HS
Hemanth Saratchandran
cs.LGcs.CVmath.DGmath.APcs.CLcs.AImath.FAmath.SPcs.NEeess.SP
On Valency
published · living versionsW_chmx6z4e·v1 · currentpublished
Language Models Can Benefit from Domain-Specific Inductive Biases
with Damien Teney, Liangze Jiang, Zachary Shinnick, Simon Lucey
1 version
Preprints & journals
48 papers in the corpus · 2016–2026Lasting Effects of Abstract Pretraining Beyond Perplexity2609.38764v1 · Zachary Shinnick, Hemanth Saratchandran, Damien Teney et al.2026 · 0 citationsarXiv
Procedural Core: A Compact Recurrent Initialization for Vision Transformers2609.37631v1 · Zachary Shinnick, Christian Interno, Hemanth Saratchandran et al.2026 · 0 citationsarXiv
Conditioned Initialization for Attention2609.07086v1 · Hemanth Saratchandran, Simon Lucey2026 · 0 citationsarXiv
Making Transformers Logical2608.00418v2 · Mingze Ma, Hemanth Saratchandran, Cameron Gordon et al.2026 · 0 citationsarXiv
Can Transformers Really Do It All? On the Compatibility of Inductive Biases Across Tasks2607.17624v1 · Damien Teney, Liangze Jiang, Hemanth Saratchandran et al.2026 · 0 citationsarXiv
Machine unlearning as a governance imperative for clinical AI.42471427 · Porter, Anthony, Kirkpatrick, Emily, Garg, Arpit et al.2026 · 0 citationsNPJ digital medicine. 2026
Parameter-Efficient Fine-Tuning with Learnable Rank2606.04325v1 · Arpit Garg, Simon Lucey, Hemanth Saratchandran2026 · 0 citationsarXiv
Mask the Target: A Plug-and-Play Regularizer Against LoRA Forgetting2605.29498v1 · Runze Xu, Arpit Garg, Hemanth Saratchandran et al.2026 · 0 citationsarXiv
The Quantization Benefits of Residual-Free Transformers2605.25880v1 · Yiping Ji, Mahalakshmi Sabanayagam, Peyman Moghadam et al.2026 · 0 citationsarXiv
Data denoising and derivative estimation for data-driven modeling of nonlinear dynamical systems.42089787 · Yao, Jiaqi, Mitchell, Lewis, Maclean, John et al.2026 · 1 citationChaos (Woodbury, N.Y.). 2026;36(5)
Preconditioned Attention: Enhancing Efficiency in Transformers2603.27153v1 · Hemanth Saratchandran2026 · 0 citationsarXiv
Spectral Conditioning of Attention Improves Transformer Performance2603.07162v1 · Hemanth Saratchandran, Simon Lucey2026 · 0 citationsarXiv
The Inlet Rank Collapse in Implicit Neural Representations: Diagnosis and Unified Remedy2602.01526v1 · Jianqiao Zheng, Hemanth Saratchandran, Simon Lucey2026 · 0 citationsarXiv
Procedural Pretraining: Warming Up Language Models with Abstract Data2601.21725v2 · Liangze Jiang, Zachary Shinnick, Anton van den Hengel et al.2026 · 0 citationsarXiv
Cutting the Skip: Training Residual-Free Transformers2510.00345v1 · Yiping Ji, James Martens, Jianqiao Zheng et al.2025 · 0 citationsarXiv
Stable Forgetting: Bounded Parameter-Efficient Unlearning in Foundation Models2509.24166v2 · Arpit Garg, Hemanth Saratchandran, Ravi Garg et al.2025 · 0 citationsarXiv
Towards Higher Effective Rank in Parameter-efficient Fine-tuning using Khatri--Rao Product2508.00230v1 · Paul Albert, Frederic Z. Zhang, Hemanth Saratchandran et al.2025 · 1 citationarXiv
SineLoRA: Sine-Activated Delta Compression2505.21895v2 · Cameron Gordon, Yiping Ji, Hemanth Saratchandran et al.2025 · 0 citationsarXiv
Transformers Pretrained on Procedural Data Contain Modular Structures for Algorithmic Reasoning2505.22308v1 · Zachary Shinnick, Liangze Jiang, Hemanth Saratchandran et al.2025 · 0 citationsarXiv
Leaner Transformers: More Heads, Less Depth2505.20802v1 · Hemanth Saratchandran, Damien Teney, Simon Lucey2025 · 0 citationsarXiv
Structured Initialization for Vision Transformers2505.19985v2 · Jianqiao Zheng, Xueqian Li, Hemanth Saratchandran et al.2025 · 0 citationsarXiv
Enhancing Transformers Through Conditioned Embedded Tokens2505.12789v2 · Hemanth Saratchandran, Simon Lucey2025 · 0 citationsarXiv
Always Skip Attention2505.01996v3 · Yiping Ji, Hemanth Saratchandran, Peyman Moghadam et al.2025 · 0 citationsarXiv
An equivariant Guillemin trace formula2502.08367v1 · Peter Hochs, Hemanth Saratchandran2025 · 0 citationsarXiv
RandLoRA: Full-rank parameter-efficient fine-tuning of large models2502.00987v2 · Paul Albert, Frederic Z. Zhang, Hemanth Saratchandran et al.2025 · 1 citationarXiv
Rethinking Attention: Polynomial Alternatives to Softmax in Transformers2410.18613v3 · Hemanth Saratchandran, Jianqiao Zheng, Yiping Ji et al.2024 · 0 citationsarXiv
Weight Conditioning for Smooth Optimization of Neural Networks2409.03424v2 · Hemanth Saratchandran, Thomas X. Wang, Simon Lucey2024 · 0 citationsarXiv
Invertible Neural Warp for NeRF2407.12354v1 · Shin-Fang Chng, Ravi Garg, Hemanth Saratchandran et al.2024 · 1 citationarXiv
D'OH: Decoder-Only Random Hypernetworks for Implicit Neural Representations2403.19163v2 · Cameron Gordon, Lachlan Ewen MacDonald, Hemanth Saratchandran et al.2024 · 2 citationsarXiv
From Activation to Initialization: Scaling Insights for Optimizing Neural Fields2403.19205v2 · Hemanth Saratchandran, Sameera Ramasinghe, Simon Lucey2024 · 0 citationsarXiv
Efficient Learning With Sine-Activated Low-rank Matrices2403.19243v5 · Yiping Ji, Hemanth Saratchandran, Cameron Gordon et al.2024 · 2 citationsarXiv
Preconditioners for the Stochastic Training of Neural Fields2402.08784v2 · Shin-Fang Chng, Hemanth Saratchandran, Simon Lucey2024 · 1 citationarXiv
A Sampling Theory Perspective on Activations for Implicit Neural Representations2402.05427v1 · Hemanth Saratchandran, Sameera Ramasinghe, Violetta Shevchenko et al.2024 · 1 citationarXiv
Analyzing the Neural Tangent Kernel of Periodically Activated Coordinate Networks2402.04783v1 · Hemanth Saratchandran, Shin-Fang Chng, Simon Lucey2024 · 1 citationarXiv
Architectural Strategies for the optimization of Physics-Informed Neural Networks2402.02711v1 · Hemanth Saratchandran, Shin-Fang Chng, Simon Lucey2024 · 0 citationsarXiv
Curvature-Aware Training for Coordinate Networks2305.08552v1 · Hemanth Saratchandran, Shin-Fang Chng, Sameera Ramasinghe et al.2023 · 2 citationsarXiv
On the effectiveness of neural priors in modeling dynamical systems2303.05728v1 · Sameera Ramasinghe, Hemanth Saratchandran, Violetta Shevchenko et al.2023 · 1 citationarXiv
A Ruelle dynamical zeta function for equivariant flows2303.00312v3 · Peter Hochs, Hemanth Saratchandran2023 · 0 citationsarXiv
On skip connections and normalisation layers in deep optimisation2210.05371v4 · Lachlan Ewen MacDonald, Jack Valmadre, Hemanth Saratchandran et al.2022 · 2 citationsarXiv
Equivariant analytic torsion for proper actions2205.04117v3 · Peter Hochs, Hemanth Saratchandran2022 · 0 citationsarXiv
A new higher order Yang--Mills--Higgs flow in Riemannian -manifold2201.02826v2 · Hemanth Saratchandran, Jiaogen Zhang, Pan Zhang2022 · 0 citationsarXiv
Generalized Ornstein--Uhlenbeck Semigroups in weighted -spaces on Riemannian Manifolds2107.03301v1 · Ognjen Milatovic, Hemanth Saratchandran2021 · 0 citationsarXiv
Realisations of elliptic operators on compact manifolds with boundary2104.01919v2 · Lashi Bandara, Magnus Goffeng, Hemanth Saratchandran2021 · 3 citationsAdvances in Mathematics 420 (2023)
Essential self-adjointness of perturbed biharmonic operators via conformally transformed metrics2003.07697v1 · Ognjen Milatovic, Hemanth Saratchandran2020 · 0 citationsarXiv
Essential self-adjointness of perturbed quadharmonic operators on Riemannian manifolds with an application to the separation problem1904.07210v1 · Hemanth Saratchandran2019 · 2 citationsarXiv
Inequalities and separation for covariant Schrödinger operators1805.06527v1 · Ognjen Milatovic, Hemanth Saratchandran2018 · 3 citationsarXiv
Higher order Seiberg-Witten functionals and their associated gradient flows1802.08573v1 · Hemanth Saratchandran2018 · 3 citationsarXiv
Essential self-adjointness of powers of first-order differential operators on non-compact manifolds with low-regularity metrics1606.06190v2 · Lashi Bandara, Hemanth Saratchandran2016 · 7 citationsJ. Funct. Anal., Volume 273, Issue 12, 2017, Pages 3719-3758
Career total: 60 works. 48 are in this corpus.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.