This repository provides PyTorch implementations of modern open-source LLM architectures, including Llama, Qwen, DeepSeek, Gemma, GPT-OSS, Kimi, and others, written from scratch for readability and learning. The implementations prioritize clarity and learning over performance, with each model implemented in a single readable file. AI summary
Firehose
Filtered to Hacker News, tagged “pytorch” · clear filters
Browse: People · Companies · Papers · Podcasts · Hacker News · Deep dives