HackerTrans
TopNewTrendsCommentsPastAskShowJobs

sbbq

no profile record

Submissions

PyTorch in One Hour: From Tensors to Training Neural Networks on Multiple GPUs

sebastianraschka.com
4 points·by sbbq·last year·0 comments

Intermediate ML and AI questions and answers for interview prep

sebastianraschka.com
3 points·by sbbq·last year·0 comments

Qwen3 Implemented from Scratch

github.com
3 points·by sbbq·last year·0 comments

Understanding and Coding the KV Cache in LLMs from Scratch

sebastianraschka.com
6 points·by sbbq·last year·0 comments

Coding LLMs from the Ground Up: A Complete Course

sebastianraschka.com
4 points·by sbbq·last year·0 comments

The State of Reasoning Models

magazine.sebastianraschka.com
4 points·by sbbq·last year·0 comments

Understanding Reasoning LLMs

sebastianraschka.com
4 points·by sbbq·last year·0 comments

Implementing a Byte Pair Encoding (BPE) Tokenizer from Scratch

sebastianraschka.com
4 points·by sbbq·last year·0 comments

AI Research Recap 2024: From New Scaling Laws to Scaling Inference Compute

magazine.sebastianraschka.com
1 points·by sbbq·last year·0 comments

Noteworthy AI Research Papers of 2024 (Part One)

magazine.sebastianraschka.com
1 points·by sbbq·2 years ago·0 comments

Collection of 1k LLM Research Papers of 2024

sebastianraschka.com
4 points·by sbbq·2 years ago·0 comments

Understanding Multimodal LLMs: The Main Techniques and Latest Models

sebastianraschka.com
4 points·by sbbq·2 years ago·0 comments

Implementing the Llama 3.2 1B and 3B Architectures from Scratch

github.com
5 points·by sbbq·2 years ago·0 comments

Converting GPT to Llama step-by-step code guide

github.com
2 points·by sbbq·2 years ago·0 comments

New LLM Pre-Training and Post-Training Paradigms: How Modern LLMs Are Trained

magazine.sebastianraschka.com
5 points·by sbbq·2 years ago·0 comments

LLM instruction finetuning from-scratch tutorial

github.com
2 points·by sbbq·2 years ago·0 comments

Tips for LLM Pretraining and Evaluating Reward Models

magazine.sebastianraschka.com
2 points·by sbbq·2 years ago·0 comments

Sharing Deep Learning Research Models: Building a Super Resolution App

sebastianraschka.com
3 points·by sbbq·4 years ago·0 comments

Taking Datasets, DataLoaders, and PyTorch’s New DataPipes for a Spin

sebastianraschka.com
2 points·by sbbq·4 years ago·0 comments

Running PyTorch on the M1 GPU

sebastianraschka.com
2 points·by sbbq·4 years ago·0 comments

comments

sbbq
·9 months ago·discuss
The chips are great. Now they just need to improve the quite stagnant laptop hardware to go with it.
sbbq
·last year·discuss
I got my first switch in 2017 and still use it as my main console. I used to be a hardcore gamer but as I got older approaching my 40s I appreciate its simplicity and catalog. I game occasionally, maybe 2 hours a week, and it allows me to use it on the couch, bed, favorite chair, whereever I feel most comfortable and relaxed after an intense day of work. That being said, i was excited about a new Switch and must say that I was a bit disappointed because Nintendo always came out with a big surprise regarding their new console designs. On the other hand, I am also just happy that it still retains the handheld form-factor and focus because that's exactly what I love about the original Switch.
sbbq
·3 years ago·discuss
*At Home
sbbq
·3 years ago·discuss
Everyone is buying up H100's (and A100's if they can't find H100's). Sure, in 18 month people may lose interest in buying H100's but then there will already be the next Nvidia model everyone wants to buy to get ahead of the competition in terms of compute capabilities.
sbbq
·3 years ago·discuss
Bluetooth ususally works pretty reliably for me these days. However, if you connect a larger number of bluetooth devices (headphones, mouse, keyboard, trackpad, etc.) it can become a bit flaky.

Since you don't move keyboards like a mouse or headphones, it helps reduce the number of peripheries connected to your computer, which in turn helps with bluetooth connectivity issues if you have a lot of devices connected.

Oh, and it's one fewer thing to charge.
sbbq
·4 years ago·discuss
I think Rapids AI's cuML tried to go into this direction (essentially scikit-learn on the GPU): https://docs.rapids.ai/api/cuml/stable/api.html#logistic-reg.... For some reason it never took really off though.

Btw., going on a tangent, you might like Hummingbird (https://github.com/microsoft/hummingbird). It allows you trained scikit-learn tree-based models to PyTorch. I watched the SciPy talk last year, and it's a super smart & elegant idea.
sbbq
·4 years ago·discuss
Author here. I agree with what you said. I wrote my first book with Packt back when I was a student and was like: "cool, a book deal!" Of course, I didn't know about the caveats :P. Yeah, there was very low (/no) quality control. In fact, they introduced a lot of typos during the layouting (apparently, they re-typed the equations by hand!). However, despite all of that, the book was quite successful, so for the subsequent editions, they gave my book much more attention. Personally, I also got much more flexibility regarding deadlines, etc.

Long story short, yeah, there are definitely issues with quality control, and it's really up to the author to make sure that the content is correct and sound. For this particular book, I must say that I worked with a great layouter who paid a lot of attention to detail this time. Also, with their new layout, they no longer had to re-type the equations, and the typesetting looks so much better now. I am pretty happy with how it turned out this time :)