Generate class logits for each prompt. This class automatically batches the given prompts, considering the memory constraint. For the best performance, put all of your prompts into a single list and pass it to this method. Args: prompts: The pro
(
self,
prompts: Union[PromptType, Sequence[PromptType]],
/,
*,
use_tqdm: bool = True,
lora_request: Optional[Union[list[LoRARequest], LoRARequest]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
)
| 1003 | return [EmbeddingRequestOutput.from_base(item) for item in items] |
| 1004 | |
| 1005 | def classify( |
| 1006 | self, |
| 1007 | prompts: Union[PromptType, Sequence[PromptType]], |
| 1008 | /, |
| 1009 | *, |
| 1010 | use_tqdm: bool = True, |
| 1011 | lora_request: Optional[Union[list[LoRARequest], LoRARequest]] = None, |
| 1012 | prompt_adapter_request: Optional[PromptAdapterRequest] = None, |
| 1013 | ) -> list[ClassificationRequestOutput]: |
| 1014 | """ |
| 1015 | Generate class logits for each prompt. |
| 1016 | |
| 1017 | This class automatically batches the given prompts, considering |
| 1018 | the memory constraint. For the best performance, put all of your prompts |
| 1019 | into a single list and pass it to this method. |
| 1020 | |
| 1021 | Args: |
| 1022 | prompts: The prompts to the LLM. You may pass a sequence of prompts |
| 1023 | for batch inference. See :class:`~vllm.inputs.PromptType` |
| 1024 | for more details about the format of each prompts. |
| 1025 | use_tqdm: Whether to use tqdm to display the progress bar. |
| 1026 | lora_request: LoRA request to use for generation, if any. |
| 1027 | prompt_adapter_request: Prompt Adapter request to use for |
| 1028 | generation, if any. |
| 1029 | |
| 1030 | Returns: |
| 1031 | A list of ``ClassificationRequestOutput`` objects containing the |
| 1032 | embedding vectors in the same order as the input prompts. |
| 1033 | """ |
| 1034 | if self.llm_engine.model_config.task != "classify": |
| 1035 | raise ValueError( |
| 1036 | "Classification API is only enabled for `--task classify`") |
| 1037 | |
| 1038 | items = self.encode(prompts, |
| 1039 | use_tqdm=use_tqdm, |
| 1040 | lora_request=lora_request, |
| 1041 | prompt_adapter_request=prompt_adapter_request) |
| 1042 | |
| 1043 | return [ClassificationRequestOutput.from_base(item) for item in items] |
| 1044 | |
| 1045 | def _embedding_score( |
| 1046 | self, |