Research Seminars @ Illinois

View Full Calendar

Tailored for undergraduate researchers, this calendar is a curated list of research seminars at the University of Illinois. Explore the diverse world of research and expand your knowledge through engaging sessions designed to inspire and enlighten.

To have your events added or removed from this calendar, please contact OUR at ugresearch@illinois.edu

CS COMPILER SEMINAR: Shubham Ugare, "SynCode: LLM Generation with Grammar Augmentation"

Event Type
Ceremony/Service
Sponsor
CS Compilers Area
Location
Room 2406(Siebel)
Date
Dec 9, 2024   4:30 - 5:30 pm  
Speaker
Shubham Ugare
Views
3
Originating Calendar
Siebel School Speakers Calendar

CS COMPILER SEMINAR: Please join us on Dec 9th, from 4:30 - 5:30pm in the Room 2406 (Siebel) where Shubham Ugare will give a talk, “SynCode: LLM Generation with Grammar Augmentation”.  

Please see their abstract below:

Abstract: LLMs are widely used in complex AI applications. These applications underscore the need for LLM outputs to adhere to a specific format, for their integration with other components in the systems. Typically the format rules e.g., for data serialization formats such as JSON, YAML, or Code in Programming Language are expressed as context-free grammar (CFG). Due to the hallucinations and unreliability of LLMs, instructing LLMs to adhere to specified syntax becomes an increasingly important challenge. We present SynCode, a novel framework for efficient and general syntactical decoding with LLMs, to address this challenge. SynCode ensures soundness and completeness with respect to the CFG of a formal language, effectively retaining valid tokens while filtering out invalid ones. SynCode uses an offline-constructed, efficient lookup table, the DFA mask store, derived from the DFA of the language's grammar for efficient generation. SynCode seamlessly integrates with any language defined by CFG, as evidenced by experiments focusing on generating JSON, Python, and Go outputs. Our experiments evaluating the effectiveness of SynCode for JSON generation demonstrate that SynCode eliminates all syntax errors and significantly outperforms state-of-the-art baselines. Furthermore, our results underscore how SynCode significantly reduces 96.07% of syntax errors in generated Python and Go code, showcasing its substantial impact on enhancing syntactical precision in LLM generation.

link for robots only