Suggestions
Abhay Zala
Undergraduate Research Assistant at MURGe-Lab (UNC-NLP Group)
Professional Background
Abhay Zala is a talented computer science professional who has made significant strides in the field of technology and research. With a robust educational foundation from the University of North Carolina at Chapel Hill, Abhay earned a Bachelor of Science degree in Computer Science, where he not only excelled academically but also was deeply engaged in various scholarly activities. His commitment to advancing knowledge in the realm of natural language processing was highlighted during his tenure as an Undergraduate Research Assistant at MURGe-Lab, part of the UNC-NLP Group. In this role, Abhay contributed to innovative projects that center on harnessing machine learning and artificial intelligence to improve communication systems and understanding of human language.
In addition to his research work, Abhay gained valuable experience as a Summer Intern at the University of North Carolina at Chapel Hill. This position allowed him to apply his theoretical knowledge in practical settings, bridging the gap between classroom learning and real-world applications. Throughout his career so far, he has demonstrated a strong capability to adapt to varying roles and responsibilities, showcasing his versatility in the tech landscape.
Education and Achievements
Abhay's academic journey began at Panther Creek High School, where he laid the groundwork for his future studies in technology and science. It was here that he discovered his passion for computer science, leading him to pursue a degree at a prestigious institution like the University of North Carolina at Chapel Hill. The Bachelor of Science program enabled him to explore various facets of computing, such as algorithms, software development, and data structures, ultimately shaping his future career path.
Additionally, Abhay played a pivotal role in the Technology Student Association, where he not only participated as a member but also took on the leadership role of President. This experience honed his leadership skills, teaching him how to guide a team, manage projects, and foster innovation among his peers. His proactive involvement in this organization reflects his dedication to cultivating future tech leaders and his commitment to the tech community at large.
Achievements
Abhay Zala's background is adorned with noteworthy achievements that reflect his passion for technology and research excellence. As President of the Technology Student Association, he spearheaded initiatives that encouraged collaboration among students interested in technology, creating a vibrant environment for learning and exploration. Under his leadership, the organization fostered several projects that not only benefited its members but also made lasting impacts on the greater community.
His research contributions at MURGe-Lab are equally impressive, as they involve cutting-edge work in natural language processing, a field that is rapidly evolving and holds immense potential for future applications.
Overall, Abhay Zala stands out as a promising young professional driven by his enthusiasm for computer science. His educational background, coupled with his leadership experiences and research expertise, positions him well for a bright future in technology. As he continues to grow and evolve in his career, there is no doubt that Abhay will leave a lasting mark in the world of computer science and beyond.
Highlights
Can we design an efficient & versatile framework to reuse+adapt existing pretrained ControlNets to accurately guide any video/image diffusion model and support diverse controls?
📢 Introducing Ctrl-Adapter:
➡ Flexible Compatibility: Adapts any pretrained ControlNet (scribble/sketch, depth, canny edge, human pose, etc.) to any image or video diffusion model (Stable Video Diffusion, I2VGen-XL, Hotshot-XL, SDXL, etc.)
➡ Diverse Capabilities: Supports image control, video control, sparse video control, multi-condition control, zeroshot adaptation to unseen control conditions, and video editing
➡ Efficient Training: Outperforms both image/video baselines in < 10 GPU hours
📃 ⏩ For a quick summary, check out our Twitter thread: https://lnkd.in/e9nR5zmz
Work with Han Lin, Jaemin Cho, and Mohit Bansal at the University of North Carolina at Chapel Hill.
ArXiv: https://lnkd.in/eRxS7k8v Project Page: https://lnkd.in/emuNt5WZ Code: https://lnkd.in/enbnDSyq Huggingface: https://lnkd.in/e_-ZBnfU
📢 Can we adaptively generate training environments with LLMs to help small embodied RL game agents learn useful skills that they are weak at? 🤔
👉 Excited to share EnvGen, an effective+efficient framework in which an LLM progressively generates and adapts training environments based on feedback from the RL agent's intermediate successes/failures.
▶ A lightweight (4M params) RL agent trained w/ EnvGen outperforms a GPT-4 agent in Crafter environment ▶ Uses only 4 LLM calls for the entire training/inference (vs. thousands of calls per episode in LLM agents) ▶ Significant performance gain & speed up in the learning of long-horizon tasks
➡ For a quick summary, check out our Twitter thread: https://lnkd.in/eH7ysn2b
Work with Jaemin Cho*, Han Lin, Jaehong Yoon, and Mohit Bansal at the University of North Carolina at Chapel Hill.
ArXiv: https://lnkd.in/eYcnthRn Project Page: https://lnkd.in/eeg23SqX Code: https://lnkd.in/eCN4zZSF
