AI Paper Atlas

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

Neural Rule-Execution Tracking Machine For Transformer-Based Text Generation

Yufei Wang, Can Xu, Huang Hu, Chongyang Tao, Stephen Wan, Mark Dras, Mark Johnson, Daxin Jiang

NeurIPS 2021 · Large language models › Language, safety and interpretability · approach: Controlled text generation · 4 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 dynamic rule-execution tracking to T5-BaseT5-Base on ROCStories with sentence-index and ordering constraints · ROCStories Test SplitCSR: 98.7 -> 98.8; 97.0 -> 97.4Improved
Add dynamic rule-execution tracking to T5-BaseT5-Base on ROCStories with sentence-length constraints · ROCStories Test SplitCSR: 18.5 -> 84.5Improved
Add dynamic rule-execution tracking to T5-BaseT5-Base on ROCStories with two length-or-stop-word constraints · ROCStories Test SplitCSR: 11.0 -> 35.9Improved
Add dynamic rule-execution tracking to T5-BaseT5-Base on ROCStories · ROCStories Test Split, simple sentence-index controlR-L: 33.1 -> 33.3; 49.6 -> 49.5No clear effect
Use morphological inflection alternatives rather than exact-form copyingT5-Large + NRETM on CommonGen · CommonGen testCIDEr: 167.3 -> 175.0Improved
Add exact-copy constraints with NRETMT5-Large on CommonGen · CommonGen testCIDEr: 163.6 -> 167.3Improved
Use inflectional alternatives in copy constraintsT5-Large + NRETM on CommonGen · CommonGen testConstraint satisfaction ALL: 93.9 -> 99.4Improved
Apply Grid Beam Search to enforce lexical constraints at inferenceT5-Base on CommonGen · CommonGen testCIDEr: 159.5 -> 110.7Hurt
Add a translated-once sentence alignment constraintDoc-mBART25 on TED15 Zh-En · TED15 Zh-En testSAR: 98.7 % -> 100 %Improved
Add a translated-once sentence alignment constraintDoc-mBART25 on TED15 Zh-En · TED15 Zh-En tests-BLEU: + 1.1 BLEUImproved
Train on storyline positions 3, 4, and 5, then test positions 1 and 2T5 versus NRETM on ROCStories · ROCStories novel sentence-index testCSA: 19.7 -> 97.7Improved
Add NRETM to T5-BaseT5-Base on CommonGen · CommonGen testCIDEr: 159.5 -> 169.8Improved

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