Pinned Loading

1. 一个以学习为目标的「Llama 3 架构复现」项目：不依赖 transformers 高层实现，只用 PyTorch 原生算子搭建核心模块（RMSNorm、RoPE、GQA、SwiGLU），加载官方权重并与 HuggingFace 实现进行数值对齐。A “Llama 3 architecture reproduction” project for learning purposes: it …

   Python 3

2. 学习 LLM 时积累的零散的，未独立成项目的代码文件。Code files accumulated during the study of Large Language Models (LLMs) that are fragmented and not developed into independent projects.

   Jupyter Notebook

3. 一个基于大语言模型的个性化学习计划生成工具，能够根据用户的技术背景和学习目标，生成符合 SMART 原则的双周粒度学习计划，并进一步细化为日粒度学习安排。A personalized learning plan generation tool based on large language models (LLMs): it can generate biweekly learning pl…

   Python 2

4. 本项目面向 ICLR 2025 与 ICLR 2026 论文集，用 LLM 提取论文中出现频次，提取年度热点主题与优化方向。统计与可视化由本项目代码自动完成。This project targets the proceedings of ICLR 2025 and ICLR 2026, using Large Language Models (LLMs) to extract the occ…

   Python 2

5. A reverse UDP proxy for Teeworlds game play

   Go 9 2

6. 🚀 Engram-PEFT: An unofficial implementation of DeepSeek Engram. Inject high-capacity conditional memory into LLMs via sparse retrieval PEFT without increasing inference FLOPs / DeepSeek Engram 架构的非…

   Python 17 5