Paper Detail

Target-Oriented Pretraining Data Selection via Neuron-Activated Graph

Zijun Wang, Haoqin Tu, Weidong Zhou, Yiyang Zhou, Xiaohuan Zhou, Bingni Zhang, Weiguo Feng, Taifeng Wang, Cihang Xie, Fengze Liu

huggingface Score 4.5

Published 2026-04-17 · First seen 2026-04-22

General AI

Abstract

Everyday tasks come with a target, and pretraining models around this target is what turns them into experts. In this paper, we study target-oriented language model (LM) pretraining by introducing Neuron-Activated Graph Ranking (NAG-based Ranking), a training-free and interpretable framework for target pretraining data selection. Rather than using black-box representations, our approach directly characterizes each target input by a sparse set of high-impact neurons in any off-the-shelf LLMs. Concretely, we quantify neuron impact and select the most influential neurons across layers into a compact Neuron-Activated Graph (NAG), and rank candidate data by NAG similarity to target examples. We conduct experiments across six benchmarks, where our NAG-based Ranking improves target-oriented pretraining by 4.9% on average over random sampling, and also outperforms state-of-the-art baselines by 5.3% accuracy on HellaSwag. It also remains effective under a more applicable multi-target setting, where our best setup surpasses two baselines by 1.1% and 4.1%, respectively. Furthermore, we provide a comprehensive analysis on why and how our NAG works, e.g., deactivating NAG-selected neurons (only 0.12% of all) causes a 23.5% performance collapse, and restricting NAG to the final layer incurs a 4.1% average drop, indicating that NAG captures a sparse "functional backbone" for learning target features. We release the code at https://github.com/asillycat/NAG.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
later
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

@misc{wang2026target,
  title = {Target-Oriented Pretraining Data Selection via Neuron-Activated Graph},
  author = {Zijun Wang and Haoqin Tu and Weidong Zhou and Yiyang Zhou and Xiaohuan Zhou and Bingni Zhang and Weiguo Feng and Taifeng Wang and Cihang Xie and Fengze Liu},
  year = {2026},
  abstract = {Everyday tasks come with a target, and pretraining models around this target is what turns them into experts. In this paper, we study target-oriented language model (LM) pretraining by introducing Neuron-Activated Graph Ranking (NAG-based Ranking), a training-free and interpretable framework for target pretraining data selection. Rather than using black-box representations, our approach directly characterizes each target input by a sparse set of high-impact neurons in any off-the-shelf LLMs. Con},
  url = {https://huggingface.co/papers/2604.15706},
  keywords = {target-oriented language model pretraining, Neuron-Activated Graph Ranking, neuron impact, Neuron-Activated Graph, data selection, sparse set of high-impact neurons, off-the-shelf LLMs, NAG similarity, target examples, multi-target setting, functional backbone, code available, huggingface daily},
  eprint = {2604.15706},
  archiveprefix = {arXiv},
}

Metadata

{}