Top all-time · #877
Sebastian Raschka
Personal blog · sebastianraschka.comaiprogrammingdata
Sebastian Raschka's personal blog at sebastianraschka.com covers ai, programming and data. Since 2014, 77 different Hacker News readers have linked it 160 times, most often “Understanding Large Language Models – A Transformative Reading List”. It is still writing: the last post was on 2026-10-24, and you can follow it by feed.
Homepage of Sebastian Raschka with recent AI and LLM articles, books, talks, research links, and project resources.
Most-linked posts?
- Understanding Large Language Models – A Transformative Reading List5 people · 6 mentions · first linked 2023-02-07
- Understanding and Coding the Self-Attention Mechanism of Large Language Models5 people · 5 mentions · first linked 2023-02-09
- Key differences between Python 2.7.x and Python 3.x (2014)4 people · 9 mentions · first linked 2014-07-12
- A Short Chronology of Deep Learning for Tabular Data4 people · 4 mentions · first linked 2022-07-24
- Implementing a Principal Component Analysis In Python (2014)4 people · 4 mentions · first linked 2014-04-14
- Running PyTorch on the M1 GPU4 people · 4 mentions · first linked 2022-05-19
- /llms-from-scratch/3 people · 3 mentions · first linked 2026-08-15
- /3 people · 3 mentions · first linked 2017-07-14
- PyTorch in One Hour: From Tensors to Training Neural Networks on Multiple GPUs3 people · 3 mentions · first linked 2025-07-02
- Numeric matrix manipulation: Cheat sheet3 people · 3 mentions · first linked 2014-07-19
What Hacker News says?
n-gram per-layer embeddings[1][2] might be it. [1] https://sebastianraschka.com/llm-architecture-gallery/per-la... [2]: See DS 4.1-Flash and Qwen-3.8-Next.
Comment linking /llm-architecture-gallery/per-layer-embeddings/
on “MiMo v2.6”Submitted Key differences between Python 2.7.x and Python 3.x (2014) as a story · 212 points
HN discussionId love an ELI5 for PLE. Im trying to work it into my back of the napikin math for compute vs memory bandwidth limitations on tok/s in PP vs TG work. My attempt at a simplification of this article on it https://sebastianraschka.com/llm-architecture-gallery/per-la... into a couple of sentences is that they are linear embeddings of the input token space projected per layer, which are then gated by the transformer outp…
Comment linking /llm-architecture-gallery/per-layer-embeddings/
on “DeepSeek v4.1 Flash”Submitted Intro to Deep Learning as a story · 189 points
HN discussionhttp://sebastianraschka.com/Articles/2014_python_2_3_key_dif...
Comment linking /Articles/2014_python_2_3_key_diff.html
on “Larry Wall Unveils Perl 6.0.0”Submitted Python, Machine Learning, and Language Wars (2015) as a story · 178 points
HN discussion1. Build an LLM from Scratch by Sebastian Raschka ( https://sebastianraschka.com/llms-from-scratch/ ) 2. LLM from 0 to Hero, and nanoGPT by Andrej Karpathy
Comment linking /llms-from-scratch/
on “I were 17, I'd learn how to build LLMs from scratch”Submitted Python, Machine Learning, and Language Wars. A Highly Subjective Point of View as a story · 167 points
HN discussionYou have to start somewhere. But looped transformers https://sebastianraschka.com/llm-architecture-gallery/looped... are more similar to your idea.
Comment linking /llm-architecture-gallery/looped-depth-sharing/
on “Hybrid Reasoning Agents: LLM agents that think in latent space, act through text”Submitted Implementing a Principal Component Analysis In Python (2014) as a story · 151 points
HN discussionGlad it was helpful. If you have time for some extracurricular study, I'd highly recommend implementing an LLM. Implementing a toy version is really not that difficult, and can be very illuminating. There are quite a few resources on the web that can help with this. One good one is Andrej Karpathy's "Let's build GPT: from scratch, in code, spelled out.": https://www.youtube.com/watch?v=kCc8FmEb1nY A somewhat more in…
Comment linking /llms-from-scratch/
on “Show HN: A working Nintendo in ~1GB of Qwen weights”Submitted Machine Learning with PyTorch and Scikit-Learn as a story · 149 points
HN discussionRosetta Code is cool, but I've long been looking for a "Rosetta stone" type of tool to show the equivalents of lower level operations in various languages. Recently I found a good resource for matrix operations ( http://sebastianraschka.com/Articles/2014_matrix_cheatsheet.... ). This however covers other topics and languages as well . Another gem to add to my bookmarks :-)
Comment linking /Articles/2014_matrix_cheatsheet.html
on “Hyperpolyglot”Submitted A Short Chronology of Deep Learning for Tabular Data as a story · 148 points
HN discussionIt's quadratic, because of the dot product in the attention mechanism. You can use K-V Caching to get rid of a lot of the quadratic runtime that comes from redundant matrix multiplications, but after you have cached everything, you still need to calculate the dot product k_i * q_j with i,j being index of the tokens. With n tokens, you will get O(n*n). But you have to remember that this is only n^2 multiplications. I…
Comment linking /blog/2023/self-attention-from-scratch.html
on “Things I don't know about AI”Submitted Naive Bayes and Text Classification I – Introduction and Theory as a story · 115 points
HN discussionpytorch can use the GPUs on M1 macs. Sebastian Raschka's post explains it nicely and shows some benchmarks too. https://sebastianraschka.com/blog/2022/pytorch-m1-gpu.html From his post: if you want to run PyTorch code on the GPU, use torch.device("mps") analogous to torch.device("cuda") on an Nvidia GPU.
Comment linking /blog/2022/pytorch-m1-gpu.html
on “Stable Diffusion 2.0 on Mac and Linux via imaginAIry Python library”Submitted Differences between Python 2.7.x and Python 3.x, with examples as a story · 108 points
HN discussionSebastian is the author of two excellent books related to LLMs Build a Large Language Model (From Scratch): https://sebastianraschka.com/llms-from-scratch/ Build a Reasoning Model (From Scratch): https://sebastianraschka.com/reasoning-from-scratch/
Comment linking /reasoning-from-scratch/
on “Language models for text classification: From bag-of-words to Jev”Submitted An introduction to parallel programming using Python's multiprocessing module as a story · 104 points
HN discussionThis post seems to cover PCA in Python quite well: http://sebastianraschka.com/Articles/2014_pca_step_by_step.h...
Comment linking /Articles/2014_pca_step_by_step.html
on “Principal component analysis explained visually”Submitted Running PyTorch on the M1 GPU as a story · 92 points
HN discussionThanks for putting this together. I recommend you check out https://sebastianraschka.com/ the author of Python Machine Learning. He's great at explaining things, wrote the bets intro to ML book IMHO and is a good Twitter follow. I also humbly submit my own guide, from the perspective of how to approach studying it amidst all the resources available: http://karlrosaen.com/ml/
Comment linking /
on “A Curated List of AI and Machine Learning Resources from Around the Web”Submitted Numeric matrix manipulation: Cheat sheet as a story · 53 points
HN discussionAlright. Point by point (although I'm considering both the beginner and "advanced" arguments for each point simultaneously): > You Should Be Able to Run 2 and 3 That makes no sense. The whole point of versions is that they are different. Maybe he means 3 should be backwards compatible. That is potentially valid criticism. That said, AFAICT, the breaking changes are pretty reasonable [0][1]. Also, I'm not sure what w…
Comment linking /Articles/2014_python_2_3_key_diff.html
on “The Case Against Python 3”Submitted Understanding Large Language Models – A Transformative Reading List as a story · 19 points
HN discussionI posted this article some while ago, and I thought it may not be a bad idea to archive it on arXiv. What do you think? The (Latex) PDF version would look like this: http://sebastianraschka.com/PDFs/articles/naive_bayes_1.pdf I just signed up on arXiv and see that the catch is that I'd need 2 recommendations for the categories Computer Science -> Learning or Statistics -> Machine Learning. Would anyone here who woul…
Comment linking /PDFs/articles/naive_bayes_1.pdf
on “Naive Bayes and Text Classification I – Introduction and Theory”Submitted Linear Discriminant Analysis bit by bit as a story · 17 points
HN discussionThe `xrange` example suggests that in Python 3.x, `range `(which is equivalent to xrange in python 2.x is slower than range in python 2.x. This is a double patently IMHO. http://sebastianraschka.com/Articles/2014_python_2_3_key_dif...
Comment linking /Articles/2014_python_2_3_key_diff.html
on “Differences between Python 2.7.x and Python 3.x, with examples”Submitted Optimizing LLMs from a Dataset Perspective as a story · 14 points
HN discussion
