Pith. sign in

REVIEW 6 cited by

Are Sixteen Heads Really Better than One?

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1905.10650 v3 pith:YEQZGOES submitted 2019-05-25 cs.CL

Are Sixteen Heads Really Better than One?

classification cs.CL
keywords attentionmodelsheadsmultipleaverageevenheadparticular
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Attention is a powerful and ubiquitous mechanism for allowing neural models to focus on particular salient pieces of information by taking their weighted average when making predictions. In particular, multi-headed attention is a driving force behind many recent state-of-the-art NLP models such as Transformer-based MT models and BERT. These models apply multiple attention mechanisms in parallel, with each attention "head" potentially focusing on different parts of the input, which makes it possible to express sophisticated functions beyond the simple weighted average. In this paper we make the surprising observation that even if models have been trained using multiple heads, in practice, a large percentage of attention heads can be removed at test time without significantly impacting performance. In fact, some layers can even be reduced to a single head. We further examine greedy algorithms for pruning down models, and the potential speed, memory efficiency, and accuracy improvements obtainable therefrom. Finally, we analyze the results with respect to which parts of the model are more reliant on having multiple heads, and provide precursory evidence that training dynamics play a role in the gains provided by multi-head attention.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. MedCore: Boundary-Preserving Medical Core Pruning for MedSAM

    cs.CV 2026-05 unverdicted novelty 7.0

    MedCore achieves 60% parameter and 58.4% FLOP reduction on MedSAM with Dice 0.9549 and preserved boundary metrics via dual-intervention pruning and a new boundary leverage principle.

  2. Voltron: Enabling Elastic Multi-Device Execution of LLM Inference for Empowered Edge Intelligence

    cs.DC 2026-07 conditional novelty 6.0

    A framework called Voltron elastically distributes LLM inference across heterogeneous edge devices using layer-wise hybrid parallelism and mixed precision, achieving up to 16.5% higher accuracy than single-device exec...

  3. HuggingFace's Transformers: State-of-the-art Natural Language Processing

    cs.CL 2019-10 accept novelty 6.0

    Hugging Face releases an open-source Python library that supplies a unified API and pretrained weights for major Transformer architectures used in natural language processing.

  4. ChessMimic: Per-Rating Transformer Models for Human Move, Clock, and Outcome Prediction in Online Blitz Chess

    cs.LG 2026-06 unverdicted novelty 5.0

    Per-100-Elo-band transformers outperform Maia-2 in move prediction accuracy across all bands and reach 0.78 AUC on outcome prediction using held-out Lichess data.

  5. AudioKV: KV Cache Eviction in Efficient Large Audio Language Models

    cs.SD 2026-04 unverdicted novelty 5.0

    AudioKV prioritizes audio-critical attention heads identified via ASR analysis and applies spectral score smoothing to evict KV cache tokens, achieving high compression with minimal accuracy loss in LALMs.

  6. Do Transformer Attention Heads Provide Transparency in Abstractive Summarization?

    cs.CL 2019-07 unverdicted novelty 5.0

    Analysis of transformer attention heads in abstractive summarization shows specialization in some heads and proposes a method to measure model reliance on learned attention distributions.