Yuanyuan Zhao — RL for LLM tool use and text reranking; early-stage profile from two 2026 arXiv preprints
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-02) 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 score yet: this axis would measure publication volume, venue quality, and yearly productivity, but the data linked to this name is too thin — only two cataloged 2026 arXiv papers — to compute an output score.
Impact—
Not scored: impact would capture citation counts, h-index, and influence on follow-up work. No citation data is available for this name's records, and both papers are recent 2026 preprints, so impact remains unmeasured.
Focus0/100
Scores 0/100. The underlying field distribution splits this name's cataloged papers exactly evenly — 50% machine learning (LLM tool learning / reinforcement learning for agents) and 50% information retrieval / natural language processing — across two 2026 arXiv papers. A perfectly balanced two-field spread naturally yields a minimal concentration score; with only two papers, it signals an emerging, still-undifferentiated agenda rather than confirmed lack of direction.
Collaboration48/100
Scores 48/100, computed from 10 distinct coauthors across the two cataloged papers. The frequent-coauthor list shows Lijun Liu, Zhengzong Chen, and Fei Huang recurring, alongside Shihao Liu, Hao Yin, and Wenyan Li, pointing to a compact, stable team. Zhao sits in a middle author position on both papers and is not corresponding author on either, consistent with a contributing rather than leading role.
Activity—
Not scored: this axis would track publishing cadence, recency, and sustained output over time. With only two 2026 papers on record and no multi-year publication history, activity cannot be estimated yet.
Reproducibility38/100
Averages 38/100 across two scored papers (reproPapers: 2). Evolve rates MATCH medium readiness — strong ablations, but single-run results without variance tests — and MERIT-Rank low, with efficiency claims and reproducibility flagged for scrutiny. The pattern suggests competent experimental design whose reporting does not yet meet variance-aware, fully reproducible standards.
Research summary
Yuanyuan Zhao is an early-stage researcher working at the intersection of reinforcement learning for large language models and information retrieval. The visible record consists of two 2026 arXiv preprints in the Evolve catalog. MATCH couples gated hierarchical rewards with a policy-tracking curriculum for LLM tool learning; MERIT-Rank (Think Thrice Before Reranking) integrates multi-perspective evidence and staged reinforcement learning to lift LLM text reranking. In both, Zhao holds a middle author position and is not the corresponding author, consistent with a contributing team member rather than a project lead. Cataloged output splits evenly across two fields — 50% machine learning (LLM tool learning / RL for agents) and 50% information retrieval / natural language processing — which surfaces as a focus score of 0/100. Collaboration scores 48/100 across 10 distinct coauthors, with Lijun Liu, Zhengzong Chen, and Fei Huang recurring on the frequent-coauthor list, pointing to a small stable group. Reproducibility is the clearest quantitative signal, averaging 38/100 over two scored papers: MATCH is rated medium (strong ablations, single-run results without variance tests) and MERIT-Rank low (efficiency and reproducibility claims needing scrutiny). Output, impact, and activity axes remain unscored for lack of citation and productivity data.
Collaboration pattern
Zhao's collaboration footprint is compact: 10 distinct coauthors across two papers, with Lijun Liu, Zhengzong Chen, and Fei Huang recurring on the frequent-coauthor list, alongside Shihao Liu, Hao Yin, and Wenyan Li. Both papers carry a middle author position with no corresponding-author role, suggesting Zhao contributes within an established group rather than leading an independent line. The recurring trio hints at a stable core team spanning both the tool-learning and reranking projects.
Research keywords
Highlights
MATCH: curriculum-scheduled RL for LLM tool learning
MERIT-Rank: multi-perspective reasoning for text reranking
Timeline
- 2026MATCH preprint posted to arXiv: curriculum scheduling with hierarchically gated rewards for RL-based LLM tool learning, with Zhao in a middle, non-corresponding author position.
- 2026Second cataloged paper, MERIT-Rank (Think Thrice Before Reranking), extends reinforcement learning to multi-perspective text reranking evaluated on BRIGHT/TREC/BEIR, again as a middle author.
- 2026The entire visible record — two papers and 10 distinct coauthors — is concentrated within 2026, marking the start of this profile's coverage window.
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.
- The name 'Yuanyuan Zhao' is common; this profile is name-resolved from public bibliographic data, with no ORCID or institutional affiliation attached, so identity confidence is limited.
- All bibliometrics (coauthor counts, field distribution, radar scores) come from Evolve's bibliometric snapshot covering only the two cataloged 2026 arXiv papers.
- Output, impact, and activity axes are unscored because citation and productivity data are insufficient; no h-index or citation counts are reported anywhere in this profile.
- Both papers are arXiv preprints as of the snapshot date; no peer-reviewed venue status is recorded for them.