Alert button
Picture for Masoud Asgharian

Masoud Asgharian

Alert button

Mitigating Outlier Activations in Low-Precision Fine-Tuning of Language Models

Dec 15, 2023
Alireza Ghaffari, Justin Yu, Mahsa Ghazvini Nejad, Masoud Asgharian, Boxing Chen, Vahid Partovi Nia

Viaarxiv icon

Statistical Hardware Design With Multi-model Active Learning

Mar 26, 2023
Alireza Ghaffari, Masoud Asgharian, Yvon Savaria

Figure 1 for Statistical Hardware Design With Multi-model Active Learning
Figure 2 for Statistical Hardware Design With Multi-model Active Learning
Figure 3 for Statistical Hardware Design With Multi-model Active Learning
Figure 4 for Statistical Hardware Design With Multi-model Active Learning
Viaarxiv icon

Mathematical Challenges in Deep Learning

Mar 24, 2023
Vahid Partovi Nia, Guojun Zhang, Ivan Kobyzev, Michael R. Metel, Xinlin Li, Ke Sun, Sobhan Hemati, Masoud Asgharian, Linglong Kong, Wulong Liu, Boxing Chen

Figure 1 for Mathematical Challenges in Deep Learning
Figure 2 for Mathematical Challenges in Deep Learning
Figure 3 for Mathematical Challenges in Deep Learning
Figure 4 for Mathematical Challenges in Deep Learning
Viaarxiv icon

On the Convergence of Stochastic Gradient Descent in Low-precision Number Formats

Jan 09, 2023
Matteo Cacciola, Antonio Frangioni, Masoud Asgharian, Alireza Ghaffari, Vahid Partovi Nia

Figure 1 for On the Convergence of Stochastic Gradient Descent in Low-precision Number Formats
Figure 2 for On the Convergence of Stochastic Gradient Descent in Low-precision Number Formats
Figure 3 for On the Convergence of Stochastic Gradient Descent in Low-precision Number Formats
Figure 4 for On the Convergence of Stochastic Gradient Descent in Low-precision Number Formats
Viaarxiv icon

EuclidNets: An Alternative Operation for Efficient Inference of Deep Learning Models

Dec 22, 2022
Xinlin Li, Mariana Parazeres, Adam Oberman, Alireza Ghaffari, Masoud Asgharian, Vahid Partovi Nia

Figure 1 for EuclidNets: An Alternative Operation for Efficient Inference of Deep Learning Models
Figure 2 for EuclidNets: An Alternative Operation for Efficient Inference of Deep Learning Models
Figure 3 for EuclidNets: An Alternative Operation for Efficient Inference of Deep Learning Models
Figure 4 for EuclidNets: An Alternative Operation for Efficient Inference of Deep Learning Models
Viaarxiv icon

Training Integer-Only Deep Recurrent Neural Networks

Dec 22, 2022
Vahid Partovi Nia, Eyyüb Sari, Vanessa Courville, Masoud Asgharian

Figure 1 for Training Integer-Only Deep Recurrent Neural Networks
Figure 2 for Training Integer-Only Deep Recurrent Neural Networks
Figure 3 for Training Integer-Only Deep Recurrent Neural Networks
Figure 4 for Training Integer-Only Deep Recurrent Neural Networks
Viaarxiv icon

Integer Fine-tuning of Transformer-based Models

Sep 20, 2022
Mohammadreza Tayaranian, Alireza Ghaffari, Marzieh S. Tahaei, Mehdi Rezagholizadeh, Masoud Asgharian, Vahid Partovi Nia

Figure 1 for Integer Fine-tuning of Transformer-based Models
Figure 2 for Integer Fine-tuning of Transformer-based Models
Figure 3 for Integer Fine-tuning of Transformer-based Models
Figure 4 for Integer Fine-tuning of Transformer-based Models
Viaarxiv icon

Is Integer Arithmetic Enough for Deep Learning Training?

Jul 18, 2022
Alireza Ghaffari, Marzieh S. Tahaei, Mohammadreza Tayaranian, Masoud Asgharian, Vahid Partovi Nia

Figure 1 for Is Integer Arithmetic Enough for Deep Learning Training?
Figure 2 for Is Integer Arithmetic Enough for Deep Learning Training?
Figure 3 for Is Integer Arithmetic Enough for Deep Learning Training?
Figure 4 for Is Integer Arithmetic Enough for Deep Learning Training?
Viaarxiv icon