News
Last updated: May 2023
2023
- May 5, 2023 SemDeDup earned a best paper award at the ICLR Multimodal Representation Learning Workshop!
- March 21, 2023 Our work on emergence of maps in blind navigation agents earned an Outstanding Paper award at ICLR 2023!
- January 20, 2023 Our work on emergence of maps in blind navigation agents was accepted as an Oral (notable top 5%) at ICLR 2023!
- January 16, 2023 Lo-fi, an approach based on model soups to remove communication during distributed training, published in TMLR!
2022
- November 21, 2022 Beyond neural scaling laws: beating power law scaling via data pruning earned an Outstanding Paper award at NeurIPS 2022!
- September 14, 2022 Our work on beating power law scaling via data pruning was accepted to NeurIPS 2022 as an oral!
- July 13, 2022: Co-organized the second Sparse NN Workshop.
- May 15, 2022: Our papers on model soups and measuring abstraction with COAT were accepted to ICML 2022!
2021
- September 28, 2021: Our paper showing that invariance in neural networks is driven by data rather than inductive biases or augmentation was accepted to NeurIPS 2021!
- May 8, 2021: Our papers on soft inductive biases with ConViT and benchmarking productive concept learning with CURI were accepted to ICML 2021!
- January 12, 2021: Our papers on and the potentially harmful impact of highly selective units and on training BatchNorm and only BatchNorm were accepted to ICLR 2021!
2020
- November 2, 2020: Our paper on representation learning with latent canonicalizations was accepted to WACV 2021!
- November 2, 2020: I had a lot of fun appearing on the Underrated ML podcast discussing my choice for an underrated paper: Are all layers created equal? by Chiyuan Zhang and collaborators.
- October 28, 2020: Our blog post on our recent line of papers evaluating the causal impact of selectivity is now out!
- October 22, 2020: Check out our new position paper/perspective on falsifiable interpretability research.
- September 25, 2020: Our paper on the generalization-stability tradeoff was accepted to NeurIPS 2020!
- June 14, 2020: I appeared on the DataCast podcast. Had a fun conversation summarizing my career trajectory and many of my papers!
- January 30, 2020: Spoke about our recent work on training batch normalization and only batch normalization at the REWORK Deep Learning Summit SF.
2019
- December 19, 2019: Our papers on the early phase of neural network training, lottery tickets in RL and NLP, and DD-PPO were accepted to ICLR 2020!
- October 24, 2019: Spoke about our recent work on lottery tickets at the REWORK Deep Learning Summit Montreal. Video available here.
- September 3, 2019: Our paper investigating the generality of winning ticket initializations across datasets was accepted at NeurIPS 2019!
- June 15, 2019: Co-organized a workshop on investigating deep phenomena at ICML 2019.
- February 19, 2019: Our review (with David GT Barrett and Jakob H Macke) on methods to analyze both biological and artifical neural networks is out in Current Opinion in Neurobiology!
2018
- December 20, 2018: Our paper on using contrasting options to learn complex generalization problems was accepted at ICLR 2019!
- September 17, 2018: I've moved back to the US and joined Facebook AI Research (FAIR) in Menlo Park, CA!
- September 5, 2018: Our paper on representational similarity was accepted at NIPS 2018!
- July 11, 2018: Our DeepMind blog post on our recent ICML 2018 paper on measuring abstract reasoning was released!
- June 21, 2018: Our Google AI blog post on our recent paper on representational similarity was released!
- June 14, 2018: Our paper describing Generative Query Networks has been published in Science!
- May 11, 2018: Our paper on measuring abstract reasoning was accepted at ICML 2018!
- March 21, 2018: Our DeepMind blog post on our ICLR 2018 paper was released!
- January 29, 2018: Our paper on the relationship between generalization, robustness, and interpretability was accepted at ICLR 2018!