NecroMOnk/khan-math-linear_algebra

---
license: mit
language:
- en
tags:
- math
- khan-academy
- latex
- education
- chatml
pretty_name: Khan Math – Linear Algebra
size_categories:
- 1M<n<10M
---

# Khan Math – Linear Algebra

Linear algebra problems from Khan Academy in ChatML format. Covers vectors, matrices, eigenvalues, transformations, and more.

## Quick Start

```python
from datasets import load_dataset

dataset = load_dataset("NecroMOnk/khan-math-linear_algebra")

print(dataset["train"][0])
```

Example record:

```json
{
"messages": [
{"role": "system", "content": "You are a mathematics tutor. Answer the following math problem."},
{"role": "user", "content": "Find the arclength of the function $f(x) = \\log(2x)$ on the interval $x=4$ to $x=5$"},
{"role": "assistant", "content": "$-\\sqrt{17}+\\sqrt{26}+\\tanh^{-1}\\left(\\sqrt{17}\\right)-\\tanh^{-1}\\left(\\sqrt{26}\\right)$"}
],
"topic": "calculus",
"subtopic": "arclength"
}
```

## Dataset Stats

- **1.295M** problems
- ChatML format
- Source: Khan Academy linear algebra materials
- Language: English
- Math notation: LaTeX

## Fields

| Field | Type | Description |
|---|---|---|
| `messages` | list | ChatML turns: system, user, assistant |
| `topic` | string | Top-level math topic (e.g. `calculus`) |
| `subtopic` | string | Specific subtopic (e.g. `arclength`) |

## Source

Problems sourced from [Khan Academy](https://www.khanacademy.org/) via the AMPS dataset.
Problems and answers are in LaTeX format.

## Versions

- **v1** – raw extraction, minor artifacts possible (e.g. degenerate intervals, `+-` notation)
- **v2** – cleaned formulas, extraction artifacts removed *(coming soon)*