School of Engineering Department of Computer Science and Engineering 115 Generative AI Supervisor: CHEN Qifeng / CSE Student: LIU Tingsen / DSCT ZHAO Yuhua / COMP Course: UROP 1100, Spring We are dedicated to researching different generative models, covering the mathematical theoretical basis, practical effects, and some detailed techniques during the implementation process of generative models. Generative AI Supervisor: CHEN Qifeng / CSE Student: MO Yunxiang / COSC Course: UROP 1100, Summer This project aims to develop and debug a web-based control system for interacting with an Ubuntu environment. Through the web interface, users can issue commands such as screen capture, simulated clicks, text input, and navigation. These commands are sent to the backend via HTTP requests and executed within the Ubuntu environment, which acts as the target operating system. While certain commands, such as screenshot capture, function correctly, click-related actions failed to trigger any observable effect during testing. This report outlines the current system architecture, the debugging process undertaken to locate the fault, technical challenges encountered, and proposed solutions. The current stage of the project is focused on isolating the cause of command execution failures in the click pipeline. Generative AI Supervisor: CHEN Qifeng / CSE Student: QIN Tsz Yui / COMP Course: UROP 1000, Summer In the summer Undergraduate Research Opportunity Program 1000, I have been following Professor Chen Qifeng and PHD student Runtao Liu in conducting UROP project titled “Generative AI”, in undergoing research on generative artificial intelligence. The research topic in my group is LLM and Text to images. My main contribution lies in assisting with the research works in 4 sub-projects, including 1. SafetyDPO, 2. flow_grpo, 3. 3D-DST and 4. video labelling. During this period, I have successfully replicated the results of all the related papers, and assisted the research work by: Building the google colab website for paper SafetyDPO, participating in Human preference annotations for AI videos labelling, and assisting the search work of related papers that may be our group’s future research topics. Now, It is believed that future works of my research will focus on flow_grpo and 3D-DST.
RkJQdWJsaXNoZXIy NDk5Njg=