@randerzander

Tech always, football in the fall

 Atlanta, GA
github.com/randerzander
Joined May 2007
  • Tweets4.8K
  • Following2K
  • Followers358
329Media
vaguely insightful, mostly confrontational tweet
2
Randy Gelhausen retweeted
Vibhu Jawa@JawaVibhu7 Sep 2023
We have been hard at work together with DGL folks to accelerate graph neural networks at scale . Checkout my new blogpost to learn more about accelerating GNN training with cugraph and dgl . developer.nvidia.com/blog/in…

Introduction to Graph Neural Networks with NVIDIA cuGraph-DGL | NVIDIA Technical Blog

Graph neural networks (GNNs) have emerged as a powerful tool for a variety of machine learning tasks on graph-structured data. These tasks range from node classification and link prediction to graph…

developer.nvidia.com
1
3
309
Randy Gelhausen retweeted
Teknium 🪽
@Teknium27 Jun 2023
Code is so much easier to understand than math notation for me. I hope that all math classes can be taught in code sometime soon lol
Simon Eskildsen
@Sirupsen26 Jun 2023
1. Find big scary equation that's hard to parse 2. Latex OCR it with Mathpix 3. Ask ChatGPT to break it down into heavily commented Python
29
24
4
316
54,584
Randy Gelhausen retweeted
Bojan Tunguz
@tunguz17 Dec 2022
Unpopular opinion: .csv files *are* a database.
191
157
50
3,012
445,457
Randy Gelhausen retweeted
Ergest Xheblati@ergestx29 Nov 2022
Transforming data is the one of the most evergreen skills. Everything you do these days is powered by data. If you know how to transform that data to make it useful you have a superpower. It doesn’t matter if you use SQL, Python, Excel or any other tool. I of course prefer SQL
4
3
37
Aaron Swartz would have been 36 today. Aaron committed suicide after being sentenced to 35 years in prison by US authorities for transferring and sharing scientific articles from JSTOR.
Readers added context they thought people might want to know
Swartz was never convicted or sentenced to prison. After declining a plea bargain in which he'd serve six months in jail, he committed suicide before trial. forbes.com/sites/timothyl…
1,062
38,769
2,736
198,494
Randy Gelhausen retweeted
David Brochart@davidbrochart26 Nov 2022
I like to think of Python as an API for compiled languages.
4
12
107
Randy Gelhausen retweeted
Graham Markall@gmarkall24 May 2022
Thanks to the hard work of Brandon Miller, the @numba_jit CUDA examples section is much-expanded with nice examples of CUDA target features: shared memory, cooperative groups, forall loops, random number generation, and CPU/GPU jit function interop: numba.readthedocs.io/en/late…
11
1
50
Randy Gelhausen retweeted
Bojan Tunguz
@tunguz4 Nov 2021
gm Remember to set index=False when saving the csv files today. You are welcome. – at Greencastle, IN
16
27
6
403
Spring has sprung, and working outdoors is a thing again. So is lawn-mowing and blowing and general gas powered noise-making :|
2
2
Is there a way to "mute" a thread I participated in once, but no longer want notifications for? @SlackHQ
3
"proceed to the route" - most useless Google maps direction ever
1
this new UI is going to be strongly correlated to RSI and carpal tunnel from forced increase in scrolling motions
tfw you forget to include "-y" when you conda install a package from inside jupyter
1
1
Randy Gelhausen retweeted
Rachel Thomas
@math_rachel12 Jul 2019
You don't have to know CUDA to use a GPU. Rapids, CuPy, & Numba provide familiar SciPy APIs around GPUs. CuPy's interface is highly compatible with NumPy; in most cases it can be used as a drop-in replacement @mrocklin #SciPy2019 cupy.chainer.org/
2
59
1
195
I think the #pydata community is in healthy shape, not only do we have a choice in dataframe libraries (@vaex_io @pandas_dev, @dask_dev, Modin, @rapidsai cudf) almost as rich as the viz ecosystem, we also have the people bringing them together, thanks @datapythonista !
This Post is from an account that no longer exists. Learn more
9
1
30
Randy Gelhausen retweeted
Pranay Pathole
@PPathole3 Jul 2019
Telling a programmer that there's already a library to do X is like telling a songwriter there's already a song about love.
9
71
4
472
there is no substitute for a concrete example
2
"I rounded the longitude and latitude numbers to two decimal places...This small change alone reduced the RMSE by 4.6%". Interesting feature engineering story by @realpaulmahler from @rapidsai medium.com/rapids-ai/when-le…

When Less is More: A brief story about XGBoost feature engineering

A glimpse into how a Data Scientist makes decisions about featuring engineering an XGBoost machine learning model

medium.com
10
2
30
Load more