Index SLM Technical Report

Bilibili has released Index-1.9B, a collection of four small language models featuring 1.9 billion parameters. Trained on a dataset of 2.8 trillion tokens, the models are specifically optimized for Chinese language processing tasks.