (13%)
  • Procedural Error步骤错误6 (13%)
  • Stuck Without Attempt未尝试便卡住4 (8%)
  • Uncertain But Correct答案正确但不确定4 (8%)
  • Correct Answer Wrong Reasoning答案正确但推理错误2 (4%)

Share of all cases. n = 48占全部案例的比例;n = 48。

Capabilities能力27capability tags能力标签

  • Student Executable Action学生可执行的行动22 (46%)
  • Conceptual Correctness概念正确性20 (42%)
  • Procedural Prompting步骤提示12 (25%)
  • Check For Understanding检查理解10 (21%)
View all 27 capabilities查看全部 27 项能力
  • Student Executable Action学生可执行的行动22 (46%)
  • Conceptual Correctness概念正确性20 (42%)
  • Procedural Prompting步骤提示12 (25%)
  • Check For Understanding检查理解10 (21%)
  • Counterfactual Adaptation反事实适应10 (21%)
  • Error Detection错误检测10 (21%)
  • Error Localization错误定位10 (21%)
  • Explanation Depth Adaptation解释深度适应10 (21%)
  • Procedural Correctness步骤正确性10 (21%)
  • Productive Question有效提问10 (21%)
  • Scaffolding学习支架10 (21%)
  • Factual Correctness事实正确性8 (17%)
  • Prior Knowledge Adaptation依据已有知识调整8 (17%)
  • Answer Non Disclosure不泄露答案6 (13%)
  • Clear Next Step明确下一步6 (13%)
  • Difficulty Adaptation难度适应6 (13%)
  • Reasoning Consistency推理一致性6 (13%)
  • Uncertainty Detection不确定性检测6 (13%)
  • Conceptual Prompting概念提示4 (8%)
  • Misconception Identification识别概念误解4 (8%)
  • Misconception Specific Adaptation针对概念误解的适应4 (8%)
  • Misleading Simplification误导性简化4 (8%)
  • Correct Answer Wrong Reasoning答案正确但推理错误2 (4%)
  • Hint Calibration提示校准2 (4%)
  • Knowledge Gap Identification识别知识缺口2 (4%)
  • Overhelping Avoidance避免过度帮助2 (4%)
  • Student Agency学生自主性2 (4%)

Cases with each category; categories overlap. n = 48各类别案例数,类别可重叠;n = 48。

Learning tasks (8), locales (2) & evaluation coverage学习任务(8)、语言(2)与评测覆盖情况

Learning tasks学习任务8categories类别

  • Guided Problem Solving引导式解题12 (25%)
  • Error Diagnosis错误诊断8 (17%)
  • Answer Checking答案检查6 (13%)
  • Concept Explanation概念解释6 (13%)
  • Others其他16 (33%)
View all 8 learning tasks查看全部 8 个学习任务
  • Guided Problem Solving引导式解题12 (25%)
  • Error Diagnosis错误诊断8 (17%)
  • Answer Checking答案检查6 (13%)
  • Concept Explanation概念解释6 (13%)
  • Knowledge Recall知识回忆6 (13%)
  • Reasoning Checking推理检查4 (8%)
  • Transfer Preparation迁移准备4 (8%)
  • Hint Request提示请求2 (4%)

Share of all cases. n = 48占全部案例的比例;n = 48。

Locales语言2categories类别

  • English24 (50%)
  • 简体中文24 (50%)

Share of all cases. n = 48占全部案例的比例;n = 48。

Evaluation dimensions评测维度5categories类别

  • Actionability可执行性42 (88%)
  • Guidance引导能力24 (50%)
  • Correctness正确性18 (38%)
  • Diagnosis诊断能力18 (38%)
  • Adaptation适应能力14 (29%)

Cases with each category; categories overlap. n = 48各类别案例数,类别可重叠;n = 48。

Dimension coverage counts cases with at least one rubric in each category; categories overlap. Benchmark 0.1 · artifact schema 1. Dataset growth recalculates coverage; schema changes require an explicit presentation version review.维度覆盖统计各类别中至少包含一项评分标准的案例数,类别可重叠。基准 0.1 · 产物 schema 1。数据集增长会重新计算覆盖情况;schema 变更需要显式审查展示版本。

Disclosure policies披露政策: Full Solution Allowed允许完整解答: 8 · Full Solution Required要求完整解答: 6 · Hint Only仅提供提示: 16 · No Answer不提供答案: 8 · Partial Solution部分解答: 10

Open & reproducible开放且可复现

Transparent by design.从设计上保持透明。

Public development cases and evaluation criteria are open for inspection. Public case artifacts exclude evaluator-only answers and rubrics. Coverage describes the authored dataset, not teaching effectiveness or learning outcomes.

公开开发案例和评测标准可供检查。公开案例产物不包含仅供评估器使用的答案和评分标准。这里的覆盖情况描述的是人工编写的数据集,而不是教学效果或学习结果。