Pith. sign in

REVIEW 5 cited by

Multi-Head Attention: Collaborate Instead of Concatenate

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 2006.16362 v2 pith:3EPBZXRX submitted 2020-06-29 cs.LG cs.CLstat.ML

Multi-Head Attention: Collaborate Instead of Concatenate

classification cs.LG cs.CLstat.ML
keywords attentionheadslayermulti-headcollaborativeprojectionsqueryaccuracy
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Attention layers are widely used in natural language processing (NLP) and are beginning to influence computer vision architectures. Training very large transformer models allowed significant improvement in both fields, but once trained, these networks show symptoms of over-parameterization. For instance, it is known that many attention heads can be pruned without impacting accuracy. This work aims to enhance current understanding on how multiple heads interact. Motivated by the observation that attention heads learn redundant key/query projections, we propose a collaborative multi-head attention layer that enables heads to learn shared projections. Our scheme decreases the number of parameters in an attention layer and can be used as a drop-in replacement in any transformer architecture. Our experiments confirm that sharing key/query dimensions can be exploited in language understanding, machine translation and vision. We also show that it is possible to re-parametrize a pre-trained multi-head attention layer into our collaborative attention layer. Collaborative multi-head attention reduces the size of the key and query projections by 4 for same accuracy and speed. Our code is public.

discussion (0)

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

Forward citations

Cited by 5 Pith papers

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

  1. Architecture, Not Scale: Circuit Localization in Large Language Models

    cs.CL 2026-05 unverdicted novelty 6.0

    Grouped query attention produces more concentrated and stable circuits than multi-head attention across tasks and scales in Pythia and Qwen2.5 models, with a phase transition in factual recall circuits.

  2. EfficientRollout: System-Aware Self-Speculative Decoding for RL Rollouts

    cs.LG 2026-06 unverdicted novelty 5.0

    EfficientRollout applies self-speculative decoding with quantized drafter induction and system-aware acceptance policies to cut RL rollout latency up to 19.6% while preserving final model quality.

  3. Understanding Inference Scaling for LLMs: Bottlenecks, Trade-offs, and Performance Principles

    cs.DC 2026-05 unverdicted novelty 5.0

    Reasoning workloads shift LLM inference to a capacity-bound regime where KV-cache fragmentation limits data parallelism, tensor parallelism unlocks memory at the 32B scale, and MoE models require hybrid strategies to ...

  4. Negative Before Positive: Asymmetric Valence Processing in Large Language Models

    cs.CL 2026-05 unverdicted novelty 5.0

    Negative valence localizes to early layers and positive valence to mid-to-late layers in LLMs, with the directions being causally steerable.

  5. Predicting Blastocyst Formation in IVF: Integrating DINOv2 and Attention-Based LSTM on Time-Lapse Embryo Images

    cs.CV 2026-04 unverdicted novelty 4.0

    DINOv2 plus attention LSTM predicts blastocyst formation from limited daily time-lapse images at 96.4% accuracy on 704 embryo videos.