Get our free extension to see links to code for papers anywhere online!

Chrome logo  Add to Chrome

Firefox logo Add to Firefox
Picture for Michal Lukasik

Michal Lukasik
Alert button

ResMem: Learn what you can and memorize the rest


Feb 03, 2023
Zitong Yang, Michal Lukasik, Vaishnavh Nagarajan, Zonglin Li, Ankit Singh Rawat, Manzil Zaheer, Aditya Krishna Menon, Sanjiv Kumar

Add code
Alert button

Figure 1 for ResMem: Learn what you can and memorize the rest
Figure 2 for ResMem: Learn what you can and memorize the rest
Figure 3 for ResMem: Learn what you can and memorize the rest
Figure 4 for ResMem: Learn what you can and memorize the rest

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Large Language Models with Controllable Working Memory


Nov 09, 2022
Daliang Li, Ankit Singh Rawat, Manzil Zaheer, Xin Wang, Michal Lukasik, Andreas Veit, Felix Yu, Sanjiv Kumar

Add code
Alert button

Figure 1 for Large Language Models with Controllable Working Memory
Figure 2 for Large Language Models with Controllable Working Memory
Figure 3 for Large Language Models with Controllable Working Memory
Figure 4 for Large Language Models with Controllable Working Memory

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Preserving In-Context Learning ability in Large Language Model Fine-tuning


Nov 01, 2022
Yihan Wang, Si Si, Daliang Li, Michal Lukasik, Felix Yu, Cho-Jui Hsieh, Inderjit S Dhillon, Sanjiv Kumar

Add code
Alert button

Figure 1 for Preserving In-Context Learning ability in Large Language Model Fine-tuning
Figure 2 for Preserving In-Context Learning ability in Large Language Model Fine-tuning
Figure 3 for Preserving In-Context Learning ability in Large Language Model Fine-tuning
Figure 4 for Preserving In-Context Learning ability in Large Language Model Fine-tuning

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Robust Distillation for Worst-class Performance


Jun 13, 2022
Serena Wang, Harikrishna Narasimhan, Yichen Zhou, Sara Hooker, Michal Lukasik, Aditya Krishna Menon

Add code
Alert button

Figure 1 for Robust Distillation for Worst-class Performance
Figure 2 for Robust Distillation for Worst-class Performance
Figure 3 for Robust Distillation for Worst-class Performance
Figure 4 for Robust Distillation for Worst-class Performance

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

HD-cos Networks: Efficient Neural Architectures for Secure Multi-Party Computation


Oct 28, 2021
Wittawat Jitkrittum, Michal Lukasik, Ananda Theertha Suresh, Felix Yu, Gang Wang

Add code
Alert button

Figure 1 for HD-cos Networks: Efficient Neural Architectures for Secure Multi-Party Computation
Figure 2 for HD-cos Networks: Efficient Neural Architectures for Secure Multi-Party Computation
Figure 3 for HD-cos Networks: Efficient Neural Architectures for Secure Multi-Party Computation
Figure 4 for HD-cos Networks: Efficient Neural Architectures for Secure Multi-Party Computation

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Leveraging redundancy in attention with Reuse Transformers


Oct 13, 2021
Srinadh Bhojanapalli, Ayan Chakrabarti, Andreas Veit, Michal Lukasik, Himanshu Jain, Frederick Liu, Yin-Wen Chang, Sanjiv Kumar

Add code
Alert button

Figure 1 for Leveraging redundancy in attention with Reuse Transformers
Figure 2 for Leveraging redundancy in attention with Reuse Transformers
Figure 3 for Leveraging redundancy in attention with Reuse Transformers
Figure 4 for Leveraging redundancy in attention with Reuse Transformers

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Teacher's pet: understanding and mitigating biases in distillation


Jul 08, 2021
Michal Lukasik, Srinadh Bhojanapalli, Aditya Krishna Menon, Sanjiv Kumar

Add code
Alert button

Figure 1 for Teacher's pet: understanding and mitigating biases in distillation
Figure 2 for Teacher's pet: understanding and mitigating biases in distillation
Figure 3 for Teacher's pet: understanding and mitigating biases in distillation
Figure 4 for Teacher's pet: understanding and mitigating biases in distillation
* 21 pages, 8 figures 
   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Eigen Analysis of Self-Attention and its Reconstruction from Partial Computation


Jun 16, 2021
Srinadh Bhojanapalli, Ayan Chakrabarti, Himanshu Jain, Sanjiv Kumar, Michal Lukasik, Andreas Veit

Add code
Alert button

Figure 1 for Eigen Analysis of Self-Attention and its Reconstruction from Partial Computation
Figure 2 for Eigen Analysis of Self-Attention and its Reconstruction from Partial Computation
Figure 3 for Eigen Analysis of Self-Attention and its Reconstruction from Partial Computation
Figure 4 for Eigen Analysis of Self-Attention and its Reconstruction from Partial Computation
* 14 pages 
   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Semantic Label Smoothing for Sequence to Sequence Problems


Oct 15, 2020
Michal Lukasik, Himanshu Jain, Aditya Krishna Menon, Seungyeon Kim, Srinadh Bhojanapalli, Felix Yu, Sanjiv Kumar

Add code
Alert button

Figure 1 for Semantic Label Smoothing for Sequence to Sequence Problems
Figure 2 for Semantic Label Smoothing for Sequence to Sequence Problems
Figure 3 for Semantic Label Smoothing for Sequence to Sequence Problems
Figure 4 for Semantic Label Smoothing for Sequence to Sequence Problems

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email
1
2
>>