vllm.models.kimi_k3.nvidia.ops

Modules:

• attn_res –
• cute_dsl –
• dcp_topk_merge –

  One-round DCP decode top-k merge for the DSA indexer.

• dsa_indexer –

  Fused Triton prologues for the Kimi K3.1 DSA layer, replacing the small

• dsa_indexer_mxfp4 –

  MXFP4 writers of the Kimi K3.1 DSA indexer: the key LayerNorm + MXFP4 quant +

• fused_mla_key_concat_kv_cache –

  Fused MLA prefill and decode epilogues for Kimi-K3.

• kda_conv –

  KDA prefill short convolution over the packed Q|K|V projection in one launch.

• latent_moe_tail –
• recoverssm –

  Kimi-K3 RecoverSSM speculative verify and accepted-state recovery.

• third_party –
• vision_fa4_warmup –

  Startup compilation of Kimi-K3 vision FA4 kernels.