Lingming Zhang @LingmingZhang Twitter profile

Pinned Tweet

Lingming Zhang

3 months

Introducing OpenAutoCoder-Agentless😺: A simple agentless solution solves 27.3% GitHub issues on SWE-bench Lite with ~$0.34 each, outperforming all open-source AI SW agents! It's fully open-source, try it out: 🧑‍💻 📝

20

135

651

Last Seen Profiles

@Rollingimpala

@bookouture

@us_thang

@oftarth

@eonmedia

@tuckerhunter_

@Archang82320459

@alfazdilse

@LisaS83607805

@eldiariony

@CowgirlFC

@bokeplokalmalam

@comparsaJoseja

@svn69_1

@KykoTakai165187

@PoonamSharma990

@rinechuvia

@An__24_

@debby04473555

@RealPersianGod

@galery_basah10

@AmutuhaC17473

@_Lyn_D

@_SayPleaseDaddy

@Syadav5429

@ta_sund

@DespinaDimp

@RN06068911

@martxdiaz

@motoe_keisuke

@onic_esports

@elfyurdakull

@NatanaelCanoNew

@ogden_robyn

@will_motion

@Los_Bunkers

Lingming Zhang

@LingmingZhang

2 years

🚨LLMs are Zero-Shot Fuzzers! Excited to share our TitanFuzz🤖 work @issta23 : #LLMs can be directly applied for both generative and mutative fuzzing, while being fully automated, generalizable, and applicable to domains challenging for traditional fuzzers (such as DL systems)🧵

2

46

208

Lingming Zhang

@LingmingZhang

18 days

🚨 Excited to share our recent survey on LLM-based Agents for Software Engineering! We will keep updating the relevant paper list at:

GitHub - FudanSELab/Agent4SE-Paper-List: Repository for the paper "Large Language Model-Based...

Repository for the paper "Large Language Model-Based Agents for Software Engineering: A Survey". - FudanSELab/Agent4SE-Paper-List

github.com

Yiling Lou

@yiling__LOU

18 days

Introducing our latest survey on LLM-based Agents for Software Engineering (SE) LLM-based agents have shown remarkable effectiveness in SE. This survey covers 106 papers and summarizes them from both SE and agent perspective. 📝 🧑‍💻

4

28

131

0

31

131

Lingming Zhang

@LingmingZhang

2 years

Super grateful to receive this award, and looking forward to working on this exciting project with @Meta ! Special thanks to my students and colleagues at @plfmse and @IllinoisCS

Siebel School of Computing and Data Science

@IllinoisCDS

2 years

Congratulations to Lingming Zhang for winning a Meta 2022 AI4AI Research award! His innovative Intelligent Software Engineering (iSE) Lab will collaborate with Meta Research on learning to fuzz tensor compilers. 🔸♾️🔹

0

23

12

4

104

Lingming Zhang

@LingmingZhang

1 month

“On SWE-bench Verified, GPT-4o resolves 33.2% of samples, with the best performing open-source scaffold, Agentless.” Wow, our Agentless😺 is used by @OpenAI as the default technique to evaluate the performance of GPT-4o on SWE-bench! 🧑‍💻

GitHub - OpenAutoCoder/Agentless: Agentless🐱: an agentless approach to automatically solve...

Agentless🐱: an agentless approach to automatically solve software development problems - OpenAutoCoder/Agentless

github.com

OpenAI

@OpenAI

2 months

We're releasing a new iteration of SWE-bench, in collaboration with the original authors, to more reliably evaluate AI models on their ability to solve real-world software issues.

409

522

3K

1

4

97

Lingming Zhang

@LingmingZhang

2 years

🏆 NNSmith just won the Distinguished Artifact Award at #ASPLOS23 ! NNSmith has found various bugs for popular deep learning systems (ONNXRuntime, PyTorch, TensorFlow, TensorRT, TVM, and more!). Try our tool if you're interested in MLSys reliability:

GitHub - ise-uiuc/nnsmith: Automatic DNN generation for fuzzing and more

Automatic DNN generation for fuzzing and more. Contribute to ise-uiuc/nnsmith development by creating an account on GitHub.

github.com

Jiawei Liu

@JiaweiLiu_

2 years

Today we will present our new work "NNSmith: Generating Diverse and Valid Test Cases for Deep Learning Compilers" @ASPLOSConf at Session 2A (1pm PT). Welcome to join our talk to learn how we build an automated tool for finding 70+ bugs (and more now!) for emerging DL compilers!

1

8

87

4

0

90

Lingming Zhang

@LingmingZhang

1 year

🥳 Great news, EvalPlus has been accepted at #NeurIPS 2023!! Joint work with my amazing students: @JiaweiLiu_ @steven_xia_ @YuyaoStarling Interested in building/assessing LLMs for code? please consider HumanEval+ for more rigorous evaluation:

GitHub - evalplus/evalplus: Rigourous evaluation of LLM-synthesized code - NeurIPS 2023

Rigourous evaluation of LLM-synthesized code - NeurIPS 2023 - evalplus/evalplus

github.com

AK

@_akhaliq

1 year

Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation extensive evaluation across 14 popular LLMs (including GPT-4 and ChatGPT) demonstrates that HUMANEVAL+ is able to catch significant amounts of previously undetected

18

113

525

3

88

Lingming Zhang

@LingmingZhang

3 months

Agentless leverages a simplistic two-phase approach: localization and repair. It first leverages hierarchical localization to decide the exact locations for editing via LLM. Then, it directly prompts LLM for patch generation using a patch format following Aider.

2

7

70

Lingming Zhang

@LingmingZhang

10 months

Very happy to share that our work "NeuRI: Diversifying DNN Generation via Inductive Rule Inference" received the ACM @sigsoft Distinguished Paper Award 🏆 at @fseconf 2023! If you are around, please come to @JiaweiLiu_ 's talk this Wednesday morning:

Jiawei Liu

@JiaweiLiu_

10 months

Interested in program synthesis for creating random DNNs? and its application on automated testing? Check our new work: “NeuRI: Diversifying DNN Generation via Inductive Rule Inference” with a Distinguished Paper Award @FSEconf ! w/ Jinjun, @YuyaoStarling , @LingmingZhang

5

6

73

5

2

65

Lingming Zhang

@LingmingZhang

2 years

Super grateful to receive an @ASE_conf Distinguished Paper Award🏆 for our work on neural bug seeding! Joint effort w/ Zhao, @JeremyChen92 , Qihao, and Junjie. Check out our paper for more details:

PL/FM/SE at Illinois

@plfmse

2 years

We're very glad to hear that a paper from @plfmse was awarded ACM SIGSOFT Distinguished Paper Award at @ASE_conf . Congratulations to @LingmingZhang for his paper "Learning to Construct Better Mutation Faults" !

1

0

25

2

3

58

Lingming Zhang

@LingmingZhang

2 years

Didn’t realize I got an account 10 years ago until multiple friends told me they found me on twitter. So, here I am:)

3

0

56

Lingming Zhang

@LingmingZhang

5 months

Just landed in the beautiful Lisbon. First conference trip outside of US since 2019. Really looking forward to meeting everyone tomorrow! #icse24

0

53

Lingming Zhang

@LingmingZhang

4 months

Proud advisor moment: Very excited to share that my PhD student @yinlin_deng is one of only two recipients worldwide of the prestigious Two Sigma PhD Fellowship this year!

Siebel School of Computing and Data Science

@IllinoisCDS

4 months

Congratulations to #ILLINOIS CS student Yinlin Deng for winning a Two Sigma PhD Fellowship that will go toward her research in software engineering and machine learning! 🔸🎉🔹 ▶️

1

2

24

2

0

50

Lingming Zhang

@LingmingZhang

5 months

Thanks for the invitation❤️ I'm honored to deliver a keynote talk at @issta_conf 2024. Looking forward to meeting everyone in Vienna this September!

ISSTA Conference

@issta_conf

5 months

📢 We are delighted to announce invited speakers of #ISSTA24 : 🟢 Lingming Zhang, University of Illinois Urbana-Champaign 🟢 Julia Rubin, University of British Columbia 🟢 René Just, University of Washington

0

34

1

50

Lingming Zhang

@LingmingZhang

1 year

🚨Is your code generated by ChatGPT correct?🚨 While more and more LLMs have shown promising results for code generation, how reliable are the evaluation datasets/results? Check out EvalPlus (w/ @JiaweiLiu_ @steven_xia_ @YuyaoStarling ) for the answer:

Is Your Code Generated by ChatGPT Really Correct? Rigorous...

Program synthesis has been long studied with recent approaches focused on directly using the power of Large Language Models (LLMs) to generate code. Programming benchmarks, with curated synthesis...

arxiv.org

AK

@_akhaliq

1 year

Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation extensive evaluation across 14 popular LLMs (including GPT-4 and ChatGPT) demonstrates that HUMANEVAL+ is able to catch significant amounts of previously undetected

18

113

525

3

4

50

Lingming Zhang

@LingmingZhang

1 year

Excited to co-organize (w/ @devanbu , @yiling__LOU , and Lin Tan) the 1st International Workshop on LLM4Code @ICSEconf 24! We hope to create a platform for researchers/practitioners to discuss the latest in #LLM4Code across various fields including SE, NLP, PL, Security, and more.

LLM4Code

@llm4code

1 year

📢🚨 Call for Papers Interested in Large Language Models for Code? Consider submitting to the 1st LLM4Code workshop at @ICSEconf 2024! #LLM4Code 🌟Website: 🌟Deadline: Dec 7, 2023 🌟8-page research paper or 4-page position paper Please help retweet! 🧵

1

15

41

1

9

48

Lingming Zhang

@LingmingZhang

5 months

Very excited to share the 1st @llm4code workshop has attracted 170+ registrations! If you’re at @ICSEconf and interested in LLMs, pls join us on April 20, we have 24 presentations and two keynotes on Code Llama ( @b_roziere ) and StarCoder2 ( @LoubnaBenAllal1 )! #icse24 #llm4code

1

5

45

Lingming Zhang

@LingmingZhang

10 months

Thank you, @WizardLM_AI ! Also, thanks for the great Evol-Instruct and WizardCoder work, which truly inspired Magicoder!! 🚀🚀

WizardLM

@WizardLM_AI

10 months

🔥🔥🔥Incredible work! OSS-Instruct + 🧙‍♂️Evol-Instruct = 🎩Magicoder Paper: Code:

12

103

583

3

5

33

Lingming Zhang

@LingmingZhang

3 months

Since Devin, more and more AI SW agents are being proposed almost every single day, achieving impressive performance on SWE-bench. But do we really need such complex agents to solve GitHub issues? Can agentless solutions approach their performance? Our paper aims to explore this.

2

31

Lingming Zhang

@LingmingZhang

2 years

We are HIRING! @IllinoisCS welcomes applications from exceptional candidates in all areas for full-time tenure-track faculty positions at all levels (Assistant/Associate/Full Professor). Applications received by Nov 15, 2022 will receive full consideration.

CS Faculty Jobs

@csfacultyjobs

2 years

Professor (Open Rank) University of Illinois Computer Science, University of Illinois Urbana-Champaign, Urbana-Champaign, Illinois

0

2

8

0

8

29

Lingming Zhang

@LingmingZhang

7 months

Want to learn about recent progress in LLMs for code? Join us at LLM4Code 2024 @ICSEconf !

LLM4Code

@llm4code

7 months

Introducing keynote speakers of #LLM4Code @ICSEconf ! 🔊"Open development of Large Language Models for code with BigCode and StarCoder2" by @LoubnaBenAllal1 from @HuggingFace 🔊"Code Llama: Open Foundation Models for Code" by @b_roziere from @Meta Don't miss these great talks!

1

7

28

0

27

Lingming Zhang

@LingmingZhang

2 years

Back from @ISSREConf , it is amazing to see how the conference goes even half of the authors cannot travel. Thanks all who made this possible. Special thanks to general chairs @BoCukic and Katerina, as well as my PC co-chair @FabrizioPastore , super proud to be part of the team!

1

0

27

Lingming Zhang

@LingmingZhang

10 months

📢Pls submit your exciting work to @llm4code , the ddl is Dec 7th 2023! 🚨We provide both archival and non-archival options. Pls check our website for more details:

Home

ICSE 2024 Workshop on Large Language Models for Code

llm4code.github.io

LLM4Code

@llm4code

10 months

🚀10 days left to submit to the LLM4Code workshop! 🧵Some updates:

1

2

10

0

2

27

Lingming Zhang

@LingmingZhang

1 year

🚨Wonder how recent LLMs for code perform on the rigorous evaluation dataset of HumanEval+? #LLM4Code Check out our EvalPlus Leaderboard! (all credits to @JiaweiLiu_ @steven_xia_ @YuyaoStarling )

Jiawei Liu

@JiaweiLiu_

1 year

Introducing the EvalPlus leaderboard! 🔥28 models have been evaluated on coding HumanEval & HumanEval+ 🔥7B CodeLlama outperforms ~16B models e.g. StarCoder&CodeGen 🔥Phind-CodeLlama-34B-v2 and WizardCoder-Python-34B-V1 as open models both beat ChatGPT 🧵

4

18

149

0

1

24

Lingming Zhang

@LingmingZhang

7 months

🤔 Curious about how the latest Claude-3 Opus compares with GPT-4 in programming beyond HumanEval? Check out our head-to-head comparison using the enhanced EvalPlus dataset! #Claude3 #GPT4 #LLM4Code

Jiawei Liu

@JiaweiLiu_

7 months

Sure! We just evaluated the fresh Claude-3 family from @AnthropicAI on the EvalPlus leaderboard () -- now we have rankings of 75 models! 🚀 On HumanEval+, Claude-3 Opus, as the best Claude-3 model, nails 🥉 #3 , topping all open models and keeping up with

14

28

175

1

2

24

Lingming Zhang

@LingmingZhang

3 months

While agent-based software development is highly promising, we believe it's time to step back and rethink its critical design/evaluation besides publishing more and more agents. We hope Agentless can help reset the baseline, starting point, and horizon for future SW agents.

3

1

23

Lingming Zhang

@LingmingZhang

5 months

@yinlin_deng is presenting her FuzzGPT work at #icse24 . FuzzGPT aims to guide LLMs (which were trained to generate more natural and common info) to generate exceptional and edge-case tests for more effective fuzzing. 📝

0

4

24

Lingming Zhang

@LingmingZhang

10 months

📢📢Excited to announce the addition of MBPP+ to the EvalPlus family! 🔥EvalPlus has been installed 5,000+ times. Please try it out if you care about rigorous evaluation of #LLM4Code ! 🚀EvalPlus leaderboard () has also been updated w/ newest LLMs!

Jiawei Liu

@JiaweiLiu_

10 months

In the past 6-mon release of HumanEval+ we have been improving its toolchain usability and dataset quality from v0.1.0 to v0.1.7 releases. 🔥 Now we release MBPP+, a new benchmark in EvalPlus v0.2.0: 🧵

2

7

53

1

0

23

Lingming Zhang

@LingmingZhang

3 months

We have compared Agentless with all SW agents, including open-source and commercial ones. Surprisingly, Agentless can outperform all existing open-source agents with much less cost! Moreover, Agentless even provides more unique fixes than the best commercial tool (LingmaAgent)!

1

2

22

Lingming Zhang

@LingmingZhang

2 years

@moarbugs Looks like a very interesting course!! I regularly teach similar topics at Illinois, feel free to check if our paper list this semester would be helpful:

1

22

Lingming Zhang

@LingmingZhang

2 years

Steven ( @steven_xia_ ) will soon present AlphaRepair, a cloze-style (or infilling-style) approach for automated program repair @FSEconf (15:00-15:15 SRC LT 53):

2

22

Lingming Zhang

@LingmingZhang

2 years

Working on Deep Learning, interested in DL system reliability? Yinlin and @cy1yang will soon present DeepREL (Fuzzing DL Libraries via Automated Relational API Inference) at @FSEconf (Nov 13 9-10:30 CST). It has helped find 100+ bugs for popular libraries ( @TensorFlow / @PyTorch )!

FSE 2025

@FSEconf

2 years

"DeepREL leverages API relations for fuzzing popular DL libraries, and has found 100+ confirmed bugs for PyTorch/TensorFlow."" - Yinlin Deng*, Chenyuan Yang*( @cy1yang ), Anjiang Wei( @AnjiangW ), Lingming Zhang

0

3

11

1

21

Lingming Zhang

@LingmingZhang

5 months

@steven_xia_ is presenting fuzz4all, the first universal fuzzer via large language models, at #icse24 ! Joint work with Matteo Paltenghi, Jiale Tian, and Michael Pradel. 🧑‍💻 📝

0

2

19

Lingming Zhang

@LingmingZhang

3 months

We also manually inspected the great SWE-bench dataset, and found some problem descriptions with ground-truth patches or misleading solutions, as well as problems with insufficient descriptions. We plan to help further mitigate such issues, please read our paper for details.

1

0

17

Lingming Zhang

@LingmingZhang

5 months

Nice view from Luis de Freitas Branco, where @llm4code 2024 will happen in an hour 🚀 #icse24

0

1

18

Lingming Zhang

@LingmingZhang

3 months

All credits go to my awesome students @steven_xia_ @yinlin_deng and Soren Dunn from @IllinoisCS . We all look forward to your feedback and suggestions!

3

1

16

Lingming Zhang

@LingmingZhang

2 years

@RiggerManuel @StarRocksLabs @CelerData Very interesting article! I highly agree that ChatGPT (or other LLMs) can at least complement existing DB fuzzers (and more). Feel free to check out more discussion in our recent work on LLM4Fuzz:

Lingming Zhang

@LingmingZhang

2 years

🚨LLMs are Zero-Shot Fuzzers! Excited to share our TitanFuzz🤖 work @issta23 : #LLMs can be directly applied for both generative and mutative fuzzing, while being fully automated, generalizable, and applicable to domains challenging for traditional fuzzers (such as DL systems)🧵

2

46

208

0

2

16

Lingming Zhang

@LingmingZhang

4 months

Congratulations Wenxi! UVA CS is very lucky to have you!

Wenxi Wang

@WenxiWang4

4 months

Excited to announce that I will be joining @UVA CS as an Assistant Professor this summer!

41

28

990

0

16

Lingming Zhang

@LingmingZhang

4 months

📢 The @ASE_conf research track submission ddl has just passed. Now it's time to focus on your demo papers! Submit your papers to ASE 2024 Demo Track by June 26 AoE: 📝 🔗 … @mboehme_ and I look forward to your submissions!

0

2

15

Lingming Zhang

@LingmingZhang

1 year

Our initial results show that the performance of widely studied commercial (like ChatGPT/GPT-4) and open-source (like CodeGen) models can drop substantially with more rigorous testing, largely affecting the result analysis for almost all recent work on LLM-based code generation!

1

0

11

Lingming Zhang

@LingmingZhang

5 months

@llm4code 2024 gets a full house! Join us at room Luis de Freitas Branco! #icse24

0

1

12

Lingming Zhang

@LingmingZhang

10 months

Thank you @_philschmid for the great thread about @Magicoder_AI ! We originate from and are all for open source❤️

Philipp Schmid

@_philschmid

10 months

Code completion tools like @github Copilot are used by over 1 million developers, helping them code 55% faster. 🧑🏻‍💻 Exciting to see open-source innovations like @Magicoder_AI OSS-Instruct outperforming @OpenAI GPT-3.5 and @GoogleDeepMind Gemini Ultra. details below ⬇️

8

27

124

0

12

Lingming Zhang

@LingmingZhang

2 years

@WenhuChen I’d say every paper should be evaluated by itself, regardless of who are the authors and whether they were, are, or will be productive or not.

1

0

12

Lingming Zhang

@LingmingZhang

10 months

To my knowledge, at least another paper (by @TaliaRinger ) from @plfmse @IllinoisCS also received the distinguished paper award at FSE 2023!

0

4

8

Lingming Zhang

@LingmingZhang

2 years

Approach: We first use a generative LLM ( #Codex , now can be #ChatGPT :) to generate high-quality seed programs. Then, we leverage an evolutionary fuzzing loop with an infilling LLM ( #InCoder ) to further mutate seed programs to generate more diverse input DL programs for fuzzing.

1

0

8

Lingming Zhang

@LingmingZhang

1 year

@mboehme_ @profbodden @mchri5taki5 @kayseesee @moarbugs @AndreasZeller Thanks all for the great event, really exciting to see people in person!

0

8

Lingming Zhang

@LingmingZhang

3 months

@gneubig @xingyaow_ Thanks a lot Graham! I couldn't agree more, and we should definitely chat about how to synergize OpenDevin and Agentless:)

0

8

Lingming Zhang

@LingmingZhang

2 years

@Reyhaneh @NSF @plfmse @IllinoisCS Congratulations Reyhan!!

0

7

Lingming Zhang

@LingmingZhang

2 years

Problem: DL systems (op-level libraries & DL compilers) are crucial, but challenging to test: it’s hard to synthesize valid input programs that satisfy both input language (e.g., Python) syntax/semantics and the tensor/operator constraints for constructing computation graphs.

1

0

7

Lingming Zhang

@LingmingZhang

2 years

This work won’t be possible without my students/interns @plfmse @IllinoisCS , including @yinlin_deng , @steven_xia_ , Haoran, and @cy1yang . Also big thanks to @Google , @Meta , and Kwai Inc. for supporting our work on ML system reliability.

0

6

Lingming Zhang

@LingmingZhang

2 years

@cestlemieux @ICSEconf Very neat idea, and look forward to it!

0

6

Lingming Zhang

@LingmingZhang

2 years

Results: TitanFuzz achieves 30.38% / 50.84% higher code coverage than state-of-the-art fuzzers on TensorFlow / PyTorch. Furthermore, TitanFuzz is able to detect 65 bugs, with 41 already confirmed as new bugs, including multiple high-priority bugs and security vulnerabilities!

1

0

5

Lingming Zhang

@LingmingZhang

2 years

Glad to know that @OpenAI released #ChatGPT API, and people are talking about the potential of LLMs for fuzzing. So it may be a good time to share the preprint of TitanFuzz:

Large Language Models are Zero-Shot Fuzzers: Fuzzing Deep-Learning...

Detecting bugs in Deep Learning (DL) libraries (e.g., TensorFlow/PyTorch) is critical for almost all downstream DL systems in ensuring effectiveness/safety for end users. Meanwhile, traditional...

arxiv.org

0

5

Lingming Zhang

@LingmingZhang

1 month

Huge shoutout to my amazing students who did all the hard work on Agentless😺: @steven_xia_ @yinlin_deng @SorenD27261

1

0

5

Lingming Zhang

@LingmingZhang

6 months

@baishakhir Congratulations!!!

0

3

Lingming Zhang

@LingmingZhang

10 months

@Reyhaneh Wow, congrats to both of you!!

1

0

1

Lingming Zhang

@LingmingZhang

4 months

@spanichella @aitorarrieta @Greg4cr @AISTWorkshop @icstconference Thanks for the great workshop and the invitation!

0

4

Lingming Zhang

@LingmingZhang

1 month

According to the extensive evaluation by OpenAI, Agentless😺 stays as the Top-1 open-source AI software engineer on all three benchmarks, including SWE-bench, SWE-bench Lite, and SWE-bench Verified!

1

0

3

Lingming Zhang

@LingmingZhang

3 months

@alxfazio @EMostaque Great question! In Agentless, LLMs are just used to process plain texts and they are not allowed for autonomous tool use or decision/planning.

1

0

4

Lingming Zhang

@LingmingZhang

11 days

@mboehme_ @PeteMarkowsky We did something along this direction in TitanFuzz. Not sure if it helps:

Large Language Models are Zero-Shot Fuzzers: Fuzzing Deep-Learning...

Detecting bugs in Deep Learning (DL) libraries (e.g., TensorFlow/PyTorch) is critical for almost all downstream DL systems in ensuring effectiveness/safety for end users. Meanwhile, traditional...

arxiv.org

0

4

Lingming Zhang

@LingmingZhang

2 years

@tianyin_xu @IllinoisCS @plfmse Cannot agree more! Interview @saikatdutta2012 please!

0

1

4

Lingming Zhang

@LingmingZhang

2 years

Generalizability: TitanFuzz can be well generalized to test arbitrary software systems (libraries/compilers/interpreters/DB). To our knowledge, this is the first end-to-end practical fuzzer directly based on #LLM4Code . We hope TitanFuzz can stimulate more work on #LLM4Fuzz .

1

0

4

Lingming Zhang

@LingmingZhang

3 months

Only three days left until the ASE'24 Demo deadline! We look forward to your submissions and seeing you in Sacramento this October! @ASE_conf @mboehme_

Lingming Zhang

@LingmingZhang

4 months

📢 The @ASE_conf research track submission ddl has just passed. Now it's time to focus on your demo papers! Submit your papers to ASE 2024 Demo Track by June 26 AoE: 📝 🔗 … @mboehme_ and I look forward to your submissions!

0

2

15

0

1

4

Lingming Zhang

@LingmingZhang

1 year

@TaliaRinger Thanks for asking. Still not sure if I could make it as I have some family medical issue to take care recently. But @steven_xia_ , @yinlin_deng , and Yicheng are already there:)

3

0

4

Lingming Zhang

@LingmingZhang

2 years

@julia_rubin1 @ASE_conf Thank you and everyone on the ASE’22 team for the great conference!!

0

1

4

Lingming Zhang

@LingmingZhang

2 years

AlphaRepair opens a new dimension for cloze-style repair via directly querying large language models. Compared with prior learning-based repair work based on NMT, AlphaRepair does not need any fine-tuning on historical bug fixes and can be easily adopted for multiple languages

1

0

4

Lingming Zhang

@LingmingZhang

7 months

@ArjunGuha @ellev3n11 @linguistcarolyn This has been an unparalleled experience for us, and we are super proud to be part of the team❤️

0

4

Lingming Zhang

@LingmingZhang

1 year

@DomSteinhoefel Thank you Dominic for sharing! This is a joint work with Michael Pradel’s group from Stuttgart.

1

0

4

Lingming Zhang

@LingmingZhang

2 months

@gneubig Great work! Cannot wait to read the full paper!

0

3

Lingming Zhang

@LingmingZhang

2 years

@taoxiease Congratulations!!

0

3

Lingming Zhang

@LingmingZhang

5 months

@margaretstorey @rmaranhao @AbhikRoychoudh1 Thank you all for the great conference!

0

3

Lingming Zhang

@LingmingZhang

2 years

Check out our paper () and code (). Also, feel free to talk to @steven_xia_ at #FSE22 !

0

3

Lingming Zhang

@LingmingZhang

3 months

@gneubig @rbren_dev @xingyaow_ Congrats Graham!! Cannot wait to see what you will build!

0

3

Lingming Zhang

@LingmingZhang

2 years

“…welcomes applications in all areas…particularly interested in candidates whose primary research areas include (1) AI and Advanced Technology for the Digital Transformation of Education; (2) CPS; (3) Cybersecurity; (4) Software Engineering; and (5) Theoretical CS.”

CS Faculty Jobs

@csfacultyjobs

2 years

Faculty Position – Tenured/Tenure-Track Faculty Positions, North Carolina State University, Raleigh, North Carolina

0

2

1

0

3

Lingming Zhang

@LingmingZhang

2 years

@tianyin_xu @JiaweiLiu_ @IllinoisCS @ASPLOSConf Thank you @tianyin_xu and @ramnatthan for hosting Jiawei’s talk!

1

0

3

Lingming Zhang

@LingmingZhang

3 months

@RickLamers Cannot agree more! We are by no means against agents, but we really hope Agentless can help inspire more open, replicable, and interpretable agentic/agentless approaches in the near future.

0

3

Lingming Zhang

@LingmingZhang

1 year

@samth @TaliaRinger We didn’t mean to embarrass anyone:) We truly appreciate existing datasets. Through this, we just want to show that besides building new code models everyday, it can be equally important to standardize/improve the evaluation datasets/platforms to truly understand where we are.

1

0

3

Lingming Zhang

@LingmingZhang

6 months

@b_vasilescu @daye_nam Congratulations to both of you!!

0

2

Lingming Zhang

@LingmingZhang

6 months

@DrWingLam Great news, congratulations!!

1

0

2

Lingming Zhang

@LingmingZhang

1 year

@tianyin_xu @IllinoisCS Thank you Tianyin!

0

3

Lingming Zhang

@LingmingZhang

1 year

@elgreco_winter @ManlingLi_ Great news, congratulations to both of you!!

0

3

Lingming Zhang

@LingmingZhang

1 year

@samth @TaliaRinger Great comment! The framework is general, and it is absolutely interesting to explore more generation strategies. We did try pure random, and are now working on including other strategies (such as feedback-driven generation) and datasets.

0

3

Lingming Zhang

@LingmingZhang

2 years

Key insight: Modern LLMs include numerous code snippets invoking DL operators in massive training corpora, and thus may implicitly learn the intricate language, tensor, and operator constraints for generating valid input DL programs for fuzzing.

1

0

3

Lingming Zhang

@LingmingZhang

5 months

You can still join us! #icse24

LLM4Code

@llm4code

5 months

We now have our first invited keynote by @b_roziere “Code Llama: Open Foundation Models for Code” 🦙

2

4

25

0

3

Lingming Zhang

@LingmingZhang

3 months

@OfirPress Thanks Ofir! I completely agree, and this is also a key motivation for our project. The agent world largely remains a mystery, and we aim to develop open, simple, and replicable techniques that could benefit both agents and non-agents.

1

0

2

Lingming Zhang

@LingmingZhang

5 months

@tianyin_xu @ASPLOSConf @charith_mendis @josep_torrellas what team are you on?:)

1

0

2

Lingming Zhang

@LingmingZhang

4 months

@SWATLab Thank you all for the great event!

0

2

Lingming Zhang

@LingmingZhang

2 years

@andrianmarcus @GMUCompSci @DrBrittJay @dsrosenblum @JeffOffutt @kevpmo @ThomasLaToza @winglam228 Cannot believe it’s already nine years, time really flies… Big congratulations to both you and @GMUCompSci !

0

2

Lingming Zhang

@LingmingZhang

2 years

@AndreasZeller @ERC_Research Congratulations, very exciting project!!

0

2

Lingming Zhang

@LingmingZhang

2 years

@davidlo2015 @BoCukic @ISSREConf @FabrizioPastore Thank you @davidlo2015 for your support! I really enjoyed your Test of Time Award talk, and congratulations again!!

0

2

Lingming Zhang

@LingmingZhang

16 days

@hengjinlp @Glaciohound Congratulations!!!

1

0

2

Lingming Zhang

@LingmingZhang

2 years

@deepakv91 @ggn_dp_sngh @Zikun__Liu @IllinoisCS @QCOMResearch Great news! Congratulations!

0

2

Lingming Zhang

@LingmingZhang

2 years

@davidlo2015 @BoCukic @aitorarrieta @ISSREConf Congratulations and very well deserved!!

0

2

Lingming Zhang

@LingmingZhang

2 years

@PradyumnaShome @tianyin_xu @ragibhasan @IllinoisCS The later, so it’s 40k for 12 months if the student stay for the summer

0

2

Lingming Zhang

@LingmingZhang

3 months

@OfirPress @steven_xia_ Thank you for the great SWE-bench project which inspires this vibrant community on AI software engineers! We are very happy to chat about our findings in detail and hope to contribute to SWE-bench.

0

2

Lingming Zhang

@LingmingZhang

2 years

@JAldrichPL Thanks for contacting ICSE. The explanation also makes some sense, but I really would not attend a conference just because I need to pay the registration fee, I attend because I want to talk to the lovely people, share our work, and more importantly when I’m available.

0

2

Lingming Zhang

@LingmingZhang

4 months

@swarat @NSF Huge congratulations!!

0

1

Lingming Zhang

@LingmingZhang

2 years

@RandomlyWalking @miltos1 Congratulations!