Dayal Singh Kalra

11 papers A* 4Journal 7
YearRankTypeTitle / Venue / Authors
2026 J jnl
A Scalable Measure of Loss Landscape Curvature for Analyzing the Training Dynamics of LLMs.
CoRR
Dayal Singh Kalra, Jean-Christophe Gagnon-Audet, Andrey Gromov, Ishita Mediratta, Kelvin Niu, Alexander H. Miller, Michael Shvartsman
2025 A* conf
(How) Can Transformers Predict Pseudo-Random Numbers?
ICML
Tao Tao, Darshil Doshi, Dayal Singh Kalra, Tianyu He, Maissam Barkeshli
2025 J jnl
(How) Can Transformers Predict Pseudo-Random Numbers?
CoRR
Tao Tao, Darshil Doshi, Dayal Singh Kalra, Tianyu He, Maissam Barkeshli
2025 J jnl
Teaching Pretrained Language Models to Think Deeper with Retrofitted Recurrence.
CoRR
Sean McLeish, Ang Li, John Kirchenbauer, Dayal Singh Kalra, Brian R. Bartoldson, Bhavya Kailkhura, Avi Schwarzschild, Jonas Geiping, Tom Goldstein, Micah Goldblum
2025 A* conf
Universal Sharpness Dynamics in Neural Network Training: Fixed Point Analysis, Edge of Stability, and Route to Chaos.
ICLR
Dayal Singh Kalra, Tianyu He, Maissam Barkeshli
2025 J jnl
When Can You Get Away with Low Memory Adam?
CoRR
Dayal Singh Kalra, John Kirchenbauer, Maissam Barkeshli, Tom Goldstein
2024 A* conf
Why Warmup the Learning Rate? Underlying Mechanisms and Improvements.
NeurIPS
Dayal Singh Kalra, Maissam Barkeshli
2024 J jnl
Why Warmup the Learning Rate? Underlying Mechanisms and Improvements.
CoRR
Dayal Singh Kalra, Maissam Barkeshli
2023 A* conf
Phase diagram of early training dynamics in deep neural networks: effect of the learning rate, depth, and width.
NeurIPS
Dayal Singh Kalra, Maissam Barkeshli
2023 J jnl
Phase diagram of training dynamics in deep neural networks: effect of learning rate, depth, and width.
CoRR
Dayal Singh Kalra, Maissam Barkeshli
2023 J jnl
Universal Sharpness Dynamics in Neural Network Training: Fixed Point Analysis, Edge of Stability, and Route to Chaos.
CoRR
Dayal Singh Kalra, Tianyu He, Maissam Barkeshli