Profiles2020-2025
Surya Ganguli
Stanford University
8Papers
3Last author
18Unique co-authors
0First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers8
- Rethinking Fine-Tuning when Scaling Test-Time Compute: Limiting Confidence Improves Mathematical ReasoningAuthor 4 of 5NeurIPS 2025 · 39 citations
- Pretraining task diversity and the emergence of non-Bayesian in-context learning for regressionAuthor 4 of 4NeurIPS 2023 · 174 citations
- Stochastic Collapse: How Gradient Noise Attracts SGD Dynamics Towards Simpler SubnetworksAuthor 4 of 4NeurIPS 2023 · 52 citations
- Information Geometry of the Retinal Representation ManifoldAuthor 5 of 6NeurIPS 2023 · 17 citations
- The Asymmetric Maximum Margin Bias of Quasi-Homogeneous Neural NetworksAuthor 4 of 4ICLR 2023 · 1 citation
- MetaMorph: Learning Universal Controllers with TransformersAuthor 3 of 4ICLR 2022 · 130 citations
- Deep Learning on a Data Diet: Finding Important Examples Early in TrainingAuthor 2 of 4NeurIPS 2021 · 806 citations
- RNNs can generate bounded hierarchical languages with optimal memoryAuthor 3 of 5EMNLP 2020 · 1 citation
