Giles Thomas
gilesthomas.com · Personal blog · on HN since 2010
Giles Thomas writes about programming and machine learning, including a long series on building an LLM from scratch and training hardware.
What HN saysHN commenters point newcomers to his LLM-from-scratch series, with links to part 1, the tag listing, and his post on the math needed.
15HN people shared it
8linked it in comments
114stories submitted
3.1Kstory points
2 months agolast shared
2010people per year2026
Most-shared posts
- 1Llm from scratch 13 people · first shared 2025
- 2Building intuition about LLM parameter counts2 people · 4 points · first shared 2026 · discussion on HN
- 3Poppy the training box, part 1: The beginnings2 people · 4 points · first shared 2026 · discussion on HN
- 4Writing an LLM from scratch, part 32h – Interventions: full fat float322 people · 4 points · first shared 2026 · discussion on HN
- 5Retro Language Models: Rebuilding Karpathy's RNN in PyTorch2 people · 4 points · first shared 2025 · discussion on HN
- 6Revisiting Karpathy's 'Unreasonable Effectiveness of Recurrent Neural Networks'2 people · 3 points · first shared 2025 · discussion on HN
- 7First Looking into Jax2 people · 3 points · first shared 2026 · discussion on HN
- 8How Do LLMs Work?2 people · 3 points · first shared 2025 · 1 comments on HN
- 9Thoughts on Role Confusion2 people · 3 points · first shared 2026 · discussion on HN
- 10Benchmarking Qwen 3.6 35B MoE (3B active) on an RTX 30902 people · 3 points · first shared 2026 · discussion on HN
- 11Jax: Commitment Issues2 people · 3 points · first shared 2026 · discussion on HN
- 12Flax debugging: making a hash of things2 people · 3 points · first shared 2026 · discussion on HN
- 13LLM from scratch, part 32j – trying to train a better model in the cloud2 people · 3 points · first shared 2026 · discussion on HN
- 1410Gb/s Ethernet: using mini-heatsinks with a 10GBASE-T SFP+ module2 people · 3 points · first shared 2026 · discussion on HN
- 15Writing an LLM from scratch, part 32B – Interventions: gradient clipping2 people · 3 points · first shared 2026 · discussion on HN
What HN people wrote when they linked it
gpjt · 2025-12-09
OP here -- agreed! I tried to summarise (at least to my current level of knowledge) those 12-18 hours here: https://www.gilesthomas.com/2025/09/maths-for-llms
nfriedly · 2025-12-09
The full list of articles is at https://www.gilesthomas.com/llm-from-scratch for anyone who's interested but wants to start at the beginning.
mettamage · 2025-10-16
Here's part 1 [1]. Since his archive goes by date, it makes it a bit easier to guestimate which part is made in which month.
[1] https://www.gilesthomas.com/2024/12/llm-from-scratch-1
ziyunli · 2025-10-16
seems like you can filter by tag https://www.gilesthomas.com/llm-from-scratch
asimovDev · 2025-10-03
andrehacker · 2025-05-11
This looks very interesting. The easiest way to navigate to the start of this series of articles seems to be https://www.gilesthomas.com/til-deep-dives/page/2
Now if I only could find some time...
Tokumei-no-hito · 2025-05-11
maybe it renders differently on mobile but this was the first entry for me. you can use the nav at the end to continue to the next part
https://www.gilesthomas.com/2024/12/llm-from-scratch-1
kbr · 2017-09-02
Alright I'm going to stop replying to these, I'm starting feel attacked.
8. All of my code is not on Github, and there is no reason for me to learn a language simply because I have previous knowledge in Ruby. I chose to learn Crystal because of it's elegance and speed.
9. I wrote that reverse proxy based on a variety of online tutorials[1], but used my knowledge of threads to make it multithreaded.
[1] http://www.gilesthomas.com/2013/08/writing-a-reverse-proxyloadbalancer…