Sarah Pratt Personal Webpage

<!DOCTYPE html> <html lang="en"> <head> <meta charset="utf-8" /> <meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no" /> <meta name="description" content="" /> <meta name="author" content="" /> <title>Sarah Pratt Personal Webpage</title> <link rel="icon" type="image/x-icon" href="assets/img/favicon.ico" />  <script src="https://use.fontawesome.com/releases/v5.15.1/js/all.js" crossorigin="anonymous"></script>  <link href="https://fonts.googleapis.com/css?family=Saira+Extra+Condensed:500,700" rel="stylesheet" type="text/css" /> <link href="https://fonts.googleapis.com/css?family=Muli:400,400i,800,800i" rel="stylesheet" type="text/css" />  <link href="css/styles.css" rel="stylesheet" /> <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/4.7.0/css/font-awesome.min.css"> </head> <body id="page-top">  <nav class="navbar navbar-expand-lg navbar-dark bg-primary fixed-top" id="sideNav"> <a class="navbar-brand js-scroll-trigger" href="#page-top"> <span class="d-block d-lg-none">Sarah Pratt</span> <span class="d-none d-lg-block"><img class="img-fluid img-profile rounded-circle mx-auto mb-2" src="assets/img/profile.jpg" alt="" /></span> </a> <button class="navbar-toggler" type="button" data-toggle="collapse" data-target="#navbarSupportedContent" aria-controls="navbarSupportedContent" aria-expanded="false" aria-label="Toggle navigation"><span class="navbar-toggler-icon"></span></button> <div class="collapse navbar-collapse" id="navbarSupportedContent"> <ul class="navbar-nav"> <li class="nav-item"><a class="nav-link js-scroll-trigger" href="#about">About</a></li> <li class="nav-item"><a class="nav-link js-scroll-trigger" href="#experience">Publications</a></li> <li class="nav-item"><a class="nav-link js-scroll-trigger" href="#talks">Talks</a></li> <li class="nav-item"><a class="nav-link js-scroll-trigger" href="#education">Education</a></li> <li class="nav-item"><a class="nav-link js-scroll-trigger" href="#teaching">Teaching</a></li>  </ul> </div> </nav>  <div class="container-fluid p-0">  <section class="resume-section" id="about"> <div class="resume-section-content"> <h1 class="mb-0"> <span class="text-primary">Sarah Pratt</span> </h1> <div class="subheading mb-0"> Third Year PhD, University of Washington </div> <div class="mb-5">spratt3 [at] cs [dot] washington [dot] edu </div> <p class="lead mb-5"> I am a PhD student working with <a href="https://homes.cs.washington.edu/~ali/">Ali Farhadi</a>. I am interested in Machine Learning and Computer Vision. My past work is in vision and language. Previously, I worked on the <a href="https://prior.allenai.org/">PRIOR team</a> at <a href="https://allenai.org/">The Allen Institute for AI</a> under the mentorship of <a href="https://anikem.github.io/">Ani Kembhavi</a>, <a href="https://lucaweihs.github.io/">Luca Weihs</a>, and <a href="http://markyatskar.com/">Mark Yatskar</a>. I completed my undergraduate degree in Applied Mathematics and Computer Science at Brown University in Providence, RI with academic advisor <a href="http://www.dam.brown.edu/people/cklivans/">Caroline Klivans</a>. </p> <p> <a href="https://twitter.com/sarahmhpratt"> <img border="0" alt="W3Schools" src="assets/img/t.png" width="50" height="50"> </a> <a href="https://github.com/sarahpratt"> <img border="0" alt="W3Schools" src="assets/img/github.png" width="50" height="50"> </a> <a href="https://scholar.google.com/citations?user=cNXj4QwAAAAJ&hl=en"> <img border="0" alt="W3Schools" src="assets/img/scholar.png" width="45" height="45"> </a> </p> </div> </section> <hr class="m-0" />  <section class="resume-section" id="experience"> <p> <div class="resume-section-content"> <h2 class="mb-5">Publications</h2> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <img border="0" alt="W3Schools" src="assets/img/cupl_thumbnail.png" width="270" height="270"> &nbsp&nbsp&nbsp&nbsp&nbsp&nbsp <div class="flex-grow-1"> <h3 class="mb-0"> What does a platypus look like? Generating customized prompts for zero-shot image classification </h3> <div class="subheading mb-0"> 2023 &nbsp&nbsp [<a href="assets/cupl.pdf"> updated pdf </a>] &nbsp&nbsp [<a href="https://arxiv.org/abs/2209.03320"> arxiv </a>] &nbsp&nbsp [<a href="https://github.com/sarahpratt/CuPL"> code </a>]</div> <p> <b>Sarah Pratt</b>, <a href="https://iancovert.com/">Ian Covert</a>, <a href="https://rosanneliu.com/">Rosanne Liu</a>, <a href="https://homes.cs.washington.edu/~ali/">Ali Farhadi</a> </p> <p> This work combines open vocabulary models with large language models (LLMs) to create Customized Prompts via Language models (CuPL, pronounced "couple"). In particular, we leverage the knowledge contained in LLMs in order to generate many descriptive sentences that are used to perform zero-shot image classification with open vocabulary models. We find that this straightforward and general approach improves accuracy on a range of zero-shot image classification benchmarks, including over one percentage point gain on ImageNet. Finally, this simple baseline requires no additional training and remains completely zero-shot. </p> </div> </div> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <img border="0" alt="W3Schools" src="assets/img/introspective_thumbnail.png" width="270" height="270"> &nbsp&nbsp&nbsp&nbsp&nbsp&nbsp <div class="flex-grow-1"> <h3 class="mb-0"> The Introspective Agent: Interdependence of Strategy, Physiology, and Sensing for Embodied Agents </h3> <div class="subheading mb-0"> 2022 &nbsp&nbsp [<a href="https://arxiv.org/abs/2201.00411"> arxiv </a>] &nbsp&nbsp [<a href="https://github.com/sarahpratt/introspective"> code </a>]</div> <p> <b>Sarah Pratt</b>, <a href="https://lucaweihs.github.io/">Luca Weihs</a>, <a href="https://homes.cs.washington.edu/~ali/">Ali Farhadi</a></p> <p> While traditional embodied agents manipulate an environment to best achieve a goal, we argue for an introspective agent, which considers its own abilities in the context of its environment. We show that different environments yield vastly different optimal designs, and increasing long-term planning is often far less beneficial than other improvements, such as increased physical ability. </p> </div> </div> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <img border="0" alt="W3Schools" src="assets/img/hs_thumnail.png" width="270" height="270"> &nbsp&nbsp&nbsp&nbsp&nbsp&nbsp <div class="flex-grow-1"> <h3 class="mb-0"> Learning Generalizable Visual Representations via Interactive Gameplay </h3> <div class="subheading mb-0">Oral Presentation at ICLR 2021 &nbsp&nbsp [<a href="https://arxiv.org/abs/1912.08195"> arxiv </a>]</div> <p> <a href="https://lucaweihs.github.io/">Luca Weihs</a>, <a href="https://anikem.github.io/">Ani Kembhavi</a>, <a href="https://ehsanik.github.io/">Kiana Ehsani</a>, <b>Sarah Pratt</b>, Winson Han, Alvaro Herrasti, Eric Kolve, Dustin Schwenk, <a href="https://roozbehm.info">Roozbeh Mottaghi</a>, <a href="https://homes.cs.washington.edu/~ali/">Ali Farhadi</a></p> </div> </div> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <img border="0" alt="W3Schools" src="assets/img/GSR_thumbnail.png" width="270" height="310"> &nbsp&nbsp&nbsp&nbsp&nbsp&nbsp <div class="flex-grow-1"> <h3 class="mb-0">Grounded Situation Recognition </h3> <div class="subheading mb-0">Spotlight at ECCV 2020 &nbsp&nbsp [<a href="https://arxiv.org/abs/2003.12058"> arxiv </a>] &nbsp&nbsp [<a href="https://github.com/allenai/swig"> code </a>] &nbsp&nbsp [<a href="https://vision-explorer.allenai.org/grounded_situation_recognition"> demo </a>]</div> <p> <b>Sarah Pratt</b>, <a href="http://markyatskar.com/"> Mark Yatskar </a>, <a href="https://lucaweihs.github.io/">Luca Weihs</a>, <a href="https://homes.cs.washington.edu/~ali/">Ali Farhadi</a>, <a href="https://anikem.github.io/">Ani Kembhavi</a></p> <p>Situation Recognition is the task of recognizing the activity happening in an image, the actors and objects involved in this activity, and the roles they play. Semantic roles describe how objects in the image participate in the activity described by the verb. While situation recognition addresses <i>what</i> is happening in an image, <i>who</i> is playing a part in this and <i>what</i> their roles are, it does not address a critical aspect of visual understanding: <b>where</b> the involved entities lie in the image. We address this shortcoming and present Grounded Situation Recognition (GSR), a task that builds upon situation recognition and requires one to not just identify the situation observed in the image but also visually ground the identified roles within the corresponding image.</p> </div> </div> </div> </p> </section>  <section class="resume-section" id="talks"> <p> <div class="resume-section-content"> <h2 class="mb-5">Talks</h2> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <img border="0" alt="W3Schools" src="assets/img/cvpr2023.png" width="270" height="270"> &nbsp&nbsp&nbsp&nbsp&nbsp&nbsp <div class="flex-grow-1"> <h3 class="mb-0"> Prompt Generation for Zero-Shot Image Classification </h3> <p>Detailing paper "What does a platypus look like? Generating customized prompts for zero-shot image classification" as well as other relavent works.</p> <div class="subheading mb-0">[<a href="assets/pratt_cvpr2023.pdf"> slides </a>]</div> </div> </div> </div> </p> </section> <hr class="m-0" />  <section class="resume-section" id="education"> <div class="resume-section-content"> <h2 class="mb-5">Education</h2> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <div class="flex-grow-1"> <h3 class="mb-0">University of Washington</h3> <div class="subheading mb-0">Third year PhD student</div> <div>Advisor: Ali Farhadi</div> </div> <div class="flex-shrink-0"><span class="text-primary">September 2020 - Current</span></div> </div> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <div class="flex-grow-1"> <h3 class="mb-0">Brown University</h3> <div class="subheading mb-0">Bachelor of Science - Honors</div> <div>Joint Concentration in Computer Science and Applied Mathemetics</div> </div> <div class="flex-shrink-0"><span class="text-primary">September 2014 - May 2018</span></div> </div> </div> </section> <hr class="m-0" />  <section class="resume-section" id="teaching"> <div class="resume-section-content"> <h2 class="mb-5">Teaching</h2> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <div class="flex-grow-1"> <h3 class="mb-0">Teaching Assistant</h3> <div class="subheading mb-0">cse493: Deep Learning</div> <div>University of Washington</div> </div> <div class="flex-shrink-0"><span class="text-primary">Spring 2018</span></div> </div> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <div class="flex-grow-1"> <h3 class="mb-0">Head Teaching Assistant</h3> <div class="subheading mb-0">cs022: Discrete Math and Probability</div> <div>Brown University</div> </div> <div class="flex-shrink-0"><span class="text-primary">Spring 2018</span></div> </div> <div class="d-flex flex-column flex-md-row justify-content-between mb-5"> <div class="flex-grow-1"> <h3 class="mb-0">Undergraduate Teaching Assistant</h3> <div class="subheading mb-0">cs015: Introduction to Object Oriented Programming</div> <div class="subheading mb-0">cs022: Discrete Math and Probability</div> <div>Brown University</div> </div> <div class="flex-shrink-0"><span class="text-primary">Fall 2016, Spring 2017</span></div> </div> <div class="d-flex flex-column flex-md-row justify-content-between"> <div class="flex-grow-1"> <h3 class="mb-0">Nonprofit tutoring</h3> <div class="subheading mb-0">Yleana Leadership Academy</div> </div> <div class="flex-shrink-0"><span class="text-primary">Summer 2015</span></div> </div> </div> </section> <hr class="m-0" /> <hr class="m-0" />   </div>  <script src="https://cdnjs.cloudflare.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script> <script src="https://cdn.jsdelivr.net/npm/bootstrap@4.5.3/dist/js/bootstrap.bundle.min.js"></script>  <script src="https://cdnjs.cloudflare.com/ajax/libs/jquery-easing/1.4.1/jquery.easing.min.js"></script>  <script src="js/scripts.js"></script> </body> <footer> <p><small>&nbsp&nbsp © Copyright 2020, Sarah Pratt. Template by <a href="https://startbootstrap.com/"> Start Bootstrap </small></a> </p> </footer> </html>

CINXE.COM

Sarah Pratt Personal Webpage