sradc/SmallPebble
View on GitHub ↗ ★ 135 Python
A minimalist deep learning library written from scratch in Python
8 mentions on Hacker News by 2 people: 3 stories, 5 comments · first 2021-02-03 · latest 2026-01-18
20082026
-
…so I could be missing something.) [1] https://github.com/apple/tensorflow_macos/issues/230 [2] https://github.com/sradc/SmallPebble/blob/2cd915c4ba72bf2d92...
comment by montebicyclelo · 2025-10-26 · in A bug that taught me more about PyTorch than years of using it -
…there's an example of making indexing differentiable in my minimal deep learning library here: https://github.com/sradc/SmallPebble/blob/2cd915c4ba72bf2d92... )
comment by montebicyclelo · 2025-07-28 · in LLM Embeddings Explained: A Visual and Intuitive Guide -
You can implement autograd as a library. Just take a look at this https://github.com/sradc/SmallPebble The first line of the description is: > SmallPebble is a minimal automatic differentiation and deep learning library written from…
comment by credit_guy · 2024-01-05 · in Fastest autograd in the West -
…operations. Wondering how it compares to the approach in this post. (I might try to benchmark.) [1] https://github.com/sradc/SmallPebble
comment by montebicyclelo · 2023-09-22 · in Compiling ML models to C for fun -
…[1] https://www.youtube.com/watch?v=q8SA3rM6ckI [2] https://sidsite.com/posts/autodiff/ [3] https://github.com/sradc/SmallPebble
comment by credit_guy · 2023-08-26 · in Understanding Automatic Differentiation in 30 lines of Python