vllm.v1

Modules:

• attention –
• core –
• cudagraph_dispatcher –
• engine –
• executor –
• fault_tolerance –
• hisparse –
• kv_cache_interface –
• kv_cache_layout –

  Physical KV cache layout descriptor.

• kv_cache_spec_registry –

  Registry for KVCacheSpec types and their associated managers.

• kv_hints –
• kv_offload –
• metrics –
• outputs –
• pool –
• request –
• sample –
• serial_utils –
• simple_kv_offload –
• spec_decode –
• structured_output –
• utils –
• watermarking –
• worker –