41,554 blogs · 5,228,115 posts Claim your blog

Record · corpus.blog/posts/01a0c530-c1dd-7199-9e27-0f1d342a6d13

NLG Evaluation 2025 vs 2015: much improved but needs to be better

Ehud Reiter · published 4 February 2025

A record is what was published and who pointed at it. The text of the post is not held here: read it at the source, or at its Wayback capture.

Read the post at ehudreiter.com Wayback capture

0blogs citing

First cited: —
Most recent: —
Rank this month: not ranked
External links: 1
Words captured: 1,002

Cited by

No blog we hold has cited this post yet.

Links from this post

Link

Host

https://aclanthology.org/2024.wmt-1.131/

aclanthology.org

Elsewhere on ehudreiter.com (12)

My New Book on NLG Read at the source ↗

https://ehudreiter.com/2018/07/02/why-bleu-poor-for-nlg/

ehudreiter.com

Why is ROUGE so popular? Read at the source ↗

Can LLM-based eval replace human evaluation? Read at the source ↗

Lets use error annotations to evaluate systems! Read at the source ↗

Unresponsive Authors and Experimental Flaws Read at the source ↗

Common Flaws in NLP Evaluation Experiments Read at the source ↗

Make your next experiment a bit better Read at the source ↗

Limits of pre-publication reviewing Read at the source ↗

We should evaluate real-world impact! Read at the source ↗

We need better LLM benchmarks Read at the source ↗

Do LLM benchmarks ignore NLG? Read at the source ↗