← all repositories

Langboat/Mengzi

Mengzi is a series of pre-trained Chinese language models (BERT, GPT-Neo, T5 variants) optimized for lightweight deployment.

Mengzi
Velocity · 7d
+0.3
★ / day
Trend
steady
star history

Mengzi provides a suite of pre-trained language models for Chinese NLP tasks. The project focuses on developing lightweight yet effective models by incorporating linguistic information and training acceleration techniques. Models maintain BERT-compatible architectures for easy deployment. The collection includes BERT variants, GPT architecture models, T5-based multi-task models, and distilled versions.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.