Junayed Mahmud

What if we could run Transformer models without worrying about context length? With our new work Unlimiformer, you can jailbreak your current models to use unlimited length inputs! Preprint: arxiv.org/abs/2305.01625 Thread 🧵 (1/6)

The phenomenal teams from Google Research’s Brain and @DeepMind have made many of the seminal research advances that underpin modern AI, from Deep RL to Transformers. Now we’re joining forces as a single unit, Google DeepMind, which I’m thrilled to lead! dpmd.ai/announcing-goo…

Codex and other LLMs generate a lotta code, for a lotta people. But do they help avoid coding errors? We study this question (with @kjnlp, @Toufique_Ahmed_ , and @emilynguist ) in our MSR 2023 paper, using the SStubs4J dataset..we find...

@ojcchar @NSF Congratulations!!🎉

🎉Oh my…! So thrilled to announce that I've received an @NSF CAREER award!🏆 Can't wait to transform the way developers make informed decisions for changing software!👨‍💻🚀 tinyurl.com/oscareer Huge thanks to my students, colleagues, mentors, friends, wife, and @NSF!

Sharing the preprint of our @FSEconf paper. We propose BURT, a task-oriented chatbot for interactive bug reporting: tinyurl.com/burtbr BURT video: tinyurl.com/burtvid Package: tinyurl.com/burtrepo Looking forward to discussing this work at @FSEconf in person!

Today at #GoogleIO @sundarpichai showed some examples of the capabilities of the PaLM 540B language model. For example, you can prompt the model with: "I will ask a question in Bengali and get English and Bengali answers" And then give it two examples of this behavior. (cont)

Open-domain dialog—where a model converses about any topic—is a key challenge for language models. Learn about LaMDA, a project to build dialog models that are more safe, grounded, high quality, and in line with our Responsible AI Principles ↓ goo.gle/3KJ2oXJ

The kind of folks at @MetaflowOSS have given me a managed infra to play around with today. If you’d like to see an intuitive take on end to end MLOps in pure Python. Join my stream tonight at 4:00 PM PST. Twitch.tv/marksaroufim

Alhamdulillah. My first conference paper, presented in IEEE UEMCON 2021, "Firefly Algorithm Based Optimized PID Controller for Stability Analysis of DC-DC SEPIC Converter" has been published in IEEE Xplore. Link for the paper: ieeexplore.ieee.org/document/96665… researchgate.net/publication/35…

🎉 Colab's markdown cells now include 💃emoji🕺 and Φ LaTeX ψ support! Start typing `:` followed by an emoji's name to select it from the drop-down menu, and `\` to trigger LaTeX completions 🎉

Expert developers have valuable internal strategies to solve programming problems. We built HowToo: A Platform for Sharing, Finding, and Using Programming Strategies presented @vlhcc #vlhcc20 Check out the video presenting our paper: youtube.com/watch?v=EaTosp…

This is a great event! I participated last year and I think the students got a lot of out of it! If you are a prospective Ph.D. student in CS, consider joining!!

Colab is now available on GCP Marketplace! This gives you the ability to connect Colab to a custom GCE VM that you can configure as you wish. console.cloud.google.com/marketplace/pr…

Great news, no GRE needed in Fall 2022 for PhD and MS program in Computer Science! Spread the word and join Virginia's most diverse and most innovative institution according to U.S. News & World Report bit.ly/3mLIdge

We're excited to open-source Mistral 🚀 - a codebase for accessible large-scale LM training, built as part of Stanford's CRFM (crfm.stanford.edu). We're releasing 10 GPT-2 Small & Medium models with different seeds & 600+ checkpoints per run! #resources" target="_blank" rel="nofollow noopener">github.com/stanford-crfm/… [1/4]

📢 Here are the key dates for #acl2022nlp: - 15 Nov 2021: latest deadline by which papers need to be submitted to @ReviewAcl - 7 Jan 2022: deadline for authors to submit reviewed papers, reviews, and meta-review to #acl2022nlp - 20 Feb 2022: notifications 2022.aclweb.org/calls

Another step forward to automated code generation...out new paper on multimodal code edit will appear in @ASE_conf (arxiv.org/pdf/2108.06645…)

Some days ago I gave a talk at the UNESCO language group (hosted @TranslationCom1) on how modern NLP changes the data requirements for translation systems (eg for communities in the process of digitizing their language) Video:youtu.be/-dLMNOb_Vmw Slides:bit.ly/3k6ZjUC

Are you curious about progress of research on Deep Learning for Software Engineering?? If so, then check out the recording of my @TheOfficialACM Webinar that offers an introduction to and retrospective of the field!! Link to the 🎥 below! youtube.com/watch?v=l_wY0t…

@PreethaChatterj Thank you @PreethaChatterj

@JunayedMahmud10 Congratulations!!

Are you curious to know about the model effectiveness & errors for the code to comment translation task? Check out our #NLP4Prog paper. We evaluate the ability of three popular models to predict natural language descriptions and manually categorize the errors for the models.