LLM agent researcher at Zhejiang University focused on evaluating and compressing self-evolving agents
2
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
No publication-volume data is available yet, so this axis is unscored. It would measure total output — number of papers over time, venue types, and authorship roles — none of which can be estimated from the current snapshot.
Impact—
Citation-based impact cannot be scored: no citation counts or h-index appear in the available data. This axis would measure scholarly influence — citations, citing works, and uptake of the agent-evaluation benchmarks — and remains unestimated here.
Focus100/100
The focus score of 100 reflects an exceptionally narrow indexed footprint: the field distribution records a single field — artificial intelligence (LLM agent evaluation) — and both cataloged 2026 papers sit inside the same self-evolving LLM agent space, one on process-level benchmarking, one on skill compression. This concentration signals a clearly defined research identity around self-evolving agents, though it rests on a very small indexed sample.
Collaboration46/100
The collaboration score of 46 sits in the mid-range, built on 9 distinct coauthors in the available data. The recurring collaborator list includes Xiaofan Bai, Chao Liu, Xuan Jin, Xipeng Cao, Yuhong Li, Di Wu, Mingli Song, and Nenggan Zheng, several of whom share the Zhejiang University affiliation. Authorship placement — first author on the evaluation paper, middle author on the compression paper — suggests both independent leadership and team-based contribution.
Activity—
Recent-activity data is insufficient to score this axis. It would measure publishing cadence — papers per recent year and gaps between outputs. Only two 2026 papers are known, so any rhythm beyond that single year cannot be inferred.
Reproducibility100/100
The reproducibility score of 100 is computed as a mean over a single paper (reproPapers: 1), so it should be read narrowly. That paper — the first-authored EvoPathBench evaluation study — carries a high reproducibility readiness rating, and the Evolve review praises it as rigorous and honest about negative results. The second paper, SkillZip Pro, is rated medium because its evidence rests on small, private evaluations.
Research summary
Hongqiang Lin is a researcher at Zhejiang University working on large language model (LLM) agents, with an emerging but sharply defined niche in self-evolving agents. The Evolve catalog holds two 2026 arXiv papers. The first, first-authored, is 'Beyond Endpoint Performance: Process-Level Evaluation of Self-Evolving Agents,' which introduces EvoPathBench, scoring self-evolving agents at intermediate checkpoints rather than only final endpoints; Evolve's review calls it rigorous and honest about negative results while noting single-domain scope and missing error bars. The second, middle-authored, is 'SkillZip Pro,' proposing execution-aware, evaluation-free compression of progressively loaded agent skill bundles with a claimed 38% size reduction without quality loss, though the review flags that evidence rests on small, private evaluations. Bibliometrically this is an early-stage profile: focus scores 100 on a field distribution concentrated in one field (LLM agent evaluation), reproducibility scores 100 over a single paper, and collaboration scores 46 across 9 distinct coauthors. Output, impact, and activity lack sufficient data and remain unscored. Recurring co-authors include Mingli Song and Nenggan Zheng. The overall picture is a focused agent researcher whose evaluation tooling shows high reproducibility readiness, with generality still to be demonstrated beyond one domain.
Collaboration pattern
The data records 9 distinct coauthors, giving a mid-range collaboration score of 46. The recurring collaborator list includes Xiaofan Bai, Chao Liu, Xuan Jin, Xipeng Cao, Yuhong Li, Di Wu, Mingli Song, and Nenggan Zheng — a mix consistent with a Zhejiang University group. Authorship placement spans first author on the evaluation benchmark paper and middle author on the systems paper, suggesting a pattern of leading evaluation work while contributing within larger engineering teams.
Research keywords
Highlights
EvoPathBench: process-level checkpoint evaluation for self-evolving agents
SkillZip Pro: claimed 38% skill-bundle compression without quality loss
Timeline
- 2026First-author arXiv paper 'Beyond Endpoint Performance' introduces EvoPathBench, process-level checkpoint evaluation for self-evolving agents.
- 2026Co-authors 'SkillZip Pro', proposing execution-aware, evaluation-free compression of progressively loaded agent skills, with a claimed 38% bundle-size reduction.
- 2026Evolve catalog profile consolidates around self-evolving LLM agents, spanning both evaluation methodology and systems/efficiency work in a single year.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric scores come from a limited OpenAlex-based snapshot; output, impact, and activity axes are unscored due to insufficient data.
- Only two reviewed arXiv papers (both 2026) underpin this profile; no journal or conference publications are in scope.
- No citation counts, h-index, or publication-volume statistics appear in the input data, so influence cannot be assessed.
- Hongqiang Lin is a common Chinese name; this profile is name-resolved from public bibliographic data, with no ORCID or verified identity linkage provided.