[Feature] Adaptation of Jet-Nemotron-2B New Hybrid Architecture Language Model

[loading66]

Checklist

• If this is not a feature request but a general question, please start a discussion at https://github.com/sgl-project/sglang/discussions. Otherwise, it will be closed.
• Please use English. Otherwise, it will be closed.

Motivation

Adapt the Jet-Nemotron-2B model. It adopts Post Neural Architecture Search, an efficient post-training architecture exploration and adaptation pipeline applicable to any pre-trained Transformer model. Its linear module JetBlock is a novel linear attention module, whose performance significantly outperforms previous designs such as Mamba2.

Related resources

No response