FORGE 2025
Sun 27 - Mon 28 April 2025 Ottawa, Ontario, Canada
co-located with ICSE 2025

Code Language Models (CLMs) have demonstrated high effectiveness in automating software engineering tasks, such as bug fixing, code generation, and code documentation. This progress has been driven by the scaling of large models, ranging from millions to trillions of parameters (e.g., GPT-4). However, as models grow in scale, sustainability concerns emerge, as they are extremely resource intensive, highlighting the need for efficient, environmentally conscious solutions. GreenAI techniques, such as QLoRA, offer a promising path for dealing with large models’ sustainability as they enable resource-efficient model fine-tuning. Previous research has shown the effectiveness of QLoRA in code-related tasks, particularly those involving natural language inputs and code as the target output (NL-to-Code), such as code generation. However, no studies have explored its application to tasks that are fundamentally similar to NL-to-Code but operate in the opposite direction, such as code summarization. This leaves a gap in understanding how well QLoRA can generalize to Code-to-NL tasks, which are equally important for supporting developers in understanding and maintaining code. To address this gap, we investigate the extent to which QLoRA’s capabilities in NL-to-Code tasks can be leveraged and transferred to code summarization–that we use as representative of Code-to-NL tasks. Our study evaluates two state-of-the-art CLMs (CodeLlama and DeepSeek-Coder) across two programming languages: Python and Java. Each model was tasked with generating a meanigful code description for the underlying code component. The findings of our research confirm previous patterns that emerged when applying QLoRA to source code generation. Notably, we observe that QLoRA not only allows efficient fine-tuning of CLMs for code summarization but also achieves the best results with minimal parameter adjustment compared to full model fine-tuning, which requires expensive recalibration of all model parameters in the traditional fine-tuning process.

Mon 28 Apr

Displayed time zone: Eastern Time (US & Canada) change

11:00 - 12:30
Session4: Human-AI Collaboration & Legal Aspects of using FMResearch Papers / Industry Papers at 207
Chair(s): Zhenhao Li York University
11:00
12m
Long-paper
Extracting Fix Ingredients using Language Models
Research Papers
Julian Prenner Free University of Bozen-Bolzano, Romain Robbes CNRS, LaBRI, University of Bordeaux
11:12
12m
Long-paper
CodeFlow: Program Behavior Prediction with Dynamic Dependencies Learning
Research Papers
Cuong Chi Le FPT Software AI Center, Hoang Nhat Phan Nanyang Technological University, Huy Nhat Phan FPT Software AI Center, Tien N. Nguyen University of Texas at Dallas, Nghi D. Q. Bui Salesforce Research
11:24
12m
Long-paper
Addressing Specific and Complex Scenarios in Semantic Parsing
Research Papers
Yu Wang Xi'an Jiaotong University, Ming Fan Xi'an Jiaotong University, Ting Liu Xi'an Jiaotong University
11:36
12m
Long-paper
Skill over Scale: The Case for Medium, Domain-Specific Models for SE
Research Papers
Manisha Mukherjee Carnegie Mellon University, Vincent J. Hellendoorn Carnegie Mellon University
Pre-print
11:48
12m
Long-paper
Resource-Efficient & Effective Code Summarization
Research Papers
Saima Afrin William & Mary, Joseph Call William & Mary, Khai Nguyen William & Mary, Oscar Chaparro William & Mary, Antonio Mastropaolo William and Mary, USA
12:00
6m
Short-paper
How Developers Interact with AI: A Taxonomy of Human-AI Collaboration in Software Engineering
Research Papers
Christoph Treude Singapore Management University, Marco Gerosa Northern Arizona University
Pre-print
12:06
6m
Short-paper
"So what if I used GenAI?” - Legal Implications of Using GenAI in Software Engineering Research
Research Papers
Gouri Ginde (Deshpande) University of Calgary
Pre-print
12:12
6m
Short-paper
Evaluating the Ability of GPT-4o to Generate Verifiable Specifications in VeriFast
Research Papers
Marilyn Rego Purdue University, Wen Fan Purdue University, Xin Hu Univeristy of Michigan - Ann Arbor, Sanya Dod , Zhaorui Ni Purdue University, Danning Xie Purdue University, Jenna DiVincenzo (Wise) Purdue University, Lin Tan Purdue University
12:18
6m
Short-paper
Towards Generating App Feature Descriptions Automatically with LLMs: the Setapp Case Study
Industry Papers
:
:
:
: