AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Relevance-Promoting Language Model for Short-Text Conversation

Xin Li, Piji Li, Wei Bi, Xiaojiang Liu, Wai Lam

AAAI 2020 · Large language models › Language, safety and interpretability · approach: Controlled text generation · 16 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 supervised source attention (SSA) and topic inference (TI) (ablation)Transformer language model with top-k sampling · Weibo STC testRelevance: 1.273 -> 1.606Improved
Add supervised source attention (SSA) (ablation)Transformer language model with top-k sampling and TI · Weibo STC testRelevance: 1.485 -> 1.606Improved
Add supervised source attention (SSA) (ablation)Transformer language model with top-k sampling and TI · Weibo STC testFluency: 1.303 -> 1.346Hurt
Add topic inference (TI) (ablation)Transformer language model with top-k sampling and SSA · Weibo STC testDist-2: 0.445 -> 0.544Improved
Add topic inference (TI) (ablation)Transformer language model with top-k sampling and SSA · Weibo STC testRelevance: 1.503 -> 1.606Improved
Use top-k sampling instead of beam searchFull relevance-promoting model · Weibo STC testDist-2: 0.398 -> 0.544Improved
Use top-k sampling instead of beam searchFull relevance-promoting model · Weibo STC testRelevance: 1.515 -> 1.606Improved
Use top-k sampling instead of beam searchTFM-S2S · Weibo STC testRelevance: 1.475 -> 1.270Hurt
Use top-k sampling instead of beam searchTFM-S2S · Weibo STC testDist-2: 0.290 -> 0.507Improved
Use component-wise Sigmoid instead of Softmax in topic inferenceTopic inference component · Weibo STC testModel performanceHurt
Train on the concatenated query-response sequence, including query-token likelihoodTransformer language model for STC · Weibo STC testRelevance: 1.412 -> 1.606Improved

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