In the 2nd paper, we addressed the computational inefficiency of language models for code stemming from Byte Pair Encoding tokenization, which generates lengthy subtoken sequences. We show that token merging can reduce BPE inefficiencies without the need for pretraining.
CONCORD, a domain-specific language (DSL), enables customizable graph-based code representations across programming languages, reduces graph size complexity through simplification heuristics, and improves scalability and reproducibility in software engineering tasks.
AI pipelines have multiple knobs for tu(r)ning computational resources, but energy often gets sidelined. Our work in IEEE Software shows: tu(r)ning these knobs strategically with energy as a design constraint achieves significant time & energy gains with minimal performance loss.
Our study (accepted in @ieeesoftware spl issue on Green Clean Sw #Sustainability)
- emphasizes on treating energy efficiency as the first-class design consideration
- demonstrates that smartly combining optimizations achieves 94% energy reduction while preserving 95% F1 score.
I am deeply honored to receive the Distinguished Service Award for my contributions as Program Co-chair of @ieeescam 2025. This recognition reflects the collective effort of the organizing committee, reviewers, and authors who made the conference an enriching experience. #icsme
SMART lab's PI, Tushar Sharma, has been honored with the Dean's Research Excellence Award for his contributions to Software Engineering and Green AI! #AcademicExcellence
Our study, fresh out of the oven, explores how various energy optimization techniques that we refer to as 'knobs' influence each other and help us attain an optimized AI pipeline with energy efficiency as the first-class consideration. #greenai https://t.co/pz0mhZRPxp
Today, Mootez @s_mootez will present our paper "An adaptive language-agnostic pruning method for greener language models for code" in the 'Fairness and Green' session at 16:00. If you are attending @FSEconf, consider attending it.
To what extent does refactoring actually remove smells? Which refactoring remove which smells? How often do smells disappear w/o any known refactoring? Do all smells live equally long?
Our paper exploring these questions has recently accepted at ESEM 2025. Preprint coming soon!
Got to present our work on "Fine-grained energy profiles of DL frameworks" at ICSE'25, met fellow researchers, attended some great talks, got to be student volunteer, saw the parliament amid elections, got some sun after a long time. Overall a good week🌞 @SMART_Dal@ICSEconf
Dalhousie@ICSE 2025:
Together, @DalhousieU software engineering researchers presented three ICSE (two research and one journal-first track), two ICPC (one research and one ERA track), two MSR (one research and one tools track), one FORGE (dataset), and one ICSEW papers.
We are presenting four papers in this week at ICSE (two in MSR, and one each in ICSE and FORGE). Two PhD students @s_mootez and @SauftwareBug will be around as student volunteers. Feel free to say "Hi!" to us 👋
🇨🇦 Attention Canadian #SoftwareEngineering researchers!
Calling new faculty & students: Consider submitting your talk proposals to CSER 2025 (Spring). A great opportunity to connect with fellow researchers and discuss your amazing research.
���� Apr 25 🔗 https://t.co/cdGSao6gwK
Excellent news from @FSEconf. Our paper "An adaptive language-agnostic pruning method for greener language models for code" is accepted. It offers a pruning method maintaining ~original accuracy with substantially less computation for LMs.
preprint: https://t.co/FIjRse0bBH
Our research introducing reproducibility smells in IaC code has been accepted in @msrconf 2025 (technical track) offering
a) a catalog of reproducibility smells
b) a tool REDUSE to identify these smells in Ansible code,
c) an empirical analysis
preprint: https://t.co/TxfLkV51PN
What a great time to introduce DPy to the world! Our paper discussing the tool has been accepted in @msrconf 2025 (tools and data track). Congrats Aryan.
Preprint: https://t.co/TTEGRWQaz0
Try it: https://t.co/JMSmMWwR2s