Flash-MSA: Accelerating Million-Token Training with Sparse Attention Kernels
https://nanduruganesh.github.io/flash-msa/
https://nanduruganesh.github.io/flash-msa/
nanduruganesh.github.io
Flash-MSA: Accelerating Million-Token Training With Sparse Attention Kernels
Since Chronium 148, Math.tanh is now fingerprintable to link underlying OS
https://scrapfly.dev/posts/browser-math-os-fingerprint/
https://scrapfly.dev/posts/browser-math-os-fingerprint/
scrapfly.dev
Your Browser Does Math Differently on Every OS, and Anti-Bot Systems Read the Bits · scrapfly.dev
Math.tanh, every CSS trig function, and the Web Audio compressor all route through the host libm, so the rounding of a cosine betrays the OS a browser actually …
MacKenzie Scott's Giving, in Quality-Adjusted Life Years (QALYs)
https://maxghenis.com/mackenzie-scott-qaly/
https://maxghenis.com/mackenzie-scott-qaly/
Maxghenis
MacKenzie Scott's giving, in QALYs | Max Ghenis
An interactive, GiveWell-style cost-effectiveness model estimating the health impact — in quality-adjusted life-years — of MacKenzie Scott's $26.3B in lifetime philanthropy. Runs entirely in your browser.
Modernizing Property Tax Assessments in Allegheny County
https://www.prohousingpgh.org/blog/new-report-modernizing-property-tax-assessments-in-allegheny-county
https://www.prohousingpgh.org/blog/new-report-modernizing-property-tax-assessments-in-allegheny-county
Pro-Housing Pittsburgh
New Report: Modernizing Property Tax Assessments in Allegheny County — Pro-Housing Pittsburgh
This report examines how modern, open-source assessment methods could be used to produce fair, transparent, and affordable countywide reassessments.
Quadrupling code performance with a "useless" if
https://purplesyringa.moe/blog/quadrupling-code-performance-with-a-useless-if/
https://purplesyringa.moe/blog/quadrupling-code-performance-with-a-useless-if/
purplesyringa's blog
Quadrupling code performance with a "useless" if
So I was optimizing a domain-specific compressor the other day, as one does.
One important problem was chunking the input string and optimally choosing the most compact encoding for each chunk (different encodings compress different characters better, so…
One important problem was chunking the input string and optimally choosing the most compact encoding for each chunk (different encodings compress different characters better, so…