awkrail / geneval

Automatic evaluation for natural language generation (NLG) systems.

geneval

Automatic evaluation for natural language generation (NLG) systems. It takes input as pairs of generated sentece and references and outputs values of metrics.

Metrics

Implemented

• BLEU
• Distinct-N

Todo

• METEOR
• ROUGE
• CIDEr
  and more

Setup

[WIP]

Usage

[WIP]

Example

[WIP]

Reference

This repo is based on the following repositories:

• nlgeval
• sumeval