AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

A Multi-Task Multi-Stage Transitional Training Framework for Neural Chat Translation

Chulun Zhou, Yunlong Liang, Fandong Meng, Jie Zhou, Jinan Xu, Hongji Wang, Min Zhang, Jinsong Su

TPAMI 2022 · Architectures and efficiency › Pruning and efficient tuning · approach: Parameter-efficient transfer analysis · 7 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 MMT training instead of two-stage Flat-NCT+FTFlat-NCT (Base) · BConTrasT En->De testBLEU: 58.15 -> 59.33Improved
add MMT training instead of two-stage Flat-NCT+FTFlat-NCT (Base) · BMELD En->Zh testBLEU: 24.76 -> 27.43Improved
add MMT training instead of two-stage Flat-NCT+FTFlat-NCT (Big) · BMELD En->Zh testBLEU: 26.45 -> 28.62Improved
use context-aware models instead of sentence-level Transformer+FTcontext-aware NMT with two-stage training · BConTrasT En->De testBLEU: 58.43 vs 58.15No clear effect
add monolingual dialogues at stages 2 and 3 (ablation)Flat-NCT+MMT · BConTrasT En->De validBLEU: 59.74 -> 60.86Improved
add source and target monolingual dialogues at stage 3 (ablation)Flat-NCT+MMT · BConTrasT En->De validBLEU: 60.51 -> 60.86Improved
add both UD and SD tasks (ablation)Flat-NCT+MMT · BConTrasT En->De validBLEU: 59.79 -> 60.86Improved
add UD task on monolingual dialogues (ablation)Flat-NCT+MMT · BConTrasT En->De validBLEU: 59.96 -> 60.86Improved
add SD task on monolingual dialogues (ablation)Flat-NCT+MMT · BConTrasT En->De validBLEU: 60.13 -> 60.86Improved
add UD task on monolingual dialogues (ablation)Flat-NCT+MMT · BConTrasT De->En validBLEU: 60.41 vs 60.94Mixed
use authentic instead of pseudo Chinese monolingual dialogues (ablation)Flat-NCT+MMT w/o SD (Base) · BMELD En->Zh testBLEU: 27.35 -> 27.80Improved
use authentic instead of pseudo Chinese monolingual dialogues (ablation)Flat-NCT+MMT w/o SD (Big) · BMELD Zh->En testBLEU: 22.87 -> 23.36Improved

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