Geeta Chauhan

68 posts

Geeta Chauhan

Geeta Chauhan

@geeta4c

Applied AI (PyTorch), Meta I am thankful for all I have

Katılım Mart 2009
364 Takip Edilen282 Takipçiler
Geeta Chauhan retweetledi
Soumith Chintala
Soumith Chintala@soumithchintala·
I think this breezes past as the fastest bs=1 inference I know. With plain and simple PyTorch code. Speculative Decoding isn't even applied, and will be purely additive...
Horace He@cHHillee

Two additions to gpt-fast this week. The first one is an optimization to tensor-parallelism added by @foofoobuggy which improves our TP perf by 20-50%. This gives us 200 => 330 tok/s for Llama-7B fp16 and 64 => 91 tok/s for Llama-70B int4 with *no* speculative decoding. (1/4)

English
0
6
54
18.9K
Geeta Chauhan retweetledi
Jeff Dean
Jeff Dean@JeffDean·
Happy 25th Birthday Google! 🎉 I have gotten incredible enjoyement from being along for the ride for 24+ of these years. When I joined, we were a handful of people wedged into a small office area in downtown Palo Alto above what is now a T-Mobile store. 1/
Jeff Dean tweet media
English
51
202
2.3K
393.8K
Andrej Karpathy
Andrej Karpathy@karpathy·
debugging in Python: - `print()`s alone: too simple - `import pdb; pdb.set_trace()`: too complex - `import code; code.interact(local=locals())`: just right simply drops you into interpreter, perfect for 95% of debugging
English
128
663
7.1K
1M
Geeta Chauhan retweetledi
Horace He
Horace He@cHHillee·
I'm going to start posting a series of PyTorch 2.0 benchmarks demonstrating what kinds of things we speed up as well as exploring a variety of ML compiler optimizations! For the first one, let's talk about good old operator fusion - the workhorse of all ML compilers. (1/8)
English
14
118
962
0
Jeremy Howard
Jeremy Howard@jeremyphoward·
@ptrblck_de Is there an example showing how to get these benefits? How about the other things mentioned on that slide (e.g. GIL)?
English
3
0
4
0
Geeta Chauhan retweetledi
PyTorch
PyTorch@PyTorch·
Introducing @PyTorchLive, an easy to use library of tools for creating on-device ML demos on Android and iOS. With Live, you can build a working mobile app ML demo in minutes. Learn more at pytorch.org/live and post your demo with #PyTorchLive.
English
9
146
519
0
Geeta Chauhan retweetledi
PyTorch
PyTorch@PyTorch·
PyTorch Developer Day brings together leading academics, researchers and developers from the Machine Learning community to learn more about software releases on PyTorch and ways PyTorch is being used in academia and industry. twitter.com/i/broadcasts/1…
English
20
92
330
0
Geeta Chauhan retweetledi
Mike Schroepfer
Mike Schroepfer@schrep·
This is a really big step forward in reproducible research. You can now link your papers to the code on papers with code! medium.com/paperswithcode…
English
2
49
244
0

Keşfet

@zou3519 @soumithchintala @karpathy @jeremyphoward @ptrblck_de @PyTorchLive @NeurIPSConf @FacebookAI