Ankit Singh Rawat

Faster Cascades via Speculative Decoding ICLR 2025 Analysis of Plan-based Retrieval for Grounded Text Generation EMNLP 2024 A Statistical Framework for Data-dependent Retrieval-Augmented Models ICML 2024 DistillSpec: Improving Speculative Decoding via Knowledge Distillation ICLR 2024 Dual-Encoders for Extreme Multi-label Classification ICLR 2024 Fine-grained Analysis of In-context Linear Estimation: Data, Architecture, and Beyond NeurIPS 2024 From Self-Attention to Markov Models: Unveiling the Dynamics of Generative Transformers ICML 2024 Language Model Cascades: Token-Level Uncertainty And Beyond ICLR 2024 Mechanics of Next Token Prediction with Self-Attention AISTATS 2024 Think before you speak: Training Language Models With Pause Tokens ICLR 2024 USTAD: Unified Single-model Training Achieving Diverse Scores for Information Retrieval ICML 2024 A Statistical Perspective on Retrieval-Based Models ICML 2023 Large Language Models with Controllable Working Memory ACL 2023 On the Role of Attention in Prompt-tuning ICML 2023 Serving Graph Compression for Graph Neural Networks ICLR 2023 Supervision Complexity and its Role in Knowledge Distillation ICLR 2023 Teacher Guided Training: An Efficient Framework for Knowledge Transfer ICLR 2023 The Lazy Neuron Phenomenon: On Emergence of Activation Sparsity in Transformers ICLR 2023 In defense of dual-encoders for neural ranking ICML 2022 A statistical perspective on distillation ICML 2021 Disentangling Sampling and Labeling Bias for Learning in Large-output Spaces ICML 2021 Long-tail learning via logit adjustment ICLR 2021 Overparameterisation and worst-case generalisation: friend or foe? ICLR 2021 RankDistil: Knowledge Distillation for Ranking AISTATS 2021 Adversarial robustness via robust low rank representations NeurIPS 2020 Are Transformers universal approximators of sequence-to-sequence functions? ICLR 2020 Can gradient clipping mitigate label noise? ICLR 2020 Federated Learning with Only Positive Labels ICML 2020 Low-Rank Bottleneck in Multi-head Attention Models ICML 2020 O(n) Connections are Expressive Enough: Universal Approximability of Sparse Transformers NeurIPS 2020 Robust large-margin learning in hyperbolic space NeurIPS 2020 Lifting high-dimensional non-linear models with Gaussian regressors AISTATS 2019 Multilabel reductions: what is my loss optimising? NeurIPS 2019 Sampled Softmax with Random Fourier Features NeurIPS 2019 Associative Memory via a Sparse Recovery Model NeurIPS 2015