University of Liverpool researcher on efficient LLM multi-agent debate and order-sensitive multimodal judgment
2
papers here
2
reviewed by Evolve
1
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 output data is available yet: the catalog holds only 2 arXiv preprints, with no publication counts, venue records, or yearly totals. This axis would measure cumulative publication volume and productivity over time.
Impact—
Impact cannot be scored yet: no citation counts, h-index, or downstream uptake are available for these 2026 preprints. This axis would measure citations, influence on follow-up work, and adoption by other researchers.
Focus0/100
The focus score is 0, reflecting an even 0.5/0.5 split across two indexed fields: multimodal large language model evaluation and artificial intelligence (LLM multi-agent systems). Neither thread yet dominates. The two are adjacent — both concern how LLM-based systems behave and how to probe them — so the split reads as parallel early exploration rather than scattered work, but a single-field concentration has not yet emerged from the two-paper record.
Collaboration32/100
The collaboration score is 32, based on a small recorded footprint: 4 distinct coauthors, with Boxuan Wang, Xiaowei Huang, and Yi Dong recurring across the catalog. Authorship roles are mixed — first author on the multimodal evaluation paper and middle author on the multi-agent debate paper, with no corresponding-author record. This points to a compact, stable working group rather than a broad external network.
Activity—
Activity is not yet measurable: both known papers date to 2026, so there is no multi-year publication cadence to score. This axis would measure publishing pace, recent output, and sustained research engagement over time.
Reproducibility38/100
The reproducibility score is 38, the mean over 2 papers with Evolve repro-readiness ratings. The multi-agent debate paper is rated medium: its cost-efficiency claims are well quantified, but accuracy gains fall within seed noise and no code was released. The evidence-order paper is rated low despite careful experimental control. Neither artifact has been independently verified, so this score reflects readiness, not replication.
Research summary
Zhuoyun Li is a researcher at the University of Liverpool working on large language model systems, with an Evolve catalog that currently spans two adjacent threads: multimodal large language model evaluation and LLM multi-agent systems, each accounting for half of the indexed field distribution (0.5/0.5). The first-authored preprint 'Same evidence, different judgments: Evidence noncommutative in vision/speech-text conflicts' (2026, arXiv) reports that swapping the order of identical evidence shifts multimodal judgments toward image and speech inputs, a claim Evolve calls well controlled while noting that 'noncommutativity' slightly outruns its two-element scope. The middle-authored 'Do We Need Complex Topology Control?' (2026, arXiv) argues that distinct-peer random routing with round-3 stopping rivals complex topology control in sparse multi-agent debate, with a solid cost-efficiency story but accuracy gains within seed noise and no code released. Both papers date to 2026, making this an early-stage profile built on two preprints. The recorded collaboration footprint is small — four distinct coauthors, with Boxuan Wang, Xiaowei Huang, and Yi Dong recurring — and Evolve scores collaboration at 32/100 and reproducibility at 38/100, the mean across two papers rated medium and low for repro readiness. Output, impact, and activity lack sufficient data and are left unscored; verified code release and venue acceptance would most strengthen this record.
Collaboration pattern
The recorded footprint is a compact team: 4 distinct coauthors, with Boxuan Wang, Xiaowei Huang, and Yi Dong recurring. Roles are mixed — first author on the multimodal evaluation paper, middle author on the debate paper, and no corresponding-author record. The collaboration score of 32 reflects this small, stable circle; no cross-institution partners appear in the current snapshot.
Research keywords
Highlights
Simpler routing rivals complex topology control in sparse multi-agent debate
Evidence order alone shifts multimodal judgments
Timeline
- 2026First-authored 'Same evidence, different judgments: Evidence noncommutative in vision/speech-text conflicts' appears on arXiv, opening the multimodal LLM evaluation thread.
- 2026Co-authored 'Do We Need Complex Topology Control?' appears on arXiv, arguing distinct-peer random routing improves cost-efficiency in sparse multi-agent debate.
- 2026Both catalog preprints land in the same year, splitting the indexed field distribution evenly between multimodal LLM evaluation and LLM multi-agent systems.
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.
- Profile is name-resolved from public bibliographic data; no ORCID or country is attached, so identity cannot be fully disambiguated.
- Coverage is limited to 2 Evolve-reviewed arXiv preprints (both 2026); no citation counts, venue acceptances, or publication totals are available.
- Reproducibility score (38) averages Evolve readiness ratings for 2 papers (medium, low); code release and independent replication were not verified.
- Dimension scores come from a sparse early snapshot and may shift as more indexed output accumulates.