Wei (Will) Feng

@weifengpy

PyTorch Distributed, FSDP, float8

United States

Joined March 2011

274Following

89Followers

Wei (Will) Feng Retweeted

verl project@verl_project · May 9

verl is embracing @PyTorch fsdp2! Better throughput, memory usage, and composability with torch.compile! Please try it out and give us feedbacks: github.com/volcengine/ver…

1.0K

Wei (Will) Feng@weifengpy · Aug 7

We have been working on PyTorch native float8 and FSDP2 for distributed training. Check out TorchTitan and TorchAO/float8 dev-discuss.pytorch.org/t/enabling-flo… with Andrew Gu, @wanchao_ , @drisspg , @vkuzo , @brian_hirsh

weifengpy's tweet image. We have been working on PyTorch native float8 and FSDP2 for distributed training. Check out TorchTitan and TorchAO/float8 dev-discuss.pytorch.org/t/enabling-flo…

with Andrew Gu, @wanchao_ , @drisspg , @vkuzo , @brian_hirsh

1.0K

Wei (Will) Feng@weifengpy · Jul 22, 2012

@mikiobraun I appreciate your jBlas. I found the pre-compiled version still has DoubleMatrix.get(Range cr, Range rs) bug. Would you update?

Wei (Will) Feng Retweeted

Data Science Central@analyticbridge · Apr 15, 2012

How to rank vectors based on their features?: ning.it/IoxQo3

Wei (Will) Feng@weifengpy · Sep 13, 2011

On @Quora: What is the best open source implementation of R-tree? Answer: http://qr.ae/7CABN