Alert button
Picture for Sachin Mehta

Sachin Mehta

Alert button

Weight subcloning: direct initialization of transformers using larger pretrained ones

Dec 14, 2023
Mohammad Samragh, Mehrdad Farajtabar, Sachin Mehta, Raviteja Vemulapalli, Fartash Faghri, Devang Naik, Oncel Tuzel, Mohammad Rastegari

Viaarxiv icon

Label-efficient Training of Small Task-specific Models by Leveraging Vision Foundation Models

Nov 30, 2023
Raviteja Vemulapalli, Hadi Pouransari, Fartash Faghri, Sachin Mehta, Mehrdad Farajtabar, Mohammad Rastegari, Oncel Tuzel

Viaarxiv icon

TiC-CLIP: Continual Training of CLIP Models

Oct 24, 2023
Saurabh Garg, Mehrdad Farajtabar, Hadi Pouransari, Raviteja Vemulapalli, Sachin Mehta, Oncel Tuzel, Vaishaal Shankar, Fartash Faghri

Viaarxiv icon

SAM-CLIP: Merging Vision Foundation Models towards Semantic and Spatial Understanding

Oct 23, 2023
Haoxiang Wang, Pavan Kumar Anasosalu Vasu, Fartash Faghri, Raviteja Vemulapalli, Mehrdad Farajtabar, Sachin Mehta, Mohammad Rastegari, Oncel Tuzel, Hadi Pouransari

Figure 1 for SAM-CLIP: Merging Vision Foundation Models towards Semantic and Spatial Understanding
Figure 2 for SAM-CLIP: Merging Vision Foundation Models towards Semantic and Spatial Understanding
Figure 3 for SAM-CLIP: Merging Vision Foundation Models towards Semantic and Spatial Understanding
Figure 4 for SAM-CLIP: Merging Vision Foundation Models towards Semantic and Spatial Understanding
Viaarxiv icon

CLIP meets Model Zoo Experts: Pseudo-Supervision for Visual Enhancement

Oct 21, 2023
Mohammadreza Salehi, Mehrdad Farajtabar, Maxwell Horton, Fartash Faghri, Hadi Pouransari, Raviteja Vemulapalli, Oncel Tuzel, Ali Farhadi, Mohammad Rastegari, Sachin Mehta

Figure 1 for CLIP meets Model Zoo Experts: Pseudo-Supervision for Visual Enhancement
Figure 2 for CLIP meets Model Zoo Experts: Pseudo-Supervision for Visual Enhancement
Figure 3 for CLIP meets Model Zoo Experts: Pseudo-Supervision for Visual Enhancement
Figure 4 for CLIP meets Model Zoo Experts: Pseudo-Supervision for Visual Enhancement
Viaarxiv icon

SHARCS: Efficient Transformers through Routing with Dynamic Width Sub-networks

Oct 18, 2023
Mohammadreza Salehi, Sachin Mehta, Aditya Kusupati, Ali Farhadi, Hannaneh Hajishirzi

Viaarxiv icon

ReLU Strikes Back: Exploiting Activation Sparsity in Large Language Models

Oct 06, 2023
Iman Mirzadeh, Keivan Alizadeh, Sachin Mehta, Carlo C Del Mundo, Oncel Tuzel, Golnoosh Samei, Mohammad Rastegari, Mehrdad Farajtabar

Figure 1 for ReLU Strikes Back: Exploiting Activation Sparsity in Large Language Models
Figure 2 for ReLU Strikes Back: Exploiting Activation Sparsity in Large Language Models
Figure 3 for ReLU Strikes Back: Exploiting Activation Sparsity in Large Language Models
Figure 4 for ReLU Strikes Back: Exploiting Activation Sparsity in Large Language Models
Viaarxiv icon

Diffusion Models as Masked Audio-Video Learners

Oct 05, 2023
Elvis Nunez, Yanzi Jin, Mohammad Rastegari, Sachin Mehta, Maxwell Horton

Figure 1 for Diffusion Models as Masked Audio-Video Learners
Figure 2 for Diffusion Models as Masked Audio-Video Learners
Figure 3 for Diffusion Models as Masked Audio-Video Learners
Figure 4 for Diffusion Models as Masked Audio-Video Learners
Viaarxiv icon

On the Efficacy of Multi-scale Data Samplers for Vision Applications

Sep 08, 2023
Elvis Nunez, Thomas Merth, Anish Prabhu, Mehrdad Farajtabar, Mohammad Rastegari, Sachin Mehta, Maxwell Horton

Figure 1 for On the Efficacy of Multi-scale Data Samplers for Vision Applications
Figure 2 for On the Efficacy of Multi-scale Data Samplers for Vision Applications
Figure 3 for On the Efficacy of Multi-scale Data Samplers for Vision Applications
Figure 4 for On the Efficacy of Multi-scale Data Samplers for Vision Applications
Viaarxiv icon

Bytes Are All You Need: Transformers Operating Directly On File Bytes

May 31, 2023
Maxwell Horton, Sachin Mehta, Ali Farhadi, Mohammad Rastegari

Figure 1 for Bytes Are All You Need: Transformers Operating Directly On File Bytes
Figure 2 for Bytes Are All You Need: Transformers Operating Directly On File Bytes
Figure 3 for Bytes Are All You Need: Transformers Operating Directly On File Bytes
Figure 4 for Bytes Are All You Need: Transformers Operating Directly On File Bytes
Viaarxiv icon