AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Pyramid Attention For Source Code Summarization

Lei Chai, Ming Li

NeurIPS 2022 · Architectures and efficiency › Attention and transformers · approach: Long-context KV and memory attention · 0 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 multi-scale pyramid processing instead of single concatenated sequence (ablation)PA-former · EMSE-DeepCom testBLEU: 37.329 -> 38.848Improved
add bottom-to-top aggregation (ablation)PA-former · EMSE-DeepCom testBLEU: 37.532 -> 38.848Improved
add cross-granularity interaction (ablation)PA-former · EMSE-DeepCom testBLEU: 37.133 -> 38.848Improved
add granular-aware relative embedding (ablation)PA-former · EMSE-DeepCom testBLEU: 36.905 -> 38.848Improved
decode from fine-grained representation only rather than serial use of all granularities (ablation)PA-former · EMSE-DeepCom testBLEU: 38.729 -> 38.848Improved
decode from fine-grained representation only rather than concatenating all granularities (ablation)PA-former · EMSE-DeepCom testBLEU: 38.556 -> 38.848Improved
add grammar-type content to semi- and coarse-grained pyramid input (ablation)PA-former · EMSE-DeepCom testBLEU: 38.394 -> 38.848Improved
add grammar-type content to pyramid input (ablation)PA-former · EMSE-DeepCom testRouge-L: 55.744 -> 56.095Improved
add multi-scale vs single-scale input (ablation)PA-former · EMSE-DeepCom testRouge-L: 55.032 -> 56.095Improved
replace TPTrans/CAST-style full-AST single-granularity models with PA-formerPA-former · EMSE-DeepCom testBLEU: 37.248 -> 38.848Improved
replace CAST with PA-former on FunComPA-former · FunCom testBLEU: 43.580 -> 44.649Improved
PA-former vs NeuralCodeSum in human evaluationPA-former · Human evalnaturalness: 5.12 -> 6.09Improved

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