Prompt Templates

These are the reference prompt templates.

We first show links to default prompts.

We then show the base prompt template class and its subclasses.

Default Prompts

Prompt Classes

pydantic model llama_index.prompts.base.BasePromptTemplate

Show JSON schema

{
   "title": "BasePromptTemplate",
   "type": "object",
   "properties": {
      "metadata": {
         "title": "Metadata",
         "type": "object"
      },
      "template_vars": {
         "title": "Template Vars",
         "type": "array",
         "items": {
            "type": "string"
         }
      },
      "kwargs": {
         "title": "Kwargs",
         "type": "object",
         "additionalProperties": {
            "type": "string"
         }
      },
      "output_parser": {
         "title": "Output Parser"
      }
   },
   "required": [
      "metadata",
      "template_vars",
      "kwargs"
   ]
}

Config

arbitrary_types_allowed: bool = True

Fields

kwargs (Dict[str, str])
metadata (Dict[str, Any])
output_parser (Optional[llama_index.types.BaseOutputParser])
template_vars (List[str])

field kwargs: Dict[str, str] [Required]

field metadata: Dict[str, Any] [Required]

field output_parser: Optional[BaseOutputParser] = None

field template_vars: List[str] [Required]

abstract format(llm: Optional[LLM] = None, **kwargs: Any) → str

abstract format_messages(llm: Optional[LLM] = None, **kwargs: Any) → List[ChatMessage]

abstract get_template(llm: Optional[LLM] = None) → str

abstract partial_format(**kwargs: Any) → BasePromptTemplate

pydantic model llama_index.prompts.base.PromptTemplate

Show JSON schema

{
   "title": "PromptTemplate",
   "type": "object",
   "properties": {
      "metadata": {
         "title": "Metadata",
         "type": "object"
      },
      "template_vars": {
         "title": "Template Vars",
         "type": "array",
         "items": {
            "type": "string"
         }
      },
      "kwargs": {
         "title": "Kwargs",
         "type": "object",
         "additionalProperties": {
            "type": "string"
         }
      },
      "output_parser": {
         "title": "Output Parser"
      },
      "template": {
         "title": "Template",
         "type": "string"
      }
   },
   "required": [
      "metadata",
      "template_vars",
      "kwargs",
      "template"
   ]
}

Config

arbitrary_types_allowed: bool = True

Fields

template (str)

field template: str [Required]

format(llm: Optional[LLM] = None, **kwargs: Any) → str: Format the prompt into a string.

format_messages(llm: Optional[LLM] = None, **kwargs: Any) → List[ChatMessage]: Format the prompt into a list of chat messages.

get_template(llm: Optional[LLM] = None) → str

partial_format(**kwargs: Any) → PromptTemplate: Partially format the prompt.

pydantic model llama_index.prompts.base.ChatPromptTemplate

Show JSON schema

{
   "title": "ChatPromptTemplate",
   "type": "object",
   "properties": {
      "metadata": {
         "title": "Metadata",
         "type": "object"
      },
      "template_vars": {
         "title": "Template Vars",
         "type": "array",
         "items": {
            "type": "string"
         }
      },
      "kwargs": {
         "title": "Kwargs",
         "type": "object",
         "additionalProperties": {
            "type": "string"
         }
      },
      "output_parser": {
         "title": "Output Parser"
      },
      "message_templates": {
         "title": "Message Templates",
         "type": "array",
         "items": {
            "$ref": "#/definitions/ChatMessage"
         }
      }
   },
   "required": [
      "metadata",
      "template_vars",
      "kwargs",
      "message_templates"
   ],
   "definitions": {
      "MessageRole": {
         "title": "MessageRole",
         "description": "Message role.",
         "enum": [
            "system",
            "user",
            "assistant",
            "function"
         ],
         "type": "string"
      },
      "ChatMessage": {
         "title": "ChatMessage",
         "description": "Chat message.",
         "type": "object",
         "properties": {
            "role": {
               "default": "user",
               "allOf": [
                  {
                     "$ref": "#/definitions/MessageRole"
                  }
               ]
            },
            "content": {
               "title": "Content",
               "default": "",
               "type": "string"
            },
            "additional_kwargs": {
               "title": "Additional Kwargs",
               "type": "object"
            }
         }
      }
   }
}

Config

arbitrary_types_allowed: bool = True

Fields

message_templates (List[llama_index.llms.base.ChatMessage])

field message_templates: List[ChatMessage] [Required]

format(llm: Optional[LLM] = None, **kwargs: Any) → str

format_messages(llm: Optional[LLM] = None, **kwargs: Any) → List[ChatMessage]

get_template(llm: Optional[LLM] = None) → str

partial_format(**kwargs: Any) → ChatPromptTemplate

pydantic model llama_index.prompts.base.SelectorPromptTemplate

Show JSON schema

{
   "title": "SelectorPromptTemplate",
   "type": "object",
   "properties": {
      "metadata": {
         "title": "Metadata",
         "type": "object"
      },
      "template_vars": {
         "title": "Template Vars",
         "type": "array",
         "items": {
            "type": "string"
         }
      },
      "kwargs": {
         "title": "Kwargs",
         "type": "object",
         "additionalProperties": {
            "type": "string"
         }
      },
      "output_parser": {
         "title": "Output Parser"
      },
      "default_template": {
         "title": "Default Template"
      },
      "conditionals": {
         "title": "Conditionals"
      }
   },
   "required": [
      "metadata",
      "template_vars",
      "kwargs"
   ]
}

Config

arbitrary_types_allowed: bool = True

Fields

conditionals (Optional[List[Tuple[Callable[[llama_index.llms.base.LLM], bool], llama_index.prompts.base.BasePromptTemplate]]])
default_template (llama_index.prompts.base.BasePromptTemplate)

field conditionals: Optional[List[Tuple[Callable[[LLM], bool], BasePromptTemplate]]] = None

field default_template: BasePromptTemplate [Required]

format(llm: Optional[LLM] = None, **kwargs: Any) → str: Format the prompt into a string.

format_messages(llm: Optional[LLM] = None, **kwargs: Any) → List[ChatMessage]: Format the prompt into a list of chat messages.

get_template(llm: Optional[LLM] = None) → str

partial_format(**kwargs: Any) → SelectorPromptTemplate

pydantic model llama_index.prompts.base.LangchainPromptTemplate

Show JSON schema

{
   "title": "LangchainPromptTemplate",
   "type": "object",
   "properties": {
      "metadata": {
         "title": "Metadata",
         "type": "object"
      },
      "template_vars": {
         "title": "Template Vars",
         "type": "array",
         "items": {
            "type": "string"
         }
      },
      "kwargs": {
         "title": "Kwargs",
         "type": "object",
         "additionalProperties": {
            "type": "string"
         }
      },
      "output_parser": {
         "title": "Output Parser"
      },
      "selector": {
         "$ref": "#/definitions/ConditionalPromptSelector"
      }
   },
   "required": [
      "metadata",
      "template_vars",
      "kwargs",
      "selector"
   ],
   "definitions": {
      "BaseOutputParser": {
         "title": "BaseOutputParser",
         "description": "Base class to parse the output of an LLM call.\n\nOutput parsers help structure language model responses.\n\nExample:\n    .. code-block:: python\n\n        class BooleanOutputParser(BaseOutputParser[bool]):\n            true_val: str = \"YES\"\n            false_val: str = \"NO\"\n\n            def parse(self, text: str) -> bool:\n                cleaned_text = text.strip().upper()\n                if cleaned_text not in (self.true_val.upper(), self.false_val.upper()):\n                    raise OutputParserException(\n                        f\"BooleanOutputParser expected output value to either be \"\n                        f\"{self.true_val} or {self.false_val} (case-insensitive). \"\n                        f\"Received {cleaned_text}.\"\n                    )\n                return cleaned_text == self.true_val.upper()\n\n                @property\n                def _type(self) -> str:\n                        return \"boolean_output_parser\"",
         "type": "object",
         "properties": {}
      },
      "BasePromptTemplate": {
         "title": "BasePromptTemplate",
         "description": "Base class for all prompt templates, returning a prompt.",
         "type": "object",
         "properties": {
            "input_variables": {
               "title": "Input Variables",
               "type": "array",
               "items": {
                  "type": "string"
               }
            },
            "output_parser": {
               "$ref": "#/definitions/BaseOutputParser"
            }
         },
         "required": [
            "input_variables"
         ]
      },
      "ConditionalPromptSelector": {
         "title": "ConditionalPromptSelector",
         "description": "Prompt collection that goes through conditionals.",
         "type": "object",
         "properties": {
            "default_prompt": {
               "$ref": "#/definitions/BasePromptTemplate"
            }
         },
         "required": [
            "default_prompt"
         ]
      }
   }
}

Config

arbitrary_types_allowed: bool = True

Fields

selector (langchain.chains.prompt_selector.ConditionalPromptSelector)

field selector: ConditionalPromptSelector [Required]

format(llm: Optional[LLM] = None, **kwargs: Any) → str: Format the prompt into a string.

format_messages(llm: Optional[LLM] = None, **kwargs: Any) → List[ChatMessage]: Format the prompt into a list of chat messages.

get_template(llm: Optional[LLM] = None) → str

partial_format(**kwargs: Any) → BasePromptTemplate: Partially format the prompt.

Subclass Prompts (deprecated)

Deprecated, but still available for reference at this link.