LaneRoPE: Positional Encoding for Collaborative Parallel Reasoning and Generation
LaneRoPE enables multiple LLM sequences to collaborate during generation via inter-sequence attention and extended RoPE, improving accuracy on math reasoning tasks with minimal architectural changes and negligible inference overhead.
Article intelligence
Key points
- Introduces inter-sequence attention mask to make sequence sampling dependent.
- Extends RoPE to capture relative positions both within and across sequences.
- Achieves accuracy gains on math reasoning under limited generation length.
- Requires minimal changes to LLM architecture and negligible overhead.
Why it matters
This matters because introduces inter-sequence attention mask to make sequence sampling dependent.
Technical impact
May affect model selection, inference cost, product capability, and evaluation benchmarks.
[2605.27570] LaneRoPE: Positional Encoding for Collaborative Parallel Reasoning and Generation
[Submitted on 26 May 2026]
Title:LaneRoPE: Positional Encoding for Collaborative Parallel Reasoning and Generation
View a PDF of the paper titled LaneRoPE: Positional Encoding for Collaborative Parallel Reasoning and Generation, by Gabriele Cesa and 5 other authors
View PDF HTML (experimental)
Abstract:Parallel LLM test-time scaling techniques (e.g., best-of-$N$) require drawing $N>1$ sequences conditioned on the same input prompt. These methods boost accuracy while exploiting the computational efficiency of batching $N$ generations. However, each sequence in the batch is traditionally generated independently and hence does not reuse intermediate generations, computations, or observations from other sequences. In this paper, we propose LaneRoPE to enable coordination and collaboration among $N>1$ sequences at generation time. LaneRoPE involves two key ideas: (a) an inter-sequence attention mask to make sampling of sequences dependent on one another; and (b) a RoPE extension that injects positional information that captures relative positions between tokens, both within and outside a particular sequence. We evaluate our approach on mathematical reasoning tasks and find promising results: LaneRoPE enables collaboration among sequences, yielding additional accuracy gains under limited generated sequence length. Importantly, since LaneRoPE enables coordination with minimal changes to the underlying LLM architecture and introduces a negligible overhead at inference time, it is appealing to rapidly incorporate parallel reasoning into existing LLM inference pipelines.
Subjects:
Artificial Intelligence (cs.AI)
Cite as: arXiv:2605.27570 [cs.AI]
(or arXiv:2605.27570v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2605.27570
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Gabriele Cesa [view email] [v1] Tue, 26 May 2026 18:43:15 UTC (631 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled LaneRoPE: Positional Encoding for Collaborative Parallel Reasoning and Generation, by Gabriele Cesa and 5 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.AI
new | recent | 2026-05
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?)
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?)