BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260402T024534Z
LOCATION:3008\, Level 3
DTSTART;TZID=America/Los_Angeles:20250624T171500
DTEND;TZID=America/Los_Angeles:20250624T173000
UID:dac_DAC 2025_sess157_RESEARCH1868@linklings.com
SUMMARY:Location is Key: Leveraging LLM for Functional Bug Localization in
  Verilog Design
DESCRIPTION:Bingkun Yao and Ning Wang (City University of Hong Kong), Jie 
 Zhou and Xi Wang (Southeast University), Hong Gao (Zhejiang Normal Univers
 ity), Zhe Jiang (Southeast University), and Nan Guan (City University of H
 ong Kong)\n\nIn Verilog code design, identifying and locating functional b
 ugs is an important yet challenging task. Existing automatic bug localizat
 ion methods have limited capabilities; they only suggest a set of potentia
 l buggy lines rather than precisely identifying the bug. Moreover, they de
 pend on verification tools like testbenches and reference models, which re
 quire expert input and are time-consuming to develop. This paper introduce
 s LiK (Location is Key), an open-source Large Language Model (LLM) to prec
 isely locate functional bugs in Verilog code without the need for expert-w
 ritten verification tools. LiK is developed from the open-source coding LL
 M Deepseek-Coder-Lite-Base-16B through a threestep\ntraining process: cont
 inuous pre-training to enhance foundational knowledge, supervised fine-tun
 ing to learn how to output localization results, and reinforcement learnin
 g to reduce output errors. Experiment\nresults demonstrate that LiK achiev
 es superior functional bug localization accuracy, outperforming both the S
 OTA traditional method Strider, and SOTA closed-source LLMs like GPT-o1-pr
 eview and Claude-3.5-Sonnet. Moreover, integrating LiK into the SOTA LLM-b
 ased Verilog debugging tool significantly boosts its functional bug fixing
  success rate from 76.47% to 90.54%. This underscores LiK's potential to e
 nhance the performance of end-to-end automatic Verilog debugging tools.\n\
 nTopics: Systems\n\nTracks: SYS3: Embedded Software\n\nSession Chairs: Hoe
 seok Yang (Santa Clara University) and Younghyun Kim (Purdue University)\n
 \n
END:VEVENT
END:VCALENDAR