tensorflow-dataset-api

Examples demonstrating dataset api capabilities.

1. Basic Placeholders
2. Basic Dataset
3. Feedable Dataset
4. Generator Dataset
5. Feedable Iterator
6. Feedable Generator Dataset
7. Feedable Generator Feedable Iterator Dataset

Note:

If facing issues in downloading the MNIST Dataset, use below bash command to update required certificates

/Applications/Python\ 3.6/Install\ Certificates.command

Choice of Benchmarks

1. EP1
2. EP3
3. EP5
4. EP11
5. EP14
6. EP17

Steps

1. Placeholder for Batch Size - For every dataset (Dataset API)
2. Handle for Iterator Switching (Model Runner / Flow API)
3. Dataset Creation - (Dataset API)
4. Initializable Iterator for Dataset - (Dataset API)
5. Feedable iterator - (Iterator Component API)
6. Placeholders with Default - (Inbuffer API)
7. String Handle for iterator (Dataset API)
8. Create Session (Model Runner / Flow API)
9. Fetch handle from running string handle (Model Runner / Flow API)
10. Initialize Iterator (Flow API)
11. Close Session (Model Runner / Flow API)