gyrovagueGeist

9 ms·

91.
by gyrovagueGeist 3y ago
...you mean like how the nogil project already has a working Numpy module?
92.
by gyrovagueGeist 3y ago
It would require no changes from pure Python apps (as the nogil implementation preserves the current behavior & thread safety of python objects). But any C extensions that rely on the GIL for thread safety would be an issue..
93.
by gyrovagueGeist 3y ago
From reading the nogil repo and related PEP, C ABI breaking does not seem to be the worst problem. An updated Cython, Swig, etc seems like it would be enough in most cases to get something running. More extreme cases might only have ~dozen
94.
by gyrovagueGeist 3y ago
Can someone give a good argument of why subinterpreters are an interesting or useful solution to concurrency in Python? It seems like all the drawbacks of multiprocessing (high memory use, slow serialized object copying) with little benefit
95.
by gyrovagueGeist 3y ago
...but the RTX 5000 was already a workstation card in the Turing series
96.
by gyrovagueGeist 3y ago
Not OP, but they likely mean a waste of brainpower in the survival or ability to function as an agent sense. Ontologically of course it's not a waste but that's not part of evolution's cost function.
97.
by gyrovagueGeist 3y ago
The book is excellent, but I also find that both adaptations drag a bit and can be a boring watch.
98.
by gyrovagueGeist 3y ago
"It doesn't stop being magic just because you know how it works" - Terry Prachett
99.
by gyrovagueGeist 3y ago
I’ve only used it on MI200 series cards but both direct API in C and through a cupy interface for matrix mul and triangular solves it’s worked well for me. There was a bit of bugginess for running on non-default streams a few months back, b
100.
by gyrovagueGeist 3y ago
All of these terms are vague. You are possibly mistaking free will with being an agent that has a (possibly deterministic) method of updating priors in response to new (unknown to the agent, and possibly deterministic) input.
101.
by gyrovagueGeist 3y ago
> as far as I can tell it is because they implemented a bunch of BLAS functions on the GPU rocBLAS and other vendor agnostic numeric libraries have made a lot of progress in the past 2 years (mostly as a result of the DoE's exascale
102.
by gyrovagueGeist 3y ago
This seems fine? Planks law was originally just a line fit with parameter tuning. It wasn't till afterwards that theoretical backing was developed with statistical mechanics. Knowing what types of models and configurations match experi
103.
by gyrovagueGeist 4y ago
Its a good baseline, but I very much doubt that openAI is paying anywhere near the public cost for their compute allocation.
104.
by gyrovagueGeist 4y ago
Yep, Obsidian's latex support makes it almost ideal for writing math notes.. but because you can't quickly drop in quick sketches from a stylus (something like the Bear notes app.. which misses too many other features) it kills Ob
105.
by gyrovagueGeist 5y ago
Huh, didn’t think I’d see the writer of The Northern Caves on the top of HN. Back to this post: I’ve written some nearest neighbor code and definitely felt some pressure to make the API sklearn compatible. But I don’t think it’s as bad as t