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

Add to Chrome

Add to Firefox

Picture for Dan Alistarh

Dan Alistarh

Optimal Brain Compression: A Framework for Accurate Post-Training Quantization and Pruning

Aug 24, 2022
Elias Frantar, Dan Alistarh

Figure 1 for Optimal Brain Compression: A Framework for Accurate Post-Training Quantization and Pruning

Figure 2 for Optimal Brain Compression: A Framework for Accurate Post-Training Quantization and Pruning

Figure 3 for Optimal Brain Compression: A Framework for Accurate Post-Training Quantization and Pruning

Figure 4 for Optimal Brain Compression: A Framework for Accurate Post-Training Quantization and Pruning

Access Paper or Ask Questions

CrAM: A Compression-Aware Minimizer

Jul 28, 2022
Alexandra Peste, Adrian Vladu, Dan Alistarh, Christoph H. Lampert

Figure 1 for CrAM: A Compression-Aware Minimizer

Figure 2 for CrAM: A Compression-Aware Minimizer

Figure 3 for CrAM: A Compression-Aware Minimizer

Figure 4 for CrAM: A Compression-Aware Minimizer

Access Paper or Ask Questions

QuAFL: Federated Averaging Can Be Both Asynchronous and Communication-Efficient

Jun 22, 2022
Hossein Zakerinia, Shayan Talaei, Giorgi Nadiradze, Dan Alistarh

Figure 1 for QuAFL: Federated Averaging Can Be Both Asynchronous and Communication-Efficient

Figure 2 for QuAFL: Federated Averaging Can Be Both Asynchronous and Communication-Efficient

Figure 3 for QuAFL: Federated Averaging Can Be Both Asynchronous and Communication-Efficient

Figure 4 for QuAFL: Federated Averaging Can Be Both Asynchronous and Communication-Efficient

Access Paper or Ask Questions

Scaling the Wild: Decentralizing Hogwild!-style Shared-memory SGD

Mar 13, 2022
Bapi Chatterjee, Vyacheslav Kungurtsev, Dan Alistarh

Figure 1 for Scaling the Wild: Decentralizing Hogwild!-style Shared-memory SGD

Figure 2 for Scaling the Wild: Decentralizing Hogwild!-style Shared-memory SGD

Figure 3 for Scaling the Wild: Decentralizing Hogwild!-style Shared-memory SGD

Figure 4 for Scaling the Wild: Decentralizing Hogwild!-style Shared-memory SGD

Access Paper or Ask Questions

SPDY: Accurate Pruning with Speedup Guarantees

Jan 31, 2022
Elias Frantar, Dan Alistarh

Figure 1 for SPDY: Accurate Pruning with Speedup Guarantees

Figure 2 for SPDY: Accurate Pruning with Speedup Guarantees

Figure 3 for SPDY: Accurate Pruning with Speedup Guarantees

Figure 4 for SPDY: Accurate Pruning with Speedup Guarantees

Access Paper or Ask Questions

How Well Do Sparse Imagenet Models Transfer?

Dec 23, 2021
Eugenia Iofinova, Alexandra Peste, Mark Kurtz, Dan Alistarh

Figure 1 for How Well Do Sparse Imagenet Models Transfer?

Figure 2 for How Well Do Sparse Imagenet Models Transfer?

Figure 3 for How Well Do Sparse Imagenet Models Transfer?

Figure 4 for How Well Do Sparse Imagenet Models Transfer?

* 19 pages, 8 figures. Latest update is a correction to Fig. 2

Access Paper or Ask Questions

Project CGX: Scalable Deep Learning on Commodity GPUs

Nov 17, 2021
Ilia Markov, Hamidreza Ramezanikebrya, Dan Alistarh

Figure 1 for Project CGX: Scalable Deep Learning on Commodity GPUs

Figure 2 for Project CGX: Scalable Deep Learning on Commodity GPUs

Figure 3 for Project CGX: Scalable Deep Learning on Commodity GPUs

Figure 4 for Project CGX: Scalable Deep Learning on Commodity GPUs

Access Paper or Ask Questions