AI News HubLIVE
Original source2 min read

Prism Transformer: Progressive Head Schedules for Hierarchical Attention Processing

A new Transformer architecture called Prism Transformer replaces uniform head allocation with a progressive head schedule, increasing head count per layer to create a local-to-global representational hierarchy. It achieves consistent improvements on multiple zero-shot benchmarks without extra parameters or compute.

SourcearXiv Machine LearningAuthor: Shubham Aggarwal

[2606.27449] Prism Transformer: Progressive Head Schedules for Hierarchical Attention Processing

[Submitted on 25 Jun 2026]

Title:Prism Transformer: Progressive Head Schedules for Hierarchical Attention Processing

View a PDF of the paper titled Prism Transformer: Progressive Head Schedules for Hierarchical Attention Processing, by Shubham Aggarwal

View PDF HTML (experimental)

Abstract:Multi-head attention conventionally partitions the hidden dimension equally across all heads at every layer, enforcing an identical representational subspace dimension (dh = dmodel/h) throughout the models depth. In this work, we identify this uniform allocation as a fundamental structural bottleneck: due to their restricted dimensional space, early-layer heads are unable to faithfully capture complex, high-dimensional contextual patterns. To resolve this, we introduce the Prism Transformer, a novel architectural paradigm that replaces the static, uniform head configuration with a progressive head schedule. By monotonically increasing the head count across layers, the Prism Transformer naturally establishes a local-to-global representational hierarchy: early layers leverage fewer, exceptionally wide heads to capture complex, local compositional patterns, while deep layers deploy many, narrow heads to decompose these patterns into specialized linguistic features. Crucially, this structural shift is parameter-neutral, compute-neutral, and introduces zero training or inference overhead, preserving identical weight matrices and FLOP budgets as the standard Transformer. Across three model scales (124M, 354M, and 757M), the Prism Transformer consistently outperforms uniform baselines, achieving consistent reductions in validation loss alongside consistent gains on downstream zero-shot benchmarks (including PIQA, HellaSwag, ARC-Easy, and WinoGrande). Our findings demonstrate that non-uniform subspace allocation unlocks latent capacity within the standard Transformer budget, enabling more effective use of model capacity.

Comments: 14 Page, 4 figure, 5 table

Subjects:

Machine Learning (cs.LG)

Cite as: arXiv:2606.27449 [cs.LG]

(or arXiv:2606.27449v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2606.27449

arXiv-issued DOI via DataCite

Submission history

From: Shubham Aggarwal [view email] [v1] Thu, 25 Jun 2026 18:18:32 UTC (1,278 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Prism Transformer: Progressive Head Schedules for Hierarchical Attention Processing, by Shubham Aggarwal

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.LG

new | recent | 2026-06

Change to browse by:

cs

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

IArxiv recommender toggle

IArxiv Recommender (What is IArxiv?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)