University of Macau researcher on LLM alignment: efficient fine-tuning, multilingual reward models, and agentic memory.
3
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
20
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—
This snapshot lacks the full publication list needed to score output volume. The axis would measure cumulative and recent paper counts; only indirect signals exist — 67 works in the last three years and an h-index of 20.
Impact—
Citation totals are not available in this snapshot, so impact is unscored; the axis would measure citations, field-normalized influence and trajectory. The record lists h-index 20 and i10-index 32, but full citation aggregates are needed.
Focus0/100
The focus score is 0: no single dominant concentration. OpenAlex topics spread across topic modeling (94 works) and NLP techniques (86), then speech recognition and synthesis (31), multimodal ML (23) and translation studies (16); the current field split is exactly even, 0.5 LLM fine-tuning vs 0.5 reward modeling. Breadth is real, but concentration is low across the record.
Collaboration62/100
Collaboration scores 62. The last three years involve 22 distinct coauthors, and 12 recurring collaborators appear in the profile, including Longyue Wang, Derek F. Wong, Jinsong Su, Runzhe Zhan and Junchao Wu. On all three Evolve-reviewed papers Luo is a middle author and not corresponding, indicating contribution within larger teams rather than sole leadership.
Activity100/100
Activity is the strongest axis at 100. The OpenAlex counts-by-year series rises from 7 works (2019) to 17 (2020) and 18 (2021), dips to 4 (2022), recovers to 10 (2024), then peaks at 28 (2025) and 29 (2026); the recent three-year window holds 67 works. The late-record acceleration is steep and current.
Reproducibility55/100
Reproducibility scores 55 (reproMean 55 over 2 repro-assessed papers). All three reviewed papers rate medium readiness. Reviewers consistently flag statistical-rigor gaps: NGFT reports broad gains without significance testing, UniRRM lacks significance tests and carries LLM-verified-data circularity risk, and CHIME shows modest 2.96–3.68 point gains with no variance tests.
Research summary
Weihua Luo is a researcher at the University of Macau whose recent work centers on aligning large language models: efficient fine-tuning, reward modeling, and LLM-based agents. An OpenAlex-based snapshot records an h-index of 20, an i10-index of 32, 67 works over the most recent three-year window, and 22 distinct coauthors. The activity curve is strongly back-loaded: from single-digit annual output through the 2000s and early 2010s, recorded works climb to 17 (2020), 18 (2021), 28 (2025) and 29 (2026) — the basis for a maximum activity score of 100. Topic coverage is broad: topic modeling (94 works) and NLP techniques (86) lead, followed by speech recognition and synthesis (31), multimodal ML (23) and translation studies (16), while the current field split is exactly even between LLM fine-tuning and reward modeling — consistent with a focus score of 0. Evolve has reviewed three 2026 arXiv papers: NGFT on neuron-guided fine-tuning; UniRRM, a unified multilingual reasoning reward model built on 64.5K preferences across 103 languages; and CHIME on credit-aware hierarchical memory for long-horizon agentic planning. All three rate medium reproducibility, with reviewers flagging missing significance or variance testing; the reproducibility score is 55 across two repro-assessed papers. On the reviewed work Luo is a middle author, never corresponding, pointing to a senior contributor role within larger teams.
Collaboration pattern
The three-year window records 22 distinct coauthors, with a recurring group of 12 including Longyue Wang, Derek F. Wong, Jinsong Su, Runzhe Zhan, Junchao Wu, Shudong Liu and Shu Yang — a roster aligned with the record's NLP and translation-studies footprint. On all three Evolve-reviewed papers Luo sits in the middle author list and is never the corresponding author, fitting a senior contributor role inside larger project teams; the collaboration dimension scores 62.
Research keywords
Highlights
Neuron-guided fine-tuning unifies data selection, sparse updates and anti-forgetting
UniRRM unifies reasoning reward models across languages and judging paradigms
CHIME adds credit-aware hierarchical memory to long-horizon agentic planning
Timeline
- 2003Earliest year in the OpenAlex counts-by-year series, with 2 recorded works; output stays sparse (roughly 0–2 works per year) through 2013.
- 2019A ramp begins: 7 recorded works, rising to 17 in 2020 and 18 in 2021.
- 2022Output dips to 4 recorded works, then recovers to 10 by 2024.
- 2025A sharp surge to 28 recorded works, the previous peak of the series.
- 202629 recorded works and all three Evolve-reviewed arXiv papers appear (NGFT, UniRRM, CHIME); the recent three-year window totals 67 works.
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.
- Bibliometrics are derived from OpenAlex only (identity: openalex); citation totals and a full publication list were not provided.
- Output and impact dimensions are unscored because the underlying aggregate data is missing from the snapshot.
- All three reviewed papers are 2026 arXiv preprints with middle-author, non-corresponding placement; peer-reviewed venue information is not in the data.
- 'Weihua Luo' is a common name; this profile is name-resolved from public bibliographic data (OpenAlex, ORCID 0000-0002-5002-6933) and may not capture every work by researchers of the same name.