Writing Together Online offers structured time to help you spring into writing and stay focused this semester. We offer writing sessions every workday, Monday through Friday. Join our daily 90-minute writing sessions and become part of a community of scholars who connect online, set realistic goals, and write together in the spirit of accountability and camaraderie. The program is open to all MIT students, postdocs, faculty, staff, and affiliates who are working on papers, proposals, thesis/dissertation chapters, application materials, and other writing projects. For more information and to register, go to this link or check the WCC website. Please spread the word and join with colleagues and friends.Register for Spring 2025 Writing Challenge 1Choose those sessions that you want to attend during Challenge 1: February 10th through March 21stMondays 9:00–10:30amTuesdays 8–9:30am and 9:30–11amWednesdays 9:00–10:30amThursdays 8–9:30am and 9:30–11amFridays 8–9:30am and 9:30–11amMIT Students and postdocs who attend at least 5 sessions per challenge will be entered into a raffle of three $25 Amazon gift cards. The raffle will take place on Friday, March 21st. The more you participate, the more times you will be entered into the raffle of prizes.For more information and to register, check the WCC website. Please spread the word and join with peers and friends.The funding support for this program comes from the Office of Graduate Education

Sun Ra, Parliament-Funkadelic, George Clinton, Erykah Badu, Octavia E. Butler, Digable Planets, Janelle Monae, Flying Lotus, Grace Jones, Missy Elliott, and moreA new exhibit in Lewis Music Library celebrates the visionary contributions of Afrofuturist artists across various genres and mediums. From the cosmic jazz of Sun Ra and the psychedelic funk of Parliament-Funkadelic and George Clinton, to the neo-soul of Erykah Badu and the sci-fi narratives of Octavia E. Butler, these artists have pushed the boundaries of creativity and imagination.This event is presented as part of Artfinity: A celebration of creativity and community at MIT.

Hidden within MIT’s Distinctive Collections, many architectural elements from the earliest days of the Institute’s architecture program still survive as part of the Rotch Art Collection. Among the artworks that conservators salvaged was a set of striking windows of gypsum and stained-glass, dating to the late 18th- to 19th c. Ottoman Empire. This exhibition illuminates the life of these historic windows, tracing their refracted histories from Egypt to MIT, their ongoing conservation, and the cutting-edge research they still prompt.The Maihaugen Gallery (14N-130) is open Monday through Thursday, 10am - 4pm, excluding Institute holidays.

The first 2 sessions in March will be held on Zoom.Meet other expecting and first time parents of infants under one year to connect, share information, and support each other. Bring your concerns, questions, and experiences to the group. And of course, your babies are welcome! This peer led group is organized by MS&PC members Kathrin and Maria.Contact Kathrin hauserkathrin1994@gmail.com or Maria maria.korompili24@gmail.com for more information.

Zoom Link: https://mit.zoom.us/j/99672193351Speaker: Cheng TangAffiliation: Jazayeri lab, 4th year PhD candidate (system neuroscience)Title: An explainable transformer circuit for compositional generalizationAbstract: Compositional generalization—the systematic combination of known elements into novel ensembles— is a hallmark of human cognition, enabling flexible problem-solving beyond rote memorization. While transformer models exhibit surprising proficiency in such tasks (Lake et al., 2023), the underlying mechanisms remain poorly understood. In this case study, we reverse-engineer how a transformer achieves compositional generalization at the circuit level, focusing on a function-primitive composition task. In this task, the model infers functions from teaching examples (e.g., interpreting “apple kiki → apple apple” to deduce that “kiki” means double) and generalizes them to new primitives (e.g., applying “kiki” to “tree” to produce “tree tree”). Our trained transformer achieves high test accuracy (~98%), demonstrating robust generalization.In the first half of the presentation, I will introduce the basics of transformer and provide an intuitive account on how attention operations perform information-routing between tokens with a slot-like data structure. Then I will present the human-interpretable algorithm implemented by the model, walk through the circuit discovery procedure, and highlight the correspondence between attention heads and the algorithm’s steps. Lastly, I will show causal perturbation experiments that validates the reverse-engineered circuit. This presentation aims to demystify the black-box impression of transformers to audience in neuroscience and invite discussion between model understanding and model control.