Project Grant 2431265
Award Date
10/1/24
Completion Date
9/30/25
Dollars Obligated
$295K
Federal Agency
Division of Computing and Communication Foundations
Federal Grant Program
47.070
Assistance Type
Project Grant
Place of Performance
Pittsburgh, PA 15213, USA
Similar Awards
This Project Grant award of $675,000 was provided by the National Science Foundation (NSF) under the Computer and Information Science and Engineering (CFDA 47.070) program. The award aims to improve the ability of large language models (LLMs) to generate high-quality source code by deeply integrating program analysis techniques into the LLM training, code generation, and evaluation processes. The key products and services to be delivered include: Developing novel quantitative program analysis...
This $567,000 Project Grant award from the National Science Foundation's (NSF) Mathematical and Physical Sciences program (CFDA 47.049) supports research at Carnegie Mellon University to develop novel techniques that combine machine learning and symbolic AI for mathematical reasoning. The project aims to enable AI systems to better understand and generate mathematical proofs, as well as create new tools to help humans formalize and digitize mathematical definitions and proofs. Key research...
This National Science Foundation (NSF) Computer and Information Science and Engineering (CISE) Federal Grant Program (CFDA 47.070) award to Purdue University will fund a $402,777 research project titled "CAREER: REGULARIZING LARGE LANGUAGE MODELS FOR SAFE AND RELIABLE PROGRAM GENERATION." The project aims to advance the understanding of limitations in using large language models (LLMs) for program generation and develop methods to enhance the correctness, safety, and robustness of...
The National Science Foundation awarded $544,003 to Carnegie Mellon University under the Computer and Information Science and Engineering federal grant program (CFDA 47.070) for a three-year Project Grant beginning October 1, 2022. The grant aims to develop formal libraries, methods, and tools to carry out tasks like encoding statements as clausal formulas and reducing search spaces in verified ways using automated reasoning and interactive theorem proving. Specifically, the university will...
This Project Grant award from the National Science Foundation (NSF) Computer and Information Science and Engineering (CISE) program (CFDA 47.070) provides $225,000 to Colorado State University to improve the code generation capabilities of large language models (LLMs). The project aims to develop novel program analysis techniques that can provide meaningful feedback to LLMs during training and decoding, enabling them to generate higher-quality, semantically-aware code. Key research activities...
This Project Grant award of $389,000.00 from the National Science Foundation (NSF) Computer and Information Science and Engineering (CISE) Federal Grant Program supports research to enable provably correct (correct-by-construction) code generation of key computational routines in a high-performance manner for computational chemistry applications. The project will develop new notations for describing operations involving structured matrices and tensors, new algorithms to identify high-performance...
This $499,999 federal Project Grant award from the National Science Foundation's (NSF) Computer and Information Science and Engineering (CISE) program (CFDA 47.070) will support collaborative research to investigate the theoretical foundations of compositional learning in large language models (LLMs) based on transformer architectures. The research aims to advance the understanding of how LLMs, such as GPT-4, LLAMA 2, and CLAUDE 3, can decompose complex tasks into simpler intermediate steps to...
This $400,000 project grant awarded by the National Science Foundation (NSF) under the Computer and Information Science and Engineering (CFDA 47.070) program supports research to develop a systematic framework for visualizing, understanding, and rewriting the learned computations of multimodal generative AI models. The key objectives are to: 1) create new methodologies to visualize the internal mechanisms and hierarchical structures of pre-trained multimodal generative models, 2) explore model...
This $875,000 Project Grant was awarded by the National Science Foundation (NSF) under the Computer and Information Science and Engineering (CISE) Federal Grant Program (CFDA 47.070) to the University of California, San Diego (UCSD). The project aims to develop new techniques for aligning large language models (LLMs) with formal specifications in order to generate high-quality computer code that provably matches user intent. Specifically, the project will: (1) develop grammar-aligned decoding...
This $100,000 Project Grant was awarded by the National Science Foundation's (NSF) Computer and Information Science and Engineering (CISE) program (CFDA 47.070). The grant supports the development of an open, community-driven evaluation infrastructure to assess the safety risks of large language models (LLMs). The project aims to: (i) conduct surveys and interviews with experts to identify critical safety concerns and evaluation gaps; (ii) organize a workshop to refine the evaluation...