---
tags:
- gsm8k
- tamil
- translation
- qa
license: mit
---

# GSM8K Tamil - Translated Version

This dataset contains **question-answer pairs from GSM8K**, fully translated into Tamil using **Deep Translate**. It can be used for training and evaluating QA models in Tamil and also for reasoning models.

## Dataset Description

- **Language:** Tamil  
- **Format:** JSON  
- **Content:** Each entry is an object with:
  - `"question"`: The math question in Tamil
  - `"answer"`: The corresponding answer in Tamil  

There are **no predefined splits** (all data is in a single JSON file).

## Usage

You can load the dataset using the `datasets` library:

```python
from datasets import load_dataset

# Login with `huggingface-cli login` if private
ds = load_dataset("Vishal0407/GSM8K_TAMIL")

# Access the first question-answer pair
print(ds[0])