Mastodawn

There exists a lot of computational models of decision making that can
give account on this or that experimental data. The question is
whether we can adapt these models to solve an actual decision task?
You'll find at https://github.com/rougier/braincraft the rules of a
simple decision making challenge.
Roughly, you have have 1000 neurons (leaky rate models), 100 seconds
of training time and 10 runs for evaluation. The difficulty comes
from the imposed constraints.

Francesc Alted 1d ago

daniel:// stenberg://2d ago

I've been talking to GitHub and giving them feedback on their "create issues with Copilot" thing they have in the works.

Today I tested a version for them and using it I asked copilot to find and report a security problem in curl and make it sound terrifying.

In about ten seconds it had a 100-line description of a "catastrophic vulnerability" it was happy to create an issue for. Entirely made up of course, but sounded plausible.

Proved my point excellently.

Francesc Alted 1d ago

Blosc Development Team 1d ago

Struggling with slow I/O in hashtag#HDF5? Try hashtag#Blosc2 as a filter or I/O data handler — faster data, less pain!

👉 https://www.blosc.org/posts/pytables-b2nd-slicing/

#Performance #DataScience

Francesc Alted Jun 24

Dr. Zilch Jun 24

#ai #meme

Francesc Alted Jun 24

Blosc Development Team Jun 24

🚀 C-Blosc2 2.19.0 is out!
We’ve added b2nd_expand_dims(), making it easy to add new dimensions to your b2nd arrays—perfect for evolving your data structures on the fly.
Big thanks to @lshaw8317 for the contribution! 🙏

Check out the release notes: https://github.com/Blosc/c-blosc2/blob/main/RELEASE_NOTES.md

#C-Blosc2 #DataCompression #OpenSource

c-blosc2/RELEASE_NOTES.md at main · Blosc/c-blosc2

A fast, compressed, persistent binary data store library for C. - Blosc/c-blosc2

GitHub

Francesc Alted Jun 23

📢 We are pleased to announce the integration of a new stack feature in #Blosc2 🚀, which allows for stacking large arrays along a new axis.

Performance benchmarks show that while aligned chunks yield the best results, #Blosc2 with unaligned chunks can still outperform #NumPy—a welcome discovery! 🎉

Many thanks to Luke Shaw for his excellent work on this new functionality. 🙏

We've updated our recent blog post:
Check it out! 🔗 https://www.blosc.org/posts/blosc2-new-concatenate/#stacking-arrays

#Python #DataScience #Performance #OpenSource

Francesc Alted Jun 19

David Beazley Jun 19

I really can't understate how much I'm enjoying my "hard way" journey with Rust this summer. The "hard way" meaning (a) Typing in things myself, (b) Not using an IDE, (c) No LLMs, (d) Making a lot of mistakes (e) Experimenting, and (f) Figuring things out on my own.

Honestly, it's been the most fun coding I've had in a long while.

Francesc Alted Jun 17

Blosc Development Team Jun 17

📢 Blosc2 just launched a super-efficient array concatenation feature! 🎉

Combine massive arrays quickly and with minimal memory. If your array chunks are aligned, it's even faster – no need to decompress first! Perfect for big data tasks.

Check out our blog post: https://www.blosc.org/posts/blosc2-new-concatenate/

Compress Better, Compute Bigger

#DataScience #Blosc2 #DataStorage #Performance #MachineLearning

Francesc Alted Jun 9

NumExpr 2.11.0 is here! 🎉

Key highlights:

🚀 Experimental support for free-threaded Python 3.13!
✨ Imaginary number evaluation like 1.1e1j is now fixed.
✅ Test suite modernized to pytest for easier contributions.
🐍 Python 3.10 is now the minimum supported version.

Check out the release notes for more details!

https://github.com/pydata/numexpr/blob/master/RELEASE_NOTES.rst

#NumExpr #Python #DataScience #Performance