A Large-scale Chinese Short-Text Conversation Dataset and Chinese pre-training dialog models
$ git clone https://github.com/thu-coai/CDial-GPT.git▲ +2 ★ this week
RWKV (pronounced RwaKuv) is an RNN with great LLM performance, which can also be directly trained like a GPT transformer (parallelizable). We are at RWKV-7 "Goose". So it's combining the best of RNN and transformer - great performance, linear time, constant space (no kv-cache), fast training, infinite ctx_len, and free sentence embedding.
▲ +13 ★ this week
Code for loralib, an implementation of "LoRA: Low-Rank Adaptation of Large Language Models"
▲ +3 ★ this week
[NeurIPS 2024 Best Paper Award][GPT beats diffusion🔥] [scaling laws in visual generation📈] Official impl. of "Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction". An *ultra-simple, user-friendly yet state-of-the-art* codebase for autoregressive image generation!
▲ +2 ★ this week
Chinese version of GPT2 training code, using BERT tokenizer.
An unnecessarily tiny implementation of GPT-2 in NumPy.
▲ +1 ★ this week
GPT2 for Chinese chitchat/用于中文闲聊的GPT2模型
Data from GitHub · snapshot Nov 13, 2026