--- title: "Pengchuan Zhang — author profile and papers" description: "Pengchuan Zhang: 2 papers on Elamur Research (2020–2023), h-index 31 — affiliations, topics, co-authors and who cites them in our library." canonical: "https://research.elamur.ai/free/en/authors/pengchuan-zhang/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/pengchuan-zhang/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / Pengchuan Zhang # Pengchuan Zhang Pengchuan Zhang works on vision-language modelling, where images and text are handled within a single learned system. The two papers here, from 2020 to 2023, ask how a model can connect what appears in a picture with the words that describe it, and how one model can serve many visual tasks at once. Both build on Transformer-based language models, in the areas of [Vision & language](https://md.research.elamur.ai/free/en/topics/multimodal/) and [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/). The most cited work here is [Oscar: Object-Semantics Aligned Pre-training](https://md.research.elamur.ai/free/en/papers/li-2020-oscar-object-semantics-aligned-pre/) (2020). It tackles a weakness of earlier pre-training, which left self-attention to discover on its own which image regions match which words. Oscar adds the object labels predicted by a detector as an extra input. Because those labels are ordinary words, a BERT-initialized model can tie them to the caption, and they then serve as anchors between visual regions and language. The later paper, [MiniGPT-v2](https://md.research.elamur.ai/free/en/papers/chen-2023-minigpt-v2-large-language-model/) (2023), turns to a large language model as a shared interface for many tasks. Its starting point is that a single request, such as locating a person, can call for a box, a description or a full list of detections. The model feeds merged tokens from a frozen image encoder into LLaMA2-chat and writes box coordinates directly as text. Taken together, the papers here move from aligning image and text during pre-training toward letting one conversational language model handle a range of grounded visual tasks. Zhang's most recent affiliation is [Meta](https://md.research.elamur.ai/free/en/companies/meta/) (2023–2024). Most of the bylines from 2018 to 2023 list [Microsoft](https://md.research.elamur.ai/free/en/companies/microsoft/), and earlier papers list the [California Institute of Technology](https://md.research.elamur.ai/free/en/universities/california-institute-of-technology/). Co-authors include [Jianfeng Gao](https://md.research.elamur.ai/free/en/authors/jianfeng-gao/), [Lijuan Wang](https://md.research.elamur.ai/free/en/authors/lijuan-wang/), [Mohamed Elhoseiny](https://md.research.elamur.ai/free/en/authors/mohamed-elhoseiny/) and [Deyao Zhu](https://md.research.elamur.ai/free/en/authors/deyao-zhu/). Pengchuan Zhang has 2 papers in our library, in 2020–2023, most of them on [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/), [Vision & language](https://md.research.elamur.ai/free/en/topics/multimodal/); often with [Chunyuan Li](https://md.research.elamur.ai/free/en/authors/chunyuan-li/), [Deyao Zhu](https://md.research.elamur.ai/free/en/authors/deyao-zhu/), [Furu Wei](https://md.research.elamur.ai/free/en/authors/furu-wei/). 2 papers in our library · 2020–2023 [OpenAlex](https://openalex.org/A5059735251) · [ORCID](https://orcid.org/0000-0003-1155-9507) · [**h-index 31**](https://openalex.org/authors/A5059735251) · [70 works](https://openalex.org/works?filter=author.id:A5059735251) · [10 125 citations](https://openalex.org/works?filter=author.id:A5059735251&sort=cited_by_count:desc) OpenAlex, 2026-10-14 [2 works in our library](https://md.research.elamur.ai/free/en/authors/pengchuan-zhang/#papers) · [410 citations in our library](https://md.research.elamur.ai/free/en/authors/pengchuan-zhang/cited-by/) · [⭐ Paper Star rank #1180](https://md.research.elamur.ai/free/en/paper-stars/page/12/#star-1180) The profile is the library's match of the name; two people with one name are not always told apart. Elsewhere: [Google Scholar](https://scholar.google.com/citations?view_op=search_authors&mauthors=Pengchuan+Zhang) · [Semantic Scholar](https://www.semanticscholar.org/search?q=Pengchuan+Zhang) · [dblp](https://dblp.org/search?q=Pengchuan+Zhang) · [ACL Anthology](https://aclanthology.org/search/?q=Pengchuan+Zhang) ## Affiliations - [Meta (United States)](https://md.research.elamur.ai/free/en/companies/meta/) · Menlo Park, United States —2023–2024 on 7 papers - [Microsoft (United States)](https://md.research.elamur.ai/free/en/companies/microsoft/) · Redmond, United States —2018–2023 on 35 papers - [California Institute of Technology](https://md.research.elamur.ai/free/en/universities/california-institute-of-technology/) · Pasadena, United States —2016–2021 on 7 papers From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 5 more institutions on single bylines are not listed. ## Topics [Language models & pretraining 2](https://md.research.elamur.ai/free/en/topics/language-models/) · [Vision & language 2](https://md.research.elamur.ai/free/en/topics/multimodal/) ## Works by year ### 2020s [**2023** 1 paper](https://md.research.elamur.ai/free/en/papers/chen-2023-minigpt-v2-large-language-model/) · [**2020** 1 paper](https://md.research.elamur.ai/free/en/papers/li-2020-oscar-object-semantics-aligned-pre/) ## Papers The most cited first (OpenAlex and our library). 1. **[Oscar: Object-Semantics Aligned Pre-training for Vision-Language Tasks](https://md.research.elamur.ai/free/en/papers/li-2020-oscar-object-semantics-aligned-pre/)** *(2020 · 1 827 citations)* Vision-language pre-training (VLP) models usually feed image region features and text tokens together into a Transformer. 2. **[MiniGPT-v2: large language model as a unified interface for vision-language multi-task learning](https://md.research.elamur.ai/free/en/papers/chen-2023-minigpt-v2-large-language-model/)** *(2023 · 192 citations)* The paper asks whether one vision-language model can handle many different tasks without confusing them. The same request, such as asking where a person is, may call for a bounding box, a verbal description or a list of all detections. MiniGPT-v2 addresses this ambiguity. A frozen EVA ViT encoder processes 448x448… ## Co-authors here 12 of 16 co-authors with a page, the most papers together first. - [Chunyuan Li](https://md.research.elamur.ai/free/en/authors/chunyuan-li/) - [Deyao Zhu](https://md.research.elamur.ai/free/en/authors/deyao-zhu/) - [Furu Wei](https://md.research.elamur.ai/free/en/authors/furu-wei/) - [Jianfeng Gao](https://md.research.elamur.ai/free/en/authors/jianfeng-gao/) - [Jun Chen](https://md.research.elamur.ai/free/en/authors/jun-chen/) - [Lei Zhang](https://md.research.elamur.ai/free/en/authors/lei-zhang/) - [Li Dong](https://md.research.elamur.ai/free/en/authors/li-dong/) - [Lijuan Wang](https://md.research.elamur.ai/free/en/authors/lijuan-wang/) - [Mohamed Elhoseiny](https://md.research.elamur.ai/free/en/authors/mohamed-elhoseiny/) - [Raghuraman Krishnamoorthi](https://md.research.elamur.ai/free/en/authors/raghuraman-krishnamoorthi/) - [Vikas Chandra](https://md.research.elamur.ai/free/en/authors/vikas-chandra/) - [Xiang Li](https://md.research.elamur.ai/free/en/authors/xiang-li/) ## Cited by Papers of our library cite their works 410 times (of the 140 884 papers whose bibliographies we read); [28 citing papers with a page here →](https://md.research.elamur.ai/free/en/authors/pengchuan-zhang/cited-by/).