A large instruct dataset

This dataset is a combination of multiple sources, including the GPT4All dataset, the Alpaca dataset from Stanford, custom generation using AllenAI augmentation, and some dataset augmentation from open-source Meta datasets. The dataset is split into 70% for training, 20% for validation, and 10% for testing.

Description

The Swype.com dataset contains prompt and completion pairs for various tasks. It's an augmented version of the following datasets:

• GPT4All: A dataset containing a wide range of tasks for training and evaluating general-purpose language models.
• Alpaca dataset from Stanford: A dataset containing prompts, completions, and annotations for controllable text generation.
• Custom generation using AllenAI augmentation: Augmentation performed using the advanced NLP tools provided by AllenAI.
• Some dataset augmentation from open-source Meta datasets: Additional augmentation from various open-source Meta datasets.

The dataset is designed for training and evaluating language models on diverse tasks, with a focus on controllable and instruction-based text generation.

Dataset Structure

The dataset contains the following columns:

• prompt: The input prompt string, representing a task or question.
• completion: The output completion string, representing the answer or generated text based on the prompt.

Citation

If you use this dataset in your research or work, please cite it as follows:

@misc{srikanth2023swypedataset, author = {Srikanth Srinivas}, title = {Swype.com Dataset}, year = {2023}, publisher = {Swype.com}, howpublished = {\url{https://swype.com}}, email = {s@swype.com} }