Darshil Doshi

4 papers · 2023–2025 · 3 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 🗺️ Taxonomy Completionist (13) ❓ The Questioner

Venues

NeurIPS (2) ICLR (1) ICML (1)

Top co-authors

Tianyu He (4) Andrey Gromov (3) Aritra Das (2) Dayal Singh Kalra (1) Tao Tao (1) Maissam Barkeshli (1)

Research topics

Deep Learning (4) Machine Learning (3) Artificial Intelligence (2) Architectures (2) Core AI (2) Learning Types (2) Optimization & Theory · Machine Learning (2) Techniques (2) Data Science & Analytics (1) Bayesian & Probabilistic (1) Methods (1) Optimization & Theory · Deep Learning (1)

Keywords

transformer architecture (1) in-context learning (1) transformer (1) out-of-distribution generalization (1) gaussian process (1) weight decay (1) training dynamics (1) attention head (1) residual connection (1) skill composition (1) modular arithmetic (1) layer normalization (1) neural network initialization (1) attention pattern (1) embedding layer (1) neural network (1) critical initialization (1) partial jacobian (1) linear congruential generators (1) modulus (1)

Papers

(How) Can Transformers Predict Pseudo-Random Numbers? ICML 2025 Learning to grok: Emergence of in-context learning and skill composition in modular arithmetic tasks NeurIPS 2024 To Grok or not to Grok: Disentangling Generalization and Memorization on Corrupted Algorithmic Datasets ICLR 2024 Critical Initialization of Wide and Deep Neural Networks using Partial Jacobians: General Theory and Applications NeurIPS 2023