ethanker/agentic_coding_dataset

---
license: mit
task_categories:
- text-generation
- question-answering
language:
- en
tags:
- code
- programming
- agentic
size_categories:
- 10K<n<100K
---

# Agentic Coding Dataset

This dataset is a compilation of various coding and instruction-following datasets, designed to train agentic coding models.

## Sources

This dataset aggregates samples from the following sources:

1. **[CodeAlpaca-20k](https://huggingface.co/datasets/sahil2801/CodeAlpaca-20k)**
* Instruction-following coding tasks.
2. **[Evol-CodeAlpaca-v1](https://huggingface.co/datasets/theblackcat102/evol-codealpaca-v1)**
* Complex evolved coding instructions (WizardCoder style).
3. **[Code Review Instruct](https://huggingface.co/datasets/Dahoas/code-review-instruct-critique-revision-python)**
* Python code review, critique, and revision examples.
4. **[APPS (Automated Programming Progress Standard)](https://huggingface.co/datasets/codeparrot/apps)**
* Coding problems with solutions.
5. **[Shell Command Instruct](https://huggingface.co/datasets/byroneverson/shell-cmd-instruct)**
* Shell command instructions and outputs.

## Structure

The dataset is provided in JSONL format with the following fields:
* `instruction`: The user prompt or problem description.
* `output`: The expected code solution or response.

## Usage

```python
from datasets import load_dataset

dataset = load_dataset("ethanker/agentic_coding_dataset")
```