Paper Detail

Ranking and Partitioning of Attention Heads for Sparse and Continual Reinforcement Learning Fine-tuning

Ruizhe Shi, Zaiwen Yang, Qiubo Wang, Banghua Zhu, Maryam Fazel

arxiv Score 16.0

Published 2026-10-15 · First seen 2026-10-16

Research Track A · General AI

Abstract

We study sparse and continual reinforcement learning fine-tuning of language models through attention-head ranking and partitioning. For sparse learning, we rank attention heads by gradient norm to identify a small subset of heads that can be trained effectively. For continual learning, we partition heads across tasks to reduce cross-task interference and preserve previously learned capabilities. Theoretically, we relate gradient norm to training effectiveness, and further characterize why sparse, separated head updates can remain stable. Empirically, across math, instruction-following, and coding tasks, tuning higher-ranked heads consistently outperforms tuning lower-ranked ones while often matching denser baselines, and head partitioning improves retention while preserving adaptation to new tasks.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
now
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@article{shi2026ranking,
  title = {Ranking and Partitioning of Attention Heads for Sparse and Continual Reinforcement Learning Fine-tuning},
  author = {Ruizhe Shi and Zaiwen Yang and Qiubo Wang and Banghua Zhu and Maryam Fazel},
  year = {2026},
  abstract = {We study sparse and continual reinforcement learning fine-tuning of language models through attention-head ranking and partitioning. For sparse learning, we rank attention heads by gradient norm to identify a small subset of heads that can be trained effectively. For continual learning, we partition heads across tasks to reduce cross-task interference and preserve previously learned capabilities. Theoretically, we relate gradient norm to training effectiveness, and further characterize why spars},
  url = {https://arxiv.org/abs/2610.21336},
  keywords = {cs.LG, cs.CL},
  eprint = {2610.21336},
  archiveprefix = {arXiv},
}

Metadata

{}