Adityanarayanan Radhakrishnan
3 papers · 2024–2025 · 2 venues · across top CS/AI venues
Authors are matched by name, so researchers who share a name ("Yang Liu", "Wei Wang") appear as a single profile and their paper counts, venues and achievements are combined. How identity is handled.
Achievements
Jump to papers ↓
🌍
Conference Polyglot · Bronze
🌉
Interdisciplinary Bridge · Bronze
Top co-authors
Research topics
Keywords
learning theory
(1)
neural network optimization
(1)
tangent kernels
(1)
quadratic models
(1)
modular arithmetic
(1)
batch size
(1)
recursive feature machine
(1)
catapult phase
(1)
sgd
(1)
agop
(1)
catapults
(1)
training loss spikes
(1)
grokking
(1)
average gradient outer product
(1)
block-circulant features
(1)
fourier multiplication algorithm
(1)
Papers
Emergence in non-neural models: grokking modular arithmetic via average gradient outer product
ICML 2025
Catapults in SGD: spikes in the training loss and their impact on generalization through feature learning
ICML 2024
Quadratic models for understanding catapult dynamics of neural networks
ICLR 2024