AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing

Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le

NeurIPS 2020 · Architectures and efficiency › Attention and transformers · approach: Efficient language transformers

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
trade sequence length for additional encoder depthFunnel-Transformer versus L12H768 · GLUE-AVG dev, base-scale MLMscore: 84.4 -> 85.3Improved
use Funnel-Transformer compressionFunnel-Transformer versus MPNet Base · GLUE-AVG dev, large-scale ELECTRAscore: 87.7 -> 88.3Improved
use Funnel-Transformer compressionFunnel-Transformer versus MPNet Base · RACE testaccuracy: 72.0 -> 79.7Improved
use Funnel-Transformer with decoderFunnel-Transformer versus ELECTRA Large · SQuAD2.0 devEM: 88.0 -> 87.6Hurt
remove the separate, unpooled [cls] state (ablation)Funnel-Transformer · GLUE-AVG devscore: 83.5 -> 82.9Hurt
remove pool-query-only attention by using pooled states for keys and values too (ablation)Funnel-Transformer · GLUE-AVG devscore: 83.5 -> 83.0Hurt
replace relative attention with absolute position embeddings (ablation)Funnel-Transformer · GLUE-AVG devscore: 83.5 -> 81.4Hurt
replace mean pooling with max pooling (ablation)Funnel-Transformer · GLUE-AVG devscore: 83.5 -> 83.4No clear effect
replace mean pooling with Top-Attn pooling (ablation)Funnel-Transformer · GLUE-AVG devscore: 83.5 -> 75.8Hurt
use a 2-block layout instead of 3 blocks (ablation)Funnel-Transformer · GLUE-AVG devscore: 83.5 -> 83.4Hurt
use a 4-block layout instead of 3 blocks (ablation)Funnel-Transformer · GLUE-AVG devscore: 83.5 -> 82.9Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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