Deep Learning Datasets

Created by GitHub user hayatoy and last updated in September 2017, this repository provides datasets specifically curated for use with TensorFlow, Keras, and Google ML Engine. The collection serves as a historical resource for deep learning training data, though specific record counts and internal schemas are not documented in the source metadata.

Use Cases

• Testing legacy 'tensorflow' and 'keras' pipeline ingestion scripts
• Benchmarking 'ml-engine' performance using 2017-era data structures
• Historical analysis of deep learning dataset curation practices from the late 2010s

Strengths

• Licensed under Apache-2.0, allowing for open commercial and private use
• Explicitly tagged for compatibility with the 'ml-engine' (Google Cloud ML) ecosystem
• Framework-specific targeting for Keras and TensorFlow users

Limitations

• Complete lack of documentation regarding data volume or file formats
• Temporal staleness with no maintenance or updates for over six years
• No descriptive metadata for fields or labels

Provenance

Source

GitHub repository hayatoy/deep-learning-datasets

Freshness

The dataset is stale, with the last recorded update occurring on September 22, 2017.