Reyhan
@Reyhaneh
Followers
2,014
Following
806
Media
14
Statuses
992
Assistant Professor of @plfmse at @IllinoisCS , Director of Intelligent CAT Lab (), PhD @UCIbrenICS , @Google PhD Fellow
Champaign, IL
Joined April 2008
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
FEMA
• 1602165 Tweets
Joker
• 180765 Tweets
Grammys
• 154376 Tweets
1005 CHEERS TO XIAO ZHAN
• 143896 Tweets
Mayorkas
• 138649 Tweets
Mancuso
• 111923 Tweets
Record of the Year
• 89629 Tweets
Best Record
• 87656 Tweets
Ülke
• 67130 Tweets
Pogba
• 66451 Tweets
Fatih
• 60691 Tweets
Uyuşturucu
• 55857 Tweets
#turkishwomenareindanger
• 51375 Tweets
Valencia
• 34917 Tweets
Marseille
• 27839 Tweets
#ikbaluzuner
• 23345 Tweets
Jamie Dimon
• 21426 Tweets
Semih Çelik
• 19984 Tweets
Celtics
• 19138 Tweets
Dolly
• 18942 Tweets
Angers
• 15490 Tweets
Leeds
• 12621 Tweets
Greenwood
• 10644 Tweets
Pinned Tweet
Very excited to share that I have been awarded
@NSF
CAREER award
@plfmse
@IllinoisCS
. At the Intelligent CAT lab, we have already made awesome progress on some of the proposed ideas and looking forward to exploring other proposed ideas in future.
37
7
324
I'm recruiting PhD students. If you are interested to do research at the intersection of AI and software engineering (specifically, using ML, DL, and RL to solve real-world software engineering problems), please apply. I appreciate retweets.
Software Engineering group (
@Reyhaneh
, Lingming Zhang, Darko Marinov) at UIUC has multiple openings for PhD students. If you are interested to engage in a cutting-edge research in Software Engineering, please apply! If you have outstanding students, encourage them to apply.
1
13
34
10
233
377
Surprisingly, "LLMs struggle with reasoning about the code they synthesize and pass the test execution" filter! So, we introduce CodeMind, a framework to evaluate LLMs for inductive code reasoning. A🧵 of findings 👇
Paper:
Code:
6
33
190
We're hiring
@IllinoisCS
! All areas, all ranks (tenure-track). Applications from senior and mid career researchers are specifically encouraged. Feel free to reach out if you have any questions.
8
28
138
It makes me sick whenever I see age restrictions to apply for a grant/award/job. This award is for early-career people, but there are people (mostly underrepresented ones) who start their careers later. A clear case of discrimination 😡
8
6
120
[please retweet] Interested in a new adventure for 2024? We've opened the applications for UIUC+ summer research in software engineering (and more related topics). Link to the application (soft deadline is on January 21st):
2
43
119
One of the high school students who researched with me told me he got an offer from Princeton, and I'm pretty sure he will get more outstanding offers! Working with high school students was something I never expected to do or even enjoy during my early career. But I'm glad I did.
6
5
92
Please spread the word: We have started recruiting for UIUC summer research program in "software engineering." The (soft) deadline for application is March 1, 2023. Looking forward to seeing your applications:
2
32
83
LLMs are good at code synthesis, when asked to implement the logic through natural language. But how about we ask them to implement an existing logic in a different programming language? Welp, it's quite disappointing. Read more about in the thread 👇
I’m happy to announce our work “Lost in Translation: A Study of Bugs Introduced by Large Language Models while Translating Code” is accepted to
@ICSEconf
2024. Joint work b/w
@IllinoisCS
and
@IBMResearch
(a 🧵)
📃 Paper:
📀 Code:
7
73
315
3
12
80
Proud advisor moment: my students
@sunniercy
(co-advised by Darko Marinov) and
@alibrahimzada
ranked 3rd and 2nd on the ACM Student Research Competition
@ICSEconf
#icse24
🎉🎉🎉
3
4
68
Academic peeps, how do you thank your outstanding TAs (in addition to nominating them for department awards)? Some of my TAs are incredible and we absolutely cannot handle 300 students without them.
6
3
60
Looking for a PhD position and attending virtual
@ICSEConf
? Come talk to me during social hours! I am looking for PhD students who are interested to apply AI techniques, mathematical optimization, and formal methods to solve real world software engineering problems.
1
20
61
Yes. You can be
@IllinoisCS
faculty or PhD student and live in Chicago, one of the greatest cities in the world! The
@IllinoisCS
Chicago campus is just across the river from Union Station and there are several daily shuttles (with wifi) between two campuses!
Posting this pic of uiuc Chicago campus neighborhood for prospective students as I’m walking around right now :)
4
0
67
0
5
57
Excited to introduce InteNSE (
@InteNSEWorkshop
), the first workshop on the interpretability and robustness of ML4Code, co-located with
@ICSEconf
in Melbourne. The submission deadline is January 13th (CFP and more: )
@XAI_Research
2
8
51
I muted the word "gpt" and I'm much happier with the quality of my feed 😊
2
0
49
I had 3 major revisions in my batch, which would have been rejected without major revision. For, I engaged in a lot of discussions with one or both other reviewers to land a major revision rather than reject. After the major revision, I increased my score (and did some others).
1
0
47
1/ So, rejections happen all the time and it's part of the academic life. But sometimes the reviews and review process make it very hard to digest the rejection:
4
0
47
Paper "Automated Construction of Energy Test Oracle" accepted
@FSEconf
. The proposed approach leverages Deep Learning for construction of an "interpretable" model to serve as an oracle, i.e., automatically determine if execution of a test reveals an energy issue or not.
5
2
45
Facebook and Instagram now show me advertisements related to my husband's Google search history 🤷. Don't do this guys. You can spoil surprises or ruin marriages.
1
0
44
My teaching evaluation results came almost the same time as ICSE results, and I'm so flattered reading that students really liked the course and it had increased their interests in ML4Code.
0
1
42
New paper from Intelligent CATs! This is special to me for several reasons: 1) It's my first last-author paper 2) It's the outcome of UIUC SE summer research program and student authors are/were undergrads (
@alibrahimzada
joins
@illinoisCS
and Dilara joins
@manningcics
for PhD)
I'm excited to announce that our paper "Perfect is The Enemy of Test Oracle" is accepted to
@FSEconf
2022. Joint work with
@Reyhaneh
,
@yigit_v4rli
, and Dilara Tekinoğlu. Preprint coming soon. (1/n)
1
1
23
3
3
42
Today was the last day of my ML for Software Engineering and Analysis class at
@IllinoisCS
. The semester-long project of the course was about exploring the strengths and weaknesses of
#ChatGPT
related to several code-related tasks.
1
1
38
If you are unhappy with the review system, then fix it. The most essential first step is to fight with your unreasonable co-reviewers. Argue with them. If they see someone stands against them, it's unlikely they show similar unreasonable behavior again.
2
4
38
Update: we received about 700 applications 🤯! The goal of this program is providing research experience in software engineering worldwide. So, to give a fair chance to everyone, Darko,
@alibrahimzada
, and I are preparing an educational evaluation task to send to the applicants.
[please retweet] Interested in a new adventure for 2024? We've opened the applications for UIUC+ summer research in software engineering (and more related topics). Link to the application (soft deadline is on January 21st):
2
43
119
5
2
38
THIS! I put an enormous amount of time into explaining formal things intuitively. I assume most of the readers of my papers are junior students who recently got interested in the topic, so all the content should be easily understandable. This, however, comes at a cost ...
@JAldrichPL
A really well explained paper is often considerably less surprising, than one that jumps about all over the place. A good narrative makes the results seem completely obvious, but then the paper gets rejected for being too obvious, discouraging good explanations.
1
2
27
0
1
34
It is a big deal! I had mine in the third year through a complicated pregnancy with three surgeries. But two of my major publications came after that, including the ICSE paper I started working on it 6 weeks postpartum and submitted <3 months afterward.
#humblebrag
#phdchat
Getting phds while growing humans like it’s no big deal.
175
2K
22K
2
0
35
Prem is a gem for our research community. I refer to his naturalness of programming languages work as a big bang of ML4Code. But he's beyond a fantastic researcher to me. He's very humble, cares about juniors, sits with them, listens about their ideas, and promotes them.
This year's prestigious Harlan D. Mills award goes to Premkumar Devanbu, “For impactful contributions to the statistical modelling of source code and development practices [...].” I am looking forward to the ICSE keynote. Congrats Prem!
11
6
92
2
1
36
Congratulations to recently named ACM Fellows, including
@taoxiease
and
@tomzimmermann
. Very well-deserved 🎉
0
0
36
Not a reality for me and won't be for my students. I specifically encourage my students to take the weekends off to refresh their minds and come back with more energy and focus.
In my first year as a PhD student my supervisor emailed me saying “most PhD students occupy their weekends with PhD work and that’s the reality of it” this expectation is extremely harmful
#AcademicMentalHealth
#AcademicChatter
#phdlife
125
208
3K
1
0
35
No paper is perfect. Judge the paper the way you want others to judge your paper. Consider the fact that most of the first authors are (PhD) students, and they need to solve a problem over the course of 5-6 years, not in a single paper!
0
2
35
Happily discarded an abstract submission
@ASE_conf
, because the paper accepted to appear
@issta_conf
. The improvements will appear in the camera-ready (thanks to great feedback from reviewers). My awesome student Yang will soon share the details.
5
0
34
Achievement of the day: I found a typo in one of my papers published 5 years ago. Now I can go to sleep.
0
0
33
If you're in the first year of your faculty job and think everything is out of control, I have some advice for you: savor your moments and enjoy the job honeymoon 🤪
2
0
32
My <<ML for Software Engineering and Analysis>> class
@IllinoisCS
ended today, with a great guest lecture by
@miltos1
. Thanks for joining us, Miltos! Lots of insightful and interesting discussions.
1
1
31
I'm on the academic job market this year. I will be at the ASE (
@ASEconf2019
) and would love to talk to you about my research and hear about opportunities at your schools.
4
10
32
Due to the financial hardship caused by the pandemic for some applicants, I plan to interview and review the materials of applicants interested to work with me "before" the application deadline. So, please email me with your application materials if you are interested.
4
2
29
Payal was one of the best students in my ML4Code class. Looking forward to seeing what she is doing next
"I'm honored to have spent the past year as a Graduate Research Assistant at NCSA. It's been an incredibly rewarding experience, both professionally and personally." -
@IllinoisCS
Master's student Payal Mantri
We are proud to work with Payal and recognize her work!
#illinoisgsaw
0
2
15
1
0
31
Special thanks should go to these awesome folks who volunteered to make
@FSEconf
great. I chose to stay in academia because I love working with students and chairing SVs was a great experience.
@burnettmmosu
@DomSteinhoefel
@CorrensonArthur
@schandra
@CaptainEmerson
@marcelodamorim
@davidlo2015
This was it from
@FSEconf
2023 in San Francisco! I hope you enjoyed the conference and everything around it. A shout-out and huge thank you to our student volunteers 🙌 🙏
Thank you also to
@Reyhaneh
and Rahul Purandare, our student volunteer co-chairs.
0
2
8
1
2
28
We have updated CodeMind with (1) additional metrics, (2) more in-depth analysis, (3) one more model (StarCoder 2), and (4) the results of another code generation task: code translation, where LLMs should showcase their code understanding in "two" programming languages.
Surprisingly, "LLMs struggle with reasoning about the code they synthesize and pass the test execution" filter! So, we introduce CodeMind, a framework to evaluate LLMs for inductive code reasoning. A🧵 of findings 👇
Paper:
Code:
6
33
190
0
3
27
In five hours, my awesome student, Yang Chen (co-advised with Darko Marinov), will present our paper Croissant
@issta_conf
.
Paper:
Tool:
3
1
26
Virtual conferences are more accessible, but still not that much for people of certain countries, e.g., Iran. Youtube and Zoom are filtered in Iran and Slack does not provide service to the people in Iran. They need VPN, which makes the slow internet connection even worse. 1/2
1
3
27
If you have to put a huge effort in prompt crafting to make/enable a model to answer a question correctly (or in a satisfactory way), then it's you who has solved the problem not the LLM. Embrace some capabilities of LLMs but be realistic about their inherent limitations.
1
4
28
One of [few] positive aspects of the pandemic was making [almost] everything online, and this, to a great extent, promoted "inclusivity" (examples in the thread). Despite a very big emphasis on inclusiveness, academia does not seem to embrace this opportunity! 1/5
5
1
27
We have already received "291 applications" in just 10 days. Please encourage others to apply if they are interested in being involved in cutting-edge PLSE research at UIUC or working with UIUC Alumni! High school students are welcome to apply as well.
[please retweet] Interested in a new adventure for 2024? We've opened the applications for UIUC+ summer research in software engineering (and more related topics). Link to the application (soft deadline is on January 21st):
2
43
119
0
7
26
Super excited about this. A lot of exciting news in the next few years.
Announcing the
#ILLINOIS
Siebel School of Computing and Data Science at
@uofigrainger
made possible with a $50 MM gift from
@TomSiebel
.
With our
#5
in-the-nation computer science program and 21 blended degree programs, the best is yet to come! 🔸🔹
▶️
0
19
74
0
2
27
Please retweet and spread the word. We are calling for applications to our software engineering summer research program. This is a fully remote and mostly unpaid research opportunity.
4
13
26
Given the very fast progress of AI, many papers accepted in the first round seem obsolete at ICSE. The authors had papers accepted with no motive to explore the impact of new models. It's something we might consider when we have a deadline 15 months before the main conference!
1
1
25
There should be a non-profit organization for evaluating ML models using complex enough benchmarks unknown to others! Current leaderboards use known (and non-complex) benchmarks. This is the only way we can properly and fairly evaluate models.
1
3
25
Lot's of great interesting research is going on in the community at the intersection of AI and SE
The
@ICSEconf
doctoral symposium focus groups are in full swing. Thanks to our mentors
@lauriewilliams
@Reyhaneh
@AndreasZeller
@xLeitix
Silvia Abrahão and Matt Dwyer
0
1
29
1
0
24
This is very important, specifically if you do not have a support network outside of the institute you decide to join. Select a place where people will work with you, care about your success not just more papers for themselves, not ignoring you, and not competing with you.
It’s faculty interview season. Reminder for y’all candidates to please also think of an interview as an opportunity to evaluate whether you would like to spend a large part of your future life with the existing cohort of faculty members at the host institution!
1
4
90
1
1
23
Excited to see people around
@ICSEconf
. I'm organizing
@InteNSEWorkshop
tomorrow and will be at the main conference. Come talk to me if you're interested in PhD or post-doc opportunities
@IllinoisCS
on ML4Code and autonomous software.
1
2
23
I love how the PC co-chairs of FSE'24 are handling the rebuttal. Solid and positive guidelines for reviewers to judge the papers and response letters, followed by enforcement from the chairs! Thank you, Lin and
@davidlo2015
.
1
1
23
How much does all the electricity and GPU burnt for ICML submissions contribute to global warming?
2
1
23
@ali
presents this work tomorrow at 4 PM
@ICSEconf
. He has "incredibly expanded" the evaluation to more models and set a leaderboard for code translation task: . If you cannot attend the talk, a pre-recorded video is at
I’m happy to announce our work “Lost in Translation: A Study of Bugs Introduced by Large Language Models while Translating Code” is accepted to
@ICSEconf
2024. Joint work b/w
@IllinoisCS
and
@IBMResearch
(a 🧵)
📃 Paper:
📀 Code:
7
73
315
3
3
22
You're from Gaza/middle east. You flee your home and leave your family and everything you have memory with, not necessarily for a better future, but to avoid you and your kids being brutally killed by inhumans. And this happens ...
We are shocked and disturbed to learn that a landlord in Chicago expressing anti-Muslim and anti-Palestinian views broke into a Muslim family's apartment and attacked them with a knife, injuring the mother and killing her 6-year-old son, Wadea Al-Fayoume.
@cairchicago
will hold a
3K
41K
86K
2
5
20
Translation bugs are real, demonstrated by the fly in the title 😄
1
1
20
Yes, the peer-review process mostly is not fair. But I wanted to thank
@AndreasZeller
and
@DanaHDamian
for the reviewer assignment
@ICSEconf
. We got three expert reviewers, who have read the paper thoroughly and asked reasonable questions. Fingers crossed for rebuttal!
0
0
21
The preprint of this paper is available at:
Paper "Automated Construction of Energy Test Oracle" accepted
@FSEconf
. The proposed approach leverages Deep Learning for construction of an "interpretable" model to serve as an oracle, i.e., automatically determine if execution of a test reveals an energy issue or not.
5
2
45
0
2
21
So, the job of people at OpenAI is creating tools that take other people's jobs? Editors, programmers, and now filmmakers? At least my job as a researcher is safe as they don't create something that takes their own job, or do they? 👀
2
0
21
Interpretation, of any form, even a simple visualization, should be a "must" for any application of ML/DL. Do not trust your black-box model! Do not trust a paper with a black-box model 🙂
0
0
20
The Programming Languages, Formal Methods, and Software Engineering group
@IllinoisCS
is on Twitter! Stay tuned :)
We (faculty and students at PL/FM/SE group) aim to improve software quality and reliability with innovative techniques in programming languages, formal verification, software testing, and automated debugging.
0
4
11
0
3
20
Yours truly is running for IEEE-TCSE ExCom (At-Large member). If you are a TCSE member, please consider voting for me (and other awesome candidates!): . The deadline is in a week (Sep. 15th).
5
1
20
It's been almost 4 years that I count every single night down until 11/3/20. Every single night! I have a hope ...
0
0
19
My toddler understands "wait for your turn" and "do not interrupt" better 🙃
0
1
18
@IllinoisCS
is a top notch Computer Science department worldwide, and Chicago is an awesome city. Some
@IllinoisCS
faculty even deliver in-person courses and you get to be in touch with them in Chicago!
Launch your career in computer science with
@IllinoisCS
' Master of Computer Science degree in Chicago! Busy professionals can complete a degree while continuing to meet current life and career obligations.
Apply by Feb. 15.
See link in bio.
#illinoisCS
#MCSChicago
0
2
5
0
5
18
Super excited about the second edition of
@InteNSEWorkshop
. We welcome papers on assessing the (un)interpretability, (poor) generalizability, and (non)robustness of LLMs for code-related tasks. Submission deadline is Dec 9, 2023 (in two months). More info:
InteNSE workshop is back! The focus of this year is on LLM for Code. So, if you have preliminary results on leveraging LLMs in code-related tasks, you should consider submitting them to this workshop.
3
0
1
0
4
17
Interested in fine-tuning language models at scale without the need for ground-truth? Read our recent work
@IllinoisCS
on using textual information as a source of feedback in an iterative fine-tuning process.
Paper:
Code:
Can pretrained language models (LMs) go beyond learning from labels and scalar rewards? Introducing LeTI, a new LM finetuning paradigm that explores LMs' potential to learn from textual interactions & feedback, allowing LMs to understand not just if they were wrong, but why. 🧵1/
2
36
188
0
1
16
Meet my awesome student Yang (co-advised with Darko Marinov). She'll be at ICSE with my other student
@alibrahimzada
. This is their first ICSE. Go talk to them 😊
Great talk by
@sunniercy
(PhD student at
@IllinoisCS
) on whether ChatGPT can fix non-order-dependent flaky tests!
Spoilers: ChatGPT cannot do it well!
0
1
13
0
1
16
Related work and threats to the validity sections are useless to me. I prefer to discuss the research gap (why related work does not solve the problem) in the intro and threats to the validity of the evaluation when I discuss the evaluation design. What am I missing?
4
0
16
[please RT] Are you an undergrad and interested to engage with cutting-edge software engineering research? Apply to our software engineering summer research program at . The deadline is March 20, but we may accept applications until positions are filled.
0
9
14
So, FSE conference is switching to a one-column format from two columns. What are tips and tricks to transform a (rejected) two-column submission into a one-column? Algorithms and one-column figures are now big and ugly, and wide tables don't fit.
3
1
16
@IllinoisCS
alumni rock!
Thrilled to announce I have received the
@NSF
CAREER Award! This grant will fuel our efforts in advancing enterprise security solutions 🔐. I am deeply grateful to my students, mentors, and collaborators who have made this journey possible. Excited for what lies ahead!
33
2
140
1
1
16
Nominations for
@sigsoft
awards are due in a few days. Out of curiosity, I looked to see how the awards (those given to individuals, not groups such as MIP) are distributed between genders (only male/female because I can only verify these two from public information). 1/n
1
0
16
What I really miss since starting my faculty job is in-depth, informal discussions about my research topics of interest, i.e., what are the research gaps, how SoTA is still not there, and what would be the next breakthrough. 1/n
1
2
16
One of the best talks I've recently attended. Full of insightful ideas and awesome work done by
@swarat
, his team, and collaborators.
The tutorial on neuro-symbolic programming is about to start. Please join if you're interested.
0
0
3
1
2
15
[Please retweet] Workshops are beyond paper presentations and keynotes! So, we added a tutorial on neuro-symbolic programming to the InteNSE workshop! Read more information about the tutorial in the thread below and consider participating if you attend
@ICSEconf
in Lisbon!
We have a new component at InteNSE'24: A Neuro-symbolic Programming Tutorial. Please find additional information about the tutorial in this thread. (1) Registration for the InteNSE workshop AND (2) filling out the following form are mandatory to attend:
1
2
3
0
4
15
@lionel_c_briand
As a reviewer and a human, I always consider I may not be able to understand some parts of the paper I'm reviewing correctly. So, I TAKE REBUTTAL SERIOUSLY! It's unfortunate that many ignore rebuttals, specifically when theirs and their co-reviewers score are low.
0
0
15
Yet another award with "age limit" (), this time from
@AnitaB_org
that aims to broaden the participation of women, many of whom undergo unconventional paths that takes them a long time to start their job in academia/tech. Sigh.
1
1
14
When my students were "solely" relied on LLMs to automate coding tasks, I used to tell them what's your contribution? It's the power of LLM. Now that we're augmenting them with symbolic techniques, I ask myself what's the power of LLM. Crazy time to be a researcher 😀
If you have to put a huge effort in prompt crafting to make/enable a model to answer a question correctly (or in a satisfactory way), then it's you who has solved the problem not the LLM. Embrace some capabilities of LLMs but be realistic about their inherent limitations.
1
4
28
1
1
14
On Monday, April 15th, Mark Harman will present one of
@Meta
's most recent LLM-based code generation and testing advancements
@InteNSEWorkshop
. The talk is in person, but we also accommodate joining on Zoom. The information on how to attend remotely is below.
Meta’s LLM for software testing work is super exciting.
This paper describes Meta’s TestGen-LLM tool, which uses LLMs to automatically improve existing human-written tests. TestGen-LLM verifies that its generated test classes successfully clear a set of filters that assure
15
270
1K
0
0
14
I just wonder if the submission rate at theory conferences have also exploded. Is that ML making research and producing papers easier or is it just research becoming a popular business (bittersweet, cause I believe research should be done out of love and passion)?
0
0
14
Yes! I missed my first paper presentation (at ISSTA'16) because as an Iranian citizen, the German consulate required more time to investigate my eligibility and background!
Dear conference organisers: I promise I will choose a better birth place next time, but in the meantime can we please 1) host conferences at more visa-friendly locations or 2) leave more time between paper decision time and conference so we have enough time to apply for visas?
18
356
3K
2
0
13