Christos Argyropoulos MD, PhD

515 Followers
314 Following
2.2K Posts

Division Chief, Nephrology, University of New Mexico
Posts:
Data Science, Bioinformatics, Computing, Statistics

#RNAsequencing #Nanopore
#GLMMs #Rstats #Bayesian
#Perl #HPC #Statistics #BigData
#fedi22 #Rstats #Clang

Redditu/ReplacementSlight413
BlueSkyhttps://bsky.app/profile/christosargyrop.bsky.social
Google Scholarhttps://scholar.google.com/citations?user=Hj15fvUAAAAJ&hl=en
Github Profilehttps://github.com/chrisarg
Imaginary gfs but the gfs are GPUs at ebay one cannot afford.
why did they call them "ai datacenters" when they could have called them "slopping malls"
The joke is really on the non #Bayesian statisticians: some of the best tools for their trade (magic formula, h-likelihood, REML, bootstrap, g-formulas) are really approximate Bayesian calculations or posterior predictive distributions. How can they even live with this massive cognitive dissonance?
One could say that our #AIslop #techdebt is decreasing, but what if the real metric should take absolute values of the number of slop #loc changes? #vibecoding
Token maxxing tonight #AIslop #TakeMyMoney
Cloned a repo that used SLURM & instead of cosplaying my 2 dual socket workstations as a cluster, I released the Copilot Kraken to rewrite it to use numactl & #rstats parallel in a single machine. This is what chatbots were made for. Keeping slurm for the actual home cluster tho

Business idea: FAFO pyramid scheme

FA at a geometrically increasing rate, so that at any one time there's much less FO in the pipeline than FA. Until it all blows up, of course.

Who am I kidding everyone's already doing this

I stand behind this #hatespeech #rstats #clang

https://blog.r-project.org/2019/03/28/use-of-c-in-packages/index.html

"A summary of the recommendation would be: don’t use C++ to interface with R. If you need to implement some computation in native code, use C (or perhaps Fortran), not C++"

Use of C++ in Packages - The R Blog

Radeon Pro W5500 provides a very cheap, reasonably low power consumption intro to the AMD HPC/AI ecosystem

The #reproducibility of p-valies is terrible (we derived the distribution of p values for GLMs with Andy Grieve through a Bayesian interpretation (π letter) here: https://arxiv.org/abs/2305.00636). I don't understand why people keep noise mining p's #stats

Formula & graph 👇