HNHacker News
TopNewBestAskShowJobs

TuckerBMorgan

37 karma · joined January 15, 2021

submissionscomments
TuckerBMorgan··on Caffè Corretto
Sounds like an italian version of a 4loco, but like chill and calm
TuckerBMorgan··on We're working on a new RuneScape MMO
It depends on what style of mmo want. Sure a full WoW one wont work out of the box, but a sharded/Guild Wars/Destiny style one would work
TuckerBMorgan··on We're working on a new RuneScape MMO
As others have pointed out, it is not just graphics. You don't need to write your own networking, your own editor, scripting engine. It really saves a lot of time
TuckerBMorgan··on Rejection Sensitivity in Gifted and Twice-Exceptional Children
Only recently learned about RS, and it was one of those moments of seeing a lot of my past through a new lens
TuckerBMorgan··on Tell HN: Bob Cringely has died
Reading accidental empires is how i learned that the tech industry even existed
TuckerBMorgan··on Backprop in Rust ML lib blogpost
Hey HN! I am back with the second in my series of blog posts about can-t my ML library in rust. Still just a hackable project for personal learning. https://github.com/TuckerBMorgan/can-t
TuckerBMorgan··on Can-t stop till you get enough: Rewriting Pytorch in Rust
Hey HN! I am back with an update on my ML Lib, can-t github.com/TuckerBMorgan/can-t

Since I last posted I have poured a lot into the library, we now support all the operations needed to run GTP2. I am currently working on supporting GPT-oss, topk and einsum have been holding my attention as of late.

I wanted to start writing about what I had learned and so this is the first in a series of posts about that subject.

TuckerBMorgan··on Show HN: Cant – Library written in Rust that provides PyTorch-like functionality
I think if I wanted to turn this into a "real" project that picking a direction would make sense. For the moment I am just trying to learn as much as I can about how something like PyTorch works.
TuckerBMorgan··on Show HN: Cant – Library written in Rust that provides PyTorch-like functionality
Basically the idea, and I need to write up something on this, is that as much as possible client usage of the library should feel like you are writing PyTorch. For better of worse, people are used to that lib, and its patterns. So if anyone wanted to bring code over, it is easier if can basically copy and paste it over.