Researcher on multimodal vision-language models and real-time video generation at Capital Medical University.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
1
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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—
The output dimension is not yet scored: a full disambiguated publication record is unavailable. It would measure total and per-year publication volume and authorship roles; only partial signals exist (19 recent-3y works, 17 indexed in 2026), so no score is estimated.
Impact—
Impact is not scored yet: citation data for this record is incomplete. The axis would measure citations, h-index trajectory, and influence on follow-up work. Available fragments — h-index 1 and i10-index 0 — indicate an early footprint, but no score is estimated.
Focus0/100
Focus scores 0 on the radar, indicating dispersed topical concentration. The OpenAlex field distribution splits evenly — 50% machine learning (multimodal vision-language models) and 50% computer vision (video generation) — and the topic profile spans five clusters: multimodal ML applications (13), GANs and image synthesis (7), advanced neural network applications (5), reinforcement learning in robotics (5), and robot manipulation and learning (4). Breadth is clear, but a consolidated single-line agenda has not yet emerged.
Collaboration75/100
Collaboration scores 75. The record lists 43 distinct coauthors, and the frequent-collaborator set includes Kaixiang Yao, Xu Wang, Yongliang Shen, Wenqi Zhang, Daniel Dahlmeier, Weishi Wang, Jintao Zhang, Kai Jiang, Xuhong Zhang, Miao Pan, and Hu Xiyue. Both catalog papers place this researcher in a middle author position rather than first or corresponding, consistent with contribution to larger team-based model-building projects rather than solo-led studies.
Activity87/100
Activity scores 87, the strongest axis in this profile. OpenAlex countsByYear shows 2 works in 2025 rising to 17 in 2026 — a steep year-over-year jump — and 19 works within the recent three-year window. This cadence matches heavy arXiv preprint-era output, though the 2026 figures include preprints whose peer-review status is not recorded in the current data.
Reproducibility38/100
Reproducibility scores 38 (mean over 2 assessed papers). The two papers diverge: Spatial-Interactor reaches medium readiness — its interaction-trajectory setup references an LSI-108K dataset and SFT+OPD training that outside teams could attempt, though Evolve notes generic spatial-QA SFT drives most ablated gains. Vidu S2 rates low: Evolve flags no ablations, no significance tests, and no released code. Independent replication would currently be difficult on average.
Research summary
Jintao Chen is an early-stage researcher at Capital Medical University whose indexed work centers on machine learning — split evenly (50/50 in the OpenAlex field distribution) between multimodal vision-language models and computer vision for video generation. The record is young and accelerating: OpenAlex indexes 19 works over the recent three-year window, with annual output rising from 2 publications in 2025 to 17 in 2026, while the citation footprint stays minimal (h-index 1, i10-index 0). Two arXiv papers reviewed by Evolve illustrate the range of involvement. Spatial-Interactor (2026) trains vision-language models for spatial reasoning through interaction trajectories, built on an LSI-108K dataset with SFT plus on-policy distillation (OPD, GRPO); Evolve notes the paper's own ablations attribute most gains to generic spatial-QA supervised fine-tuning. Vidu S2 (2026) targets real-time 720p interactive avatars and streaming video editing through autoregressive diffusion, Self-Replay Forcing, and distribution matching distillation, though it reports no ablations, significance tests, or code. The radar places activity at 87 and collaboration at 75, reflecting 43 distinct coauthors, while focus scores 0 — the portfolio spans five topic clusters from multimodal learning to reinforcement learning in robotics — and reproducibility sits at 38 (mean over 2 assessed papers). Evidence depth is limited, so treat this as an early-stage profile.
Collaboration pattern
With 43 distinct coauthors and a collaboration score of 75, this researcher works in sizeable teams. Recurring names include Kaixiang Yao, Xu Wang, Yongliang Shen, Wenqi Zhang, Daniel Dahlmeier, Weishi Wang, and Hu Xiyue. Both catalog papers carry a middle author position, suggesting contribution within larger engineering-driven groups rather than first- or corresponding-author leadership. Coauthor lists built from bibliographic data may include homonyms, so individual edges should be verified before use.
Research keywords
Highlights
Interaction-trajectory training for VLM spatial reasoning (Spatial-Interactor)
Real-time interactive and editable video generation (Vidu S2)
Timeline
- 2025OpenAlex records the first indexed works for this profile: 2 publications appear in 2025.
- 2026Output accelerates sharply to 17 indexed works in 2026, up from 2 in 2025.
- 2026Contributes as a middle author to the arXiv paper 'Spatial-Interactor', on interaction-trajectory training (LSI-108K; SFT+OPD) for vision-language spatial reasoning.
- 2026Contributes as a middle author to the arXiv paper 'Vidu S2', on real-time 720p interactive avatars and streaming, editable video generation.
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 indicators are computed from OpenAlex only (identity: openalex); no ORCID link was available for disambiguation.
- Output and impact dimensions are unscored due to insufficient record/citation data; the only signals present are h-index 1 and i10-index 0.
- The indexed topic mix (robotics, reinforcement learning) diverges from the listed medical-university affiliation; 'Jintao Chen' is a common name, so this profile may blend homonymous records and is name-resolved from public bibliographic data.
- The reproducibility mean (38) and deep assessments rest on only 2 Evolve-reviewed papers; other indexed works are not individually assessed.