gpt2-webgl on HackerShow

language model inference

SHOW HN2025

GPT-2 implemented using graphics shaders

Shown by nathan-barry on May 2, 2025. Click the card to turn it over.

228points

25comments

#269of 3,837 in 2025

Top 8%of 2025 by points

How HN took it

MixedNo clear lean

6 positive, 7 neutral or asking, 0 mixed, 2 critical, out of 15 comments about the project. Thread score +0.33.

Execution: 0 for, 2 againstIdea: 2 for, 0 againstPerformance: 1 for, 0 against

Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.

positive neutral or mixed critical about another commenter or something else the maker

Cool, impressive project
From 3 comments
- As the guy who did GPT2 in Excel, very cool and kudos!! Curious why you chose WebGL over WebGPU? Just to show it can be done? (Also see my other comment about fetching weights…
  ianand on HN
- Nice, that takes some serious grinding to pull off in shaders. Makes me kinda wonder - does pushing old tools like WebGL actually force deeper understanding of how these models…
  gitroom on HN
- Would be really cool to have it hosted on a website if it can be a static page so we could try it out
  SamosaGuru on HN
Interesting link to GPU programming history
From 1 comment
- > Back in the old days, people used to do general-purpose GPU programming by using shaders like GLSL. This is what inspired NVIDIA (and other companies) to eventually create CUDA…
  supuun on HN

Host a working demo on GitHub Pages
From 2 comments
- Would be really cool to have it hosted on a website if it can be a static page so we could try it out
  SamosaGuru on HN
- Incredible. Was going to suggest tossing up a github page with it, but looks like one is currently in-progress? (https://nathan-barry.github.io/gpt2-webgl/) but doesn't seem to…
  jasonjmcghee on HN
Load weights from Hugging Face instead of bundling them
From 2 comments
- You may be able to fetch the weights directly from Hugging Face. I'd try that first.
  Philpax on HN
- I haven't looked at your implementation, but if you can read ONNX weights... https://huggingface.co/Xenova/gpt2
  jasonjmcghee on HN
Write-up of how it was made
From 1 comment
- Would absolutely read a 'making of' post on this project!
  helloplanets on HN

Why WebGL instead of WebGPU
From 1 comment
- As the guy who did GPT2 in Excel, very cool and kudos!! Curious why you chose WebGL over WebGPU? Just to show it can be done? (Also see my other comment about fetching weights…
  ianand on HN
How much inference time is GPU-bound versus CPU-bound
From 1 comment
- Super cool project! How much of the inference time is actually GPU-bound vs CPU-bound in the browser? Curious if WebGPU would offer a big perf boost over WebGL2 here.
  badmonster on HN

transformers.js (ONNX runtime)
From 2 comments
- > Curious why you chose WebGL over WebGPU? Just to show it can be done? For a WebGPU implementation, one can use transformers.js directly (or many other libraries actually),…
  littlestymaar on HN
- Transformers.js wraps the ONNX runtime which is rather versatile (WASM, WebGL, WebGPU, and WebNN). It's not the backend that makes it novel.
  85392_school on HN
gpt2-tfjs
From 1 comment
- Checkout https://github.com/jseeio/gpt2-tfjs fetches the weights for GPT2 from huggingface on the fly.
  ianand on HN
GPT-2 in Excel
From 1 comment
- As the guy who did GPT2 in Excel, very cool and kudos!! Curious why you chose WebGL over WebGPU? Just to show it can be done? (Also see my other comment about fetching weights…
  ianand on HN
Qwen2-0.5B shader world in VRChat
From 1 comment
- I'm reminded of this VRChat world, which runs Qwen2-0.5B in a shader: https://vrchat.com/home/launch?worldId=wrld_e919c1f4-8399-43cc-8305-d890dc8e2e98
  Philpax on HN