o áÙ·iýã@sždZddlmZddlZddlZddlmZmZmZGdd„dej ƒZ Gdd„dej ƒZ Gd d „d ej ƒZ Gd d „d eƒZ Gd d„deƒZGdd„deƒZdS)z" Engine components for vLLM-Omni. é)ÚAnyN)ÚEngineCoreOutputÚEngineCoreOutputsÚEngineCoreRequestc@s.eZdZUdZeed<eeed<eed<dS)ÚPromptEmbedsPayloadzÏSerialized prompt embeddings payload for direct transfer. data: raw bytes of the tensor in row-major order shape: [seq_len, hidden_size] dtype: torch dtype name (e.g., "float16", "float32") ÚdataÚshapeÚdtypeN) Ú__name__Ú __module__Ú __qualname__Ú__doc__ÚbytesÚ__annotations__ÚlistÚintÚstr©rrúM/home/ubuntu/.local/lib/python3.10/site-packages/vllm_omni/engine/__init__.pyrs   rc@sZeZdZUdZdZedBed<dZee dBed<dZ e dBed<dZ ee dBed<dS)ÚAdditionalInformationEntryzãOne entry of additional_information. Two supported forms are encoded: - tensor: data/shape/dtype - list: a Python list (msgspec-serializable) Exactly one of (tensor_data, list_data) should be non-None. NÚ tensor_dataÚ tensor_shapeÚ tensor_dtypeÚ list_data)r r r r rrrrrrrrrrrrrrrs  rc@s"eZdZUdZeeefed<dS)ÚAdditionalInformationPayloadz‡Serialized dictionary payload for additional_information. Keys are strings; values are encoded as AdditionalInformationEntry. ÚentriesN)r r r r Údictrrrrrrrr/s rc@s2eZdZUdZdZedBed<dZedBed<dS)ÚOmniEngineCoreRequesta(Engine core request for omni models with embeddings support. Extends the base EngineCoreRequest with support for prompt embeddings and additional information payloads, enabling direct transfer of pre-computed embeddings between pipeline stages. Attributes: prompt_embeds: Optional serialized prompt embeddings payload for direct transfer between stages additional_information: Optional serialized additional information dictionary containing tensors or lists to pass along with the request NÚ prompt_embedsÚadditional_information) r r r r rrrrrrrrrr8s rc@s(eZdZUdZeeejfdBed<dS)ÚOmniEngineCoreOutputNÚpooling_output) r r r r!rrÚtorchÚTensorrrrrrr Ls r c@seZdZUgZeeed<dS)ÚOmniEngineCoreOutputsÚoutputsN)r r r r%rr rrrrrr$Ps r$)r ÚtypingrÚmsgspecr"Úvllm.v1.enginerrrÚStructrrrrr r$rrrrÚs