Picture for Anuj Diwan

Anuj Diwan

When to Use Efficient Self Attention? Profiling Text, Speech and Image Transformer Variants

Add code
Jun 14, 2023
Viaarxiv icon

Unit-based Speech-to-Speech Translation Without Parallel Data

Add code
May 24, 2023
Viaarxiv icon

Continual Learning for On-Device Speech Recognition using Disentangled Conformers

Add code
Dec 13, 2022
Viaarxiv icon

Why is Winoground Hard? Investigating Failures in Visuolinguistic Compositionality

Add code
Nov 11, 2022
Viaarxiv icon

Zero-shot Video Moment Retrieval With Off-the-Shelf Models

Add code
Nov 03, 2022
Viaarxiv icon

Multilingual and code-switching ASR challenges for low resource Indian languages

Add code
Apr 01, 2021
Figure 1 for Multilingual and code-switching ASR challenges for low resource Indian languages
Figure 2 for Multilingual and code-switching ASR challenges for low resource Indian languages
Figure 3 for Multilingual and code-switching ASR challenges for low resource Indian languages
Figure 4 for Multilingual and code-switching ASR challenges for low resource Indian languages
Viaarxiv icon

Reduce and Reconstruct: Improving Low-resource End-to-end ASR Via Reconstruction Using Reduced Vocabularies

Add code
Oct 19, 2020
Figure 1 for Reduce and Reconstruct: Improving Low-resource End-to-end ASR Via Reconstruction Using Reduced Vocabularies
Figure 2 for Reduce and Reconstruct: Improving Low-resource End-to-end ASR Via Reconstruction Using Reduced Vocabularies
Figure 3 for Reduce and Reconstruct: Improving Low-resource End-to-end ASR Via Reconstruction Using Reduced Vocabularies
Figure 4 for Reduce and Reconstruct: Improving Low-resource End-to-end ASR Via Reconstruction Using Reduced Vocabularies
Viaarxiv icon