Computer scienceData scienceNLPNLP metricsClassic NLP metrics

ROUGE

Meaningful evaluation

Report a typo

Which text generation models could be evaluated with the ROUGE metric so that the ROUGE scores are somewhat reflective of the model performance?

Select one option from the list

Models that are designed to capture the meaning from the source, not the text unit overlaps

Models that are designed to produce outputs that overlap with the human references

Any models could be meaningfully evaluated with the ROUGE scores

Models that produce the outputs that are considered high quality by the human experts

Create a free account to access the full topic