AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Conclusion-Supplement Answer Generation for Non-Factoid Questions

Makoto Nakatsuji, Sohei Okui

AAAI 2020 · Large language models › Retrieval and knowledge · approach: Explanatory and multiple-choice QA benchmarks · 9 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add conclusion-to-supplement attention (ablation)NAGM · Oshiete-gooROUGE-L: 0.291 -> 0.299Improved
add conclusion-to-supplement attention (ablation)NAGM · Oshiete-gooBLEU-4: 0.147 -> 0.158Improved
add conclusion-to-supplement attention (ablation)NAGM · nfL6ROUGE-L: 0.383 -> 0.402Improved
add conclusion-to-supplement attention (ablation)NAGM · nfL6BLEU-4: 0.157 -> 0.181Improved
add sentence-type embeddings (ablation)NAGM · Oshiete-gooROUGE-L: 0.235 -> 0.299Improved
add sentence-type embeddings (ablation)NAGM · Oshiete-gooBLEU-4: 0.090 -> 0.158Improved
add sentence-type embeddings (ablation)NAGM · nfL6ROUGE-L: 0.349 -> 0.402Improved
add sentence-type embeddings (ablation)NAGM · nfL6BLEU-4: 0.067 -> 0.181Improved
set ensemble-loss weight alpha to 1 instead of 0 (ablation)NAGM · Oshiete-gooROUGE-L: 0.251 -> 0.299Improved
set ensemble-loss weight alpha to 1 instead of 0 (ablation)NAGM · nfL6BLEU-4: 0.062 -> 0.181Improved
increase ensemble-loss weight alpha from 1 to 2 (ablation)NAGM · Oshiete-gooROUGE-L: 0.299 -> 0.211Hurt
increase ensemble-loss weight alpha from 1 to 2 (ablation)NAGM · nfL6BLEU-4: 0.181 -> 0.023Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy