license: mit
Dataset Summary
The Programmatic Ad Creatives dataset contains 7097 samples of online programmatic ad creatives along with their ad sizes. The dataset includes 8 unique ad sizes, such as (300, 250), (728, 90), (970, 250), (300, 600), (160, 600), (970, 90), (336, 280), and (320, 50). The dataset is in a tabular format and represents a random sample from Project300x250.com's complete creative data set. It is primarily used for training and evaluating natural language processing models in the context of advertising creatives.
Supported Tasks
This dataset supports a range of tasks, including language modeling, text generation, and text augmentation. The full dataset has been utilized to fine-tune open-source models for creative ad copy. We hope this dataset will inspire contributors to join Project 300x250 in creating open-source alternatives to Google and Meta, ensuring the existence of independent advertising.
Languages
The dataset primarily consists of English language text.
Dataset Structure
Data Fields
The dataset contains the following fields:
- 'text': Represents the text collected from the programmatic ad creative.
- 'dimensions': Represents the dimensions of the creative ad size.
Data Splits
The data is not split into separate subsets; it is provided as a whole.
Dataset Creation
Curation Rationale
The dataset of online programmatic ad creatives was curated to serve as a valuable resource for researchers and developers. It provides a unique collection of advertising creative text that is typically only available within walled gardens. The dataset aims to foster the development of independent advertising alternatives to Google and Meta, particularly in the field of AI, by promoting open-source solutions in the advertising domain.
Source Data
The data is generated from a vast collection of programmatic creative images hosted by Project 300x250 . The text was extracted from each creative image.
Dataset Use
Use Cases
The dataset can be used for various tasks related to language understanding, natural language processing, machine learning model training, and model performance evaluation. Initially, the dataset has been utilized to fine-tune open-source models using programmatic ad text to generate unique ad copy. These models were created to inspire ad creatives and provide a starting point for developing effective marketing content.
Usage Caveats
As this dataset is a sampled subset, it is recommended to regularly check for updates and improvements or reach out to the author for access to the full dataset.