NanoEuler – GPT-2 scale model in pure C/CUDA from scratch
Hi everyone, I started working on nanoeuler after the ban of anthropic's fable because my ambition and dream is to work in the AI field in anthropic. The two interesting reasons that led me to create nanoeuler were (1) interfacing with llm does not mean understanding how they are composed and (2), working on llm with a very low-level layer to understand the correlation between parameters and data and growth of the model and how the GPU works and how some layers can be optimized. So I started working on it with a research aspect by making nanoeuler grow more and more but doing one step after…
NanoEuler – GPT-2 scale model in pure C/CUDA from scratch launched on June 28, 2026 with 55 votes, #647 of 22,647 launches that month and more than 95% of that year's launches.
In plain words
NanoEuler is a GPT-2 scale language model built from scratch in pure C and CUDA. Designed for developers and researchers who want to understand how language models work at a low level, it demonstrates the relationship between parameters, data, and GPU optimization through hands-on implementation. The project starts with small-scale text generation and progressively grows in complexity, allowing users to observe how models learn language patterns from the ground up.
written from the facts on this page · September 2026
From the sources
GPT-2-style LLM built from scratch in C/CUDA with hand-written backprop, BPE tokenizer, FlashAttention, pretraining, and SFT. - JustVugg/nanoeuler
A GPT-2-class language model built entirely from scratch in C/CUDA — no PyTorch, no autograd, no ML libraries. The forward and backward passes are written and verified by hand, and the whole training pipeline lives in this repo: a hand-written byte-level BPE tokenizer , pretraining on a books + web corpus, and supervised fine-tuning into a chat model (RLHF/DPO planned). It runs on CPU ( libm + OpenMP) for a small showcase model, and a full from-scratch CUDA engine — cuBLAS matmuls, a hand-written FlashAttention , validated against a CPU reference by a full-model gradient check — trains a ~116M-parameter model on a single RTX 4070. Status & honesty. This is a research/educational artifact,…from github.com
In the maker’s words, at launch
Hi everyone, I started working on nanoeuler after the ban of anthropic's fable because my ambition and dream is to work in the AI field in anthropic. The two interesting reasons that led me to create nanoeuler were (1) interfacing with llm does not mean understanding how they are composed and (2), working on llm with…
Does a similar job
all alternatives →
GPT‑5.4 mini and nanoMar 2026 · openai.com · ▲267Fast and efficient models optimized for coding and subagents
GPT-4.1 in the API2025 · openai.com · ▲375Announcing GPT-4.1, GPT-4.1 mini, & GPT-4.1 nano in the API
Getting GLM 5.2 running on my slow computerJul 2026 · github.com · ▲937A few days ago I found myself trying out GLM 5.2 and was really positively impressed. The capabilities and security I was getting from this LLM are similar to those I've gotten from models like Claude or GPT, and this really surprised me. But then I thought, "I wonder how it would work on a normal computer like mine," and above all, "I wonder if it would work without going into OOM on a computer like mine." So I started working with the help of agents to test this possibility. I started converting the model to int4, understanding MTP usage, and if possible implementing DSA for long context.…
Nano Banana 2Feb 2026 · blog.google · ▲416Hi HN — I’m an indie developer. I built Nano Banana 2 Playground, a simple web demo for generating images from text and transforming existing images (image-to-image). My goal was a fast, minimal UI that lets you iterate on prompts quickly and compare results without extra friction. I’d love feedback on: Prompt UX (what feels missing/confusing) Which controls/presets are actually useful How to communicate limitations and expected results clearly Thanks for taking a look — happy to answer questions and improve it based on your suggestions.
I built a tiny LLM to demystify how language models workApr 2026 · github.com · ▲915Built a ~9M param LLM from scratch to understand how they actually work. Vanilla transformer, 60K synthetic conversations, ~130 lines of PyTorch. Trains in 5 min on a free Colab T4. The fish thinks the meaning of life is food. Fork it and swap the personality for your own character.
More ai this month
the category →
Paintings by AI models: programs run through a simulation of oil paint.
AI · 21d ago · stillwet.art
PixVerse R2▲381A real-time world model you can explore and change
AI · 30d ago · world.pixverse.video
Source: Show HN post ↗ · launch text quoted in part
Launched alongside, June 2026
the whole month →
Fundraisly▲1,537AI fundraising agent that finds investors and books meetings
AI · Jun 2026 · fundraisly.com
Homebrew 6.0.0▲1,481Today, I’m proud to announce Homebrew 6.0.0. The most significant changes since 5.1.0 are a new tap trust security mechanism, the new faster, smaller, default internal Homebrew JSON API, sandboxing on Linux, better defaults informed by our user survey, many brew bundle improvements, improved performance and initial support for macOS 27 (Golden Gate). Happy to discuss any questions here!
Dev tools · Jun 2026 · brew.sh
Hi HN! I was super frustrated at apps like tiktok and wechat preventing me from saving or downloading the animated gifs, so I made a little tool to turn a screen recording into a slack-sized emoji. Hope you get some value from this!
Life & fun · Jun 2026 · vorpus.github.io
Search 18 years of Hacker News and chart how any topic, tool, or person trended. Overlay terms to compare their rise and fall across 45M posts and comments.
Life & fun · Jun 2026 · hackernewstrends.com