AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Multi-Head Adapter Routing for Cross-Task Generalization

Lucas Caccia, Edoardo Ponti, Zhan Su, Matheus Pereira, Nicolas Le Roux, Alessandro Sordoni

NeurIPS 2023 · Architectures and efficiency › Pruning and efficient tuning · approach: Adapter-based parameter-efficient tuning · 3 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
replace single-head routing Poly with multi-head routing Poly-SPoly · T0 few-shot (T5-XL)avg test accuracy: 68.0 -> 69.1Improved
route only the multi-head routing function (Poly-S-z) vs Poly-zPoly-z · T0 few-shot (T5-XL)avg test accuracy: 66.4 -> 68.3Improved
replace learned routing with random routing (Random-mu) (ablation)Poly · T0 (T5-XL)test accuracy: 68.0 -> 66.0Hurt
use Poly-S instead of Poly at 11B backbonePoly · T0-11B few-shotavg test accuracy: 74.9 -> 74.7No clear effect
routing-only tuning on 11B: Poly-S-z vs Poly-zPoly-z · T0-11B few-shotavg test accuracy: 70.0 -> 72.9Improved
Poly-S vs LoRA on SuperNILoRA · SuperNIRouge-L: 67.6 -> 68.5Improved
Poly-S vs Poly on SuperNIPoly · SuperNIRouge-L: 67.8 -> 68.5Improved
routing-only tuning on SuperNI: Poly-S-z vs Poly-zPoly-z · SuperNIRouge-L: 64.6 -> 68.0Improved
larger LoRA rank (LoRA-big) instead of LoRALoRA · SuperNIRouge-L: 67.6 -> 67.2No clear effect
average pre-trained modules (Poly-mu) instead of learned routing at adaptation (ablation)Poly · T0 (T5-XL)test accuracy: 68.0 -> 67.8No clear effect
Poly-mu averaging instead of shared LoRA adapterLoRA · T0 (T5-XL)test accuracy: 66.0 -> 67.8Improved
zero-shot: fine-tune averaged MHR-mu adapters for k=1000 steps vs k=0 (ablation)MHR-mu · T0 zero-shot (T5-XL-LM)avg accuracy: 48.0 -> 58.0Improved

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