GPT-2 training in pure Mojo with hand-written CUDA and Metal GPU kernels. llm.c parity in bf16 on NVIDIA, 1.72x faster than PyTorch MPS on Apple Silicon.
-
Updated
Aug 13, 2026 - Mojo
#
llm-c
Here are 4 public repositories matching this topic...
TensorBoard live logger for llm.c training: parses loss, lr, grad norm from log file → real-time visualisation
-
Updated
Apr 1, 2026 - Python
A GPT/Llama-style language model built from scratch in Rust — hand-written forward & backprop, every gradient numerically verified, trained on CPU, exported to GGUF for llama.cpp & Ollama.
rust machine-learning cpu deep-learning neural-network autograd transformer llama gpt language-model backpropagation from-scratch rope llm nanogpt rmsnorm gguf swiglu from-scratch-neural-network llm-c
-
Updated
Aug 6, 2026 - Rust
Andrej Karpathy 技術學習庫 — 七階段循序漸進，從反向傳播到 LLM 實戰、C/CUDA 優化與 AI Agent，也是 Obsidian Vault
machine-learning deep-learning cuda transformers neural-networks traditional-chinese gpt obsidian zh-tw ai-agents karpathy vercel micrograd vitepress obsidian-vault llm llm-c llama2-c teaching-site
-
Updated
Aug 15, 2026 - Cuda
Improve this page
Add a description, image, and links to the llm-c topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the llm-c topic, visit your repo's landing page and select "manage topics."