Harvard & ETH Launch TinyTorch: Build a Working ML Framework in Pure Python, 20 Modules, No GPU
lmoroney · x · 2026-10-12
- TinyTorch, a free open-source curriculum from Harvard's CS 249r (now maintained at ETH Zurich), has you build a working ML framework from scratch in pure Python using PyTorch's own API—20 modules spanning tensors to transformers.
- The rationale follows a long tradition: MINIX for Unix, xv6 for OS, Tiger for compilers. Production systems hide internals that only a teaching version can expose.
- Runs on a laptop with 4 GB RAM and no GPU; ships with grading infrastructure universities can adopt for credit, and self-learners can finish it without a compute budget.
- Site: tinytorch.ai, with an official PyTorch blog post by Vijay Janapa Reddi and Andrea Mattia Garavagno.
Related event: Harvard and ETH Launch TinyTorch, a Pure-Python ML Framework(2 posts)→
More from coding & agent
- "I don't want to look at code, just state machines and live docs": one giant diagram — _AustinCalvert_ · 2026-10-12
- VectorDash: open-source Windows 11 search replacement powered by EmbeddingGemma 2 — andy2na · 2026-10-12
- Using AI agents to reverse engineer and patch games with byte-exact verification — arpit_bhayani · 2026-10-12
- Power user burns 86% of Codex weekly quota in 25 hours, complains resets can't keep up — AlchainHust · 2026-10-12
- dsco: an agentic LLM CLI written in C with tool use, swarms, and MCP — arthurcolle · 2026-10-12
- RLHF sawed off 'Decline': why LLMs hallucinate APIs and the one-line Flores fix — RandalSchwartz · 2026-10-12