vllm.models.glm5next.nvidia

Modules:

• mega_moe –

  GLM-5.3-Flash adapter for the DeepGEMM MegaMoE kernel.

• model_state –

  GLM-5.3-Flash's model runner state: the hybrid (KDA + MLA) state, plus the

• ops –
• sparse_indexer –

  Custom Sparse Attention Indexer layers.

• specialized_decode –