AI researcher working on autonomous LLM research agents and human-aware language models, with two 2026 arXiv papers.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-20) 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-count or output metric is available yet in the bibliometric record, so this axis is unscored. It would measure total research output over time, such as papers per year across publication venues.
Impact—
No citation counts, h-index, or influence metrics are present in the input data, so impact cannot be scored. This axis would measure citation-based influence and uptake of the researcher's work by the wider community.
Focus0/100
The focus score of 0 reflects the bibliometric field distribution rather than thin output: the reviewed work splits evenly at 0.5 natural language processing and 0.5 artificial intelligence (autonomous LLM research agents). The two 2026 papers sit in different fields, one on effort allocation for autonomous research agents and one on human-aware dialogue modeling, so attention is currently divided across two lines rather than concentrated in a single specialization.
Collaboration66/100
Collaboration scores 66 out of 100, based on 26 distinct coauthors in the bibliometric record, with 12 recurring collaborators including Xinle Yu, Zhen Wang, and Zixuan Wang. Both reviewed 2026 papers are middle-author, non-corresponding contributions, indicating work within multi-author teams. The pattern suggests active but distributed collaboration: a moderately large coauthor network with a core group, rather than solo work or a single dominant partnership.
Activity—
The bibliometric record contains no publication-frequency or year-by-year activity data, so this axis is unscored. It would measure recent publishing cadence and ongoing engagement with the research community over time.
Reproducibility38/100
Reproducibility scores 38 out of 100, computed as a mean of 38 across the 2 reviewed papers carrying reproducibility assessments. PrimeScientist is rated medium readiness, while Mind2Dialogue is rated low, largely because its personalization gains rest on a closed teacher model that others cannot easily access. The below-midpoint score signals that independent replication of both research lines would currently require author support or additional released artifacts.
Research summary
Fan Bai is an early-stage AI researcher whose visible output in the Evolve catalog centers on two 2026 arXiv preprints that sit at the intersection of large language model agents and natural language processing. The first, PrimeScientist, addresses strategic allocation of research effort in autonomous research, using budget-adaptive Monte Carlo tree search over executable plan trees; Evolve's review judges the attempt-efficiency gains solid while noting that headline reward gains rest on single runs. The second, Mind2Dialogue, trains human-aware language models by distilling an Oracle that sees simulated user mental states; personalization gains hold across three backbones but depend on a closed teacher model. The bibliometric record shows the work evenly split between natural language processing and artificial intelligence for autonomous LLM research agents (0.5/0.5 field distribution), a collaboration score of 66 built on 26 distinct coauthors, and a reproducibility mean of 38 across the two reviewed papers. Both catalog entries list Fan Bai as a middle-position, non-corresponding author, suggesting a contributor role within larger multi-author teams rather than an independent group lead — a picture consistent with an early-career, dual-track profile.
Collaboration pattern
Fan Bai's bibliometric record lists 26 distinct coauthors, with a recurring core of 12 frequent collaborators including Xinle Yu, Zhen Wang, Zixuan Wang, Yufan Zhou, and Lyumanshan Ye. Both reviewed 2026 papers are middle-author, non-corresponding contributions, so the pattern points to collaborative work inside multi-author teams spanning two fields — autonomous LLM research agents and natural language processing — rather than solo authorship or a single dominant partnership.
Research keywords
Highlights
Budget-adaptive MCTS allocates research effort in autonomous science agents
Human-aware language models trained from a simulated-user Oracle
Timeline
- 2026PrimeScientist appears on arXiv, proposing budget-adaptive Monte Carlo tree search over executable plan trees to allocate research effort in autonomous research.
- 2026Mind2Dialogue appears on arXiv, training human-aware language models from an Oracle that sees simulated user mental states.
- 2026Both papers are reviewed into the Evolve catalog with reproducibility readiness of medium (PrimeScientist) and low (Mind2Dialogue), forming a dual-field profile across autonomous research agents and natural language processing.
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; 'Fan Bai' is a relatively common name, so the record may aggregate homonymous authors.
- Institution, country, and ORCID are absent from the input data, so affiliation and identity cannot be verified.
- No citation, output, or activity metrics were available, so the impact, output, and activity radar axes remain unscored.
- The profile rests on only two reviewed papers, both 2026 arXiv preprints; peer-reviewed venue placement is not yet represented.