Meta tags:

Headings (most frequently used words):

vivekramanujan, research, service,

Text of the page (most frequently used words):
vivek (18), #ramanujan (18), ali (13), farhadi (13), neurips (10), for (9), the (9), learning (8), code (8), aditya (7), with (7), kusupati (6), matthew (6), wallingford (6), better (5), image (5), aniruddha (5), kembhavi (5), training (5), from (5), 2024 (4), mitchell (4), wortsman (4), sham (4), kakade (4), 2023 (4), alex (4), ludwig (4), schmidt (4), generation (4), data (4), and (4), cvpr (3), 2020 (3), prateek (3), jain (3), neural (3), networks (3), codes (3), imagenet (3), krishna (3), 2022 (3), representations (3), view (3), reconstruction (3), fang (3), roozbeh (3), mottaghi (3), dataset (3), wei (3), real (3), images (3), 2025 (2), iclr (2), improving (2), shape (2), deformation (2), between (2), soft (2), threshold (2), enabling (2), but (2), trained (2), models (2), raghav (2), somani (2), that (2), network (2), mohammad (2), rastegari (2), hidden (2), tasks (2), low (2), dimensional (2), binary (2), while (2), accuracy (2), multi (2), inductive (2), bias (2), model (2), adapt (2), downstream (2), object (2), novel (2), using (2), multimodal (2), datacomp (2), achieves (2), datasets (2), ranjay (2), pang (2), koh (2), sewoong (2), thao (2), nguyen (2), robustness (2), spotlight (2), synthetic (2), 360 (2), world (2), when (2), worse (2), seattle, brown, noise, reaction, game, fun, cs146, computer, vision, cs142, cs141, applied, cs2951k, deep, teaching, reviewer, service, cyclegan, allowing, more, disparate, domains, eccv, 2018, aaron, gokaslan, kwang, kim, daniel, ritchie, james, tompkin, unsupervised, translation, pruning, strategy, based, reparametrization, very, sparse, highly, performant, icml, weight, reparameterization, learnable, sparsity, finding, untrained, subnetworks, initialization, match, performance, what, randomly, weighted, blog, continual, thousands, without, catastrophic, forgetting, rosanne, liu, jason, yosinski, supermasks, superposition, extremely, bits, maintaining, near, optimal, 2021, jae, sung, park, pillutla, llc, accurate, purpose, learnt, parameters, grow, magnitude, approximate, discretized, saturated, activations, new, characterization, emnlp, oral, will, merrill, yoav, goldberg, roy, schwartz, noah, smith, effects, parameter, norm, growth, during, transformer, gradient, descent, preparing, future, versions, saving, cheap, auxiliary, information, about, present, pavan, kumar, anasosalu, vasu, oncel, tuzel, hadi, pouransari, forward, compatible, representation, flexible, multiple, varying, computational, resources, gantavya, bhatt, aniket, rege, sinha, william, howard, snyder, kaifeng, chen, matryoshka, adaptive, deployment, centric, through, dictionary, radiance, field, codebooks, benchmark, creation, text, pairs, zero, shot, track, samir, yitzhak, gadre, gabriel, ilharco, vaishaal, shankar, yonatan, bitton, kalyani, marathe, stephen, mussmann, richard, vencu, mehdi, cherti, olga, saukh, alexander, ratner, shuran, song, hannaneh, hajishirzi, romain, beaumont, dimakis, jenia, jitsev, yair, carmon, jonathan, hayase, georgios, smyrnis, ryan, marten, dhruba, ghosh, jieyu, zhang, eyal, orgad, rahim, entezari, giannis, daras, sarah, pratt, search, next, large, pretrained, distribution, shifts, minimal, labeled, priming, sample, efficient, adaptation, quantity, primary, factor, influencing, other, factors, have, limited, impact, connection, pre, diversity, fine, tuning, can, beneficial, consistently, matched, outperformed, simple, retrieval, baseline, scott, geng, cheng, hsieh, chun, liang, unmet, promise, retrieved, performs, odin, synthesis, largest, date, anand, bhattad, matt, deitke, chiu, scene, imagine, million, videos, studies, how, leads, crt, state, art, fid, improved, compute, efficiency, even, kushal, tirumala, armen, aghajanyan, luke, zettlemoyer, navigating, compression, tradeoff, visual, tokenization, denotes, equal, contribution, research, scholar, github, email, phd, student, working, robust, machine, previously, predoctoral, researcher, ai2, university, washington,

Text of the page (random words):
vivek ramanujan vivek ramanujan phd student at the university of washington working with ali farhadi and ludwig schmidt on robust machine learning previously predoctoral researcher at ai2 email cv github scholar research denotes equal contribution when worse is better navigating the compression generation tradeoff in visual tokenization vivek ramanujan kushal tirumala armen aghajanyan luke zettlemoyer ali farhadi neurips 2025 spotlight studies when and how better image reconstruction leads to better generation crt achieves state of the art imagenet generation 2 18 fid with 2 3x improved compute efficiency even with worse reconstruction from an image to a scene learning to imagine the world from a million 360 videos matthew wallingford anand bhattad aditya kusupati vivek ramanujan matt deitke sham kakade aniruddha kembhavi roozbeh mottaghi wei chiu ma ali farhadi neurips 2024 360 1m dataset and odin model for novel view synthesis from the largest real world multi view dataset to date the unmet promise of synthetic training images using retrieved real images performs better scott geng cheng yu hsieh vivek ramanujan matthew wallingford chun liang li pang wei koh ranjay krishna neurips 2024 synthetic data can be beneficial but is consistently matched or outperformed by real images from a simple retrieval baseline code on the connection between pre training data diversity and fine tuning robustness vivek ramanujan thao nguyen sewoong oh ludwig schmidt ali farhadi neurips 2023 spotlight data quantity is the primary factor influencing downstream robustness while other factors have limited impact neural priming for sample efficient adaptation matthew wallingford vivek ramanujan alex fang aditya kusupati roozbeh mottaghi aniruddha kembhavi ludwig schmidt ali farhadi neurips 2023 enabling large pretrained models to adapt to distribution shifts with minimal labeled data code datacomp in search of the next generation of multimodal datasets samir yitzhak gadre gabriel ilharco alex fang jonathan hayase georgios smyrnis thao nguyen ryan marten mitchell wortsman dhruba ghosh jieyu zhang eyal orgad rahim entezari giannis daras sarah pratt vivek ramanujan yonatan bitton kalyani marathe stephen mussmann richard vencu mehdi cherti ranjay krishna pang wei koh olga saukh alexander ratner shuran song hannaneh hajishirzi ali farhadi romain beaumont sewoong oh alex dimakis jenia jitsev yair carmon vaishaal shankar ludwig schmidt neurips 2023 datasets track a benchmark for multimodal dataset creation with 12 8b image text pairs datacomp 1b achieves 79 2 zero shot imagenet accuracy code neural radiance field codebooks matthew wallingford aditya kusupati alex fang vivek ramanujan aniruddha kembhavi roozbeh mottaghi ali farhadi iclr 2023 learning object centric representations through novel view reconstruction using a dictionary of object codes matryoshka representations for adaptive deployment aditya kusupati gantavya bhatt aniket rege matthew wallingford aditya sinha vivek ramanujan william howard snyder kaifeng chen sham kakade prateek jain ali farhadi neurips 2022 flexible representations that adapt to multiple downstream tasks with varying computational resources code forward compatible training for representation learning vivek ramanujan pavan kumar anasosalu vasu ali farhadi oncel tuzel hadi pouransari cvpr 2022 preparing for future model versions by saving cheap auxiliary information about present training code effects of parameter norm growth during transformer training inductive bias from gradient descent will merrill vivek ramanujan yoav goldberg roy schwartz noah smith emnlp 2022 oral as parameters grow in magnitude networks approximate discretized networks with saturated activations a new characterization of inductive bias in gd llc accurate multi purpose learnt low dimensional binary codes aditya kusupati matthew wallingford vivek ramanujan raghav somani jae sung park krishna pillutla prateek jain sham kakade ali farhadi neurips 2021 learning extremely low dimensional binary codes 20 bits for imagenet 1k while maintaining near optimal accuracy code supermasks in superposition mitchell wortsman vivek ramanujan rosanne liu aniruddha kembhavi mohammad rastegari jason yosinski ali farhadi neurips 2020 hidden networks for continual learning learning thousands of tasks without catastrophic forgetting code blog what s hidden in a randomly weighted neural network vivek ramanujan mitchell wortsman aniruddha kembhavi ali farhadi mohammad rastegari cvpr 2020 finding untrained subnetworks at initialization that match trained network performance code soft threshold weight reparameterization for learnable sparsity aditya kusupati vivek ramanujan raghav somani mitchell wortsman prateek jain sham kakade ali farhadi icml 2020 a pruning strategy based on soft threshold reparametrization enabling very sparse but highly performant trained models improving shape deformation in unsupervised image to image translation aaron gokaslan vivek ramanujan kwang in kim daniel ritchie james tompkin eccv 2018 improving on cyclegan by allowing better shape deformation between more disparate domains service reviewer cvpr 2025 iclr 2024 neurips 2024 teaching cs146 computer vision cs142 ml cs141 applied ai cs2951k deep learning fun reaction game brown noise vivek ramanujan seattle wa