Alert button
Picture for Eugene Belilovsky

Eugene Belilovsky

Alert button

Mila

Channel-Selective Normalization for Label-Shift Robust Test-Time Adaptation

Feb 07, 2024
Pedro Vianna, Muawiz Chaudhary, Paria Mehrbod, An Tang, Guy Cloutier, Guy Wolf, Michael Eickenberg, Eugene Belilovsky

Viaarxiv icon

Model Breadcrumbs: Scaling Multi-Task Model Merging with Sparse Masks

Dec 11, 2023
MohammadReza Davari, Eugene Belilovsky

Viaarxiv icon

Can We Learn Communication-Efficient Optimizers?

Dec 02, 2023
Charles-Étienne Joseph, Benjamin Thérien, Abhinav Moudgil, Boris Knyazev, Eugene Belilovsky

Viaarxiv icon

DragD3D: Vertex-based Editing for Realistic Mesh Deformations using 2D Diffusion Priors

Oct 06, 2023
Tianhao Xie, Eugene Belilovsky, Sudhir Mudur, Tiberiu Popa

Figure 1 for DragD3D: Vertex-based Editing for Realistic Mesh Deformations using 2D Diffusion Priors
Figure 2 for DragD3D: Vertex-based Editing for Realistic Mesh Deformations using 2D Diffusion Priors
Figure 3 for DragD3D: Vertex-based Editing for Realistic Mesh Deformations using 2D Diffusion Priors
Figure 4 for DragD3D: Vertex-based Editing for Realistic Mesh Deformations using 2D Diffusion Priors
Viaarxiv icon

Continual Pre-Training of Large Language Models: How to (re)warm your model?

Aug 08, 2023
Kshitij Gupta, Benjamin Thérien, Adam Ibrahim, Mats L. Richter, Quentin Anthony, Eugene Belilovsky, Irina Rish, Timothée Lesort

Figure 1 for Continual Pre-Training of Large Language Models: How to (re)warm your model?
Figure 2 for Continual Pre-Training of Large Language Models: How to (re)warm your model?
Figure 3 for Continual Pre-Training of Large Language Models: How to (re)warm your model?
Figure 4 for Continual Pre-Training of Large Language Models: How to (re)warm your model?
Viaarxiv icon

$\textbf{A}^2\textbf{CiD}^2$: Accelerating Asynchronous Communication in Decentralized Deep Learning

Jun 14, 2023
Adel Nabli, Eugene Belilovsky, Edouard Oyallon

Figure 1 for $\textbf{A}^2\textbf{CiD}^2$: Accelerating Asynchronous Communication in Decentralized Deep Learning
Figure 2 for $\textbf{A}^2\textbf{CiD}^2$: Accelerating Asynchronous Communication in Decentralized Deep Learning
Figure 3 for $\textbf{A}^2\textbf{CiD}^2$: Accelerating Asynchronous Communication in Decentralized Deep Learning
Figure 4 for $\textbf{A}^2\textbf{CiD}^2$: Accelerating Asynchronous Communication in Decentralized Deep Learning
Viaarxiv icon

Adversarial Attacks on the Interpretation of Neuron Activation Maximization

Jun 12, 2023
Geraldin Nanfack, Alexander Fulleringer, Jonathan Marty, Michael Eickenberg, Eugene Belilovsky

Figure 1 for Adversarial Attacks on the Interpretation of Neuron Activation Maximization
Figure 2 for Adversarial Attacks on the Interpretation of Neuron Activation Maximization
Figure 3 for Adversarial Attacks on the Interpretation of Neuron Activation Maximization
Figure 4 for Adversarial Attacks on the Interpretation of Neuron Activation Maximization
Viaarxiv icon

Can Forward Gradient Match Backpropagation?

Jun 12, 2023
Louis Fournier, Stéphane Rivaud, Eugene Belilovsky, Michael Eickenberg, Edouard Oyallon

Figure 1 for Can Forward Gradient Match Backpropagation?
Figure 2 for Can Forward Gradient Match Backpropagation?
Figure 3 for Can Forward Gradient Match Backpropagation?
Figure 4 for Can Forward Gradient Match Backpropagation?
Viaarxiv icon

Guiding The Last Layer in Federated Learning with Pre-Trained Models

Jun 06, 2023
Gwen Legate, Nicolas Bernier, Lucas Caccia, Edouard Oyallon, Eugene Belilovsky

Figure 1 for Guiding The Last Layer in Federated Learning with Pre-Trained Models
Figure 2 for Guiding The Last Layer in Federated Learning with Pre-Trained Models
Figure 3 for Guiding The Last Layer in Federated Learning with Pre-Trained Models
Figure 4 for Guiding The Last Layer in Federated Learning with Pre-Trained Models
Viaarxiv icon

Re-Weighted Softmax Cross-Entropy to Control Forgetting in Federated Learning

Apr 11, 2023
Gwen Legate, Lucas Caccia, Eugene Belilovsky

Figure 1 for Re-Weighted Softmax Cross-Entropy to Control Forgetting in Federated Learning
Figure 2 for Re-Weighted Softmax Cross-Entropy to Control Forgetting in Federated Learning
Figure 3 for Re-Weighted Softmax Cross-Entropy to Control Forgetting in Federated Learning
Figure 4 for Re-Weighted Softmax Cross-Entropy to Control Forgetting in Federated Learning
Viaarxiv icon