Edit model card
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/model-cards#model-card-metadata)

Pre-trained evaluator in EMNLP 2022 paper

Towards a Unified Multi-Dimensional Evaluator for Text Generation

Introduction

Multi-dimensional evaluation is the dominant paradigm for human evaluation in Natural Language Generation (NLG), i.e., evaluating the generated text from multiple explainable dimensions, such as coherence and fluency.

However, automatic evaluation in NLG is still dominated by similarity-based metrics (e.g., ROUGE, BLEU), but they are not sufficient to portray the difference between the advanced generation models.

Therefore, we propose UniEval to bridge this gap so that a more comprehensive and fine-grained evaluation of NLG systems can be achieved.

Pre-trained Evaluator

unieval-dialog is the pre-trained evaluator for the dialogue response generation task. It can evaluate the model output from five dimensions:

  • naturalness
  • coherence
  • engagingness
  • groundedness
  • understandability

Usage

Please refer to our GitHub repository.

Downloads last month
2,888
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.