Picture for Abhay Gupta

Abhay Gupta

AAVENUE: Detecting LLM Biases on NLU Tasks in AAVE via a Novel Benchmark

Add code
Aug 27, 2024
Viaarxiv icon

Enabling High-Sparsity Foundational Llama Models with Efficient Pretraining and Deployment

Add code
May 06, 2024
Viaarxiv icon

Sparse Iso-FLOP Transformations for Maximizing Training Efficiency

Add code
Mar 25, 2023
Figure 1 for Sparse Iso-FLOP Transformations for Maximizing Training Efficiency
Figure 2 for Sparse Iso-FLOP Transformations for Maximizing Training Efficiency
Figure 3 for Sparse Iso-FLOP Transformations for Maximizing Training Efficiency
Figure 4 for Sparse Iso-FLOP Transformations for Maximizing Training Efficiency
Viaarxiv icon

SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models

Add code
Mar 18, 2023
Figure 1 for SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models
Figure 2 for SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models
Figure 3 for SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models
Figure 4 for SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models
Viaarxiv icon

RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network

Add code
Jun 28, 2022
Figure 1 for RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network
Figure 2 for RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network
Figure 3 for RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network
Figure 4 for RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network
Viaarxiv icon

DAiSEE: Towards User Engagement Recognition in the Wild

Add code
Apr 13, 2018
Figure 1 for DAiSEE: Towards User Engagement Recognition in the Wild
Figure 2 for DAiSEE: Towards User Engagement Recognition in the Wild
Figure 3 for DAiSEE: Towards User Engagement Recognition in the Wild
Figure 4 for DAiSEE: Towards User Engagement Recognition in the Wild
Viaarxiv icon

Approximating Wisdom of Crowds using K-RBMs

Add code
Nov 17, 2016
Figure 1 for Approximating Wisdom of Crowds using K-RBMs
Viaarxiv icon