Implementation:Explodinggradients Ragas BasePrompt Class

Field	Value
source	Explodinggradients_Ragas (GitHub)
domains	Prompts, Framework
last_updated	2026-02-10 00:00 GMT

Overview

The base prompt module defines the abstract BasePrompt class and lightweight I/O models (StringIO, BoolIO, StringPrompt) that form the foundation of the Ragas prompt hierarchy.

Description

BasePrompt is an abstract base class requiring subclasses to implement generate (single completion) and generate_multiple (multiple completions) async methods. It provides save and load methods for JSON-based prompt persistence with Ragas version tracking. The class stores a name (auto-derived from the class name via camel-to-snake conversion), language, and an optional original_hash for integrity checking. StringIO and BoolIO are simple Pydantic models wrapping a text string and a boolean respectively. StringPrompt is a concrete subclass of BasePrompt that sends raw text to an LLM via StringPromptValue and returns plain string output, suitable for simple text-in/text-out use cases.

Usage

Use StringPrompt directly for simple string-based LLM generation. Subclass BasePrompt to create custom prompt types with structured input/output handling. Use StringIO and BoolIO as I/O models in PydanticPrompt subclasses.

Code Reference

Item	Detail
Source Location	`src/ragas/prompt/base.py` L23-227
Classes	`BasePrompt(ABC)`, `StringIO(BaseModel)`, `BoolIO(BaseModel)`, `StringPrompt(BasePrompt)`
BasePrompt Signature	`class BasePrompt(ABC)` with abstract methods `generate()` and `generate_multiple()`
Import	`from ragas.prompt.base import BasePrompt, StringIO, StringPrompt`

I/O Contract

Inputs (StringPrompt.generate)

Parameter	Type	Description
`llm`	`BaseRagasLLM`	The language model for text generation
`data`	`str`	Raw text input to send to the LLM
`temperature`	`Optional[float]`	Temperature for generation randomness
`stop`	`Optional[List[str]]`	Stop sequences to end generation
`callbacks`	`Callbacks`	Callback handlers for the generation process

Outputs

Method	Return Type	Description
`StringPrompt.generate()`	`str`	Single generated text string
`StringPrompt.generate_multiple()`	`List[str]`	List of n generated text strings
`BasePrompt.save()`	`None`	Writes JSON file with prompt metadata
`BasePrompt.load()`	`BasePrompt`	Restored prompt instance from JSON

Usage Examples

from ragas.prompt.base import StringPrompt, StringIO, BoolIO

# Using StringPrompt for simple text generation
prompt = StringPrompt()
result = await prompt.generate(
    llm=my_llm,
    data="Summarize the following text: ..."
)
print(result)  # Generated summary text

# Multiple generations
results = await prompt.generate_multiple(
    llm=my_llm,
    data="Give me a creative title for this article: ...",
    n=3,
    temperature=0.9,
)
print(results)  # ['Title 1', 'Title 2', 'Title 3']

# Using StringIO as a Pydantic I/O model
io = StringIO(text="hello world")
print(io.text)  # 'hello world'

Related Pages

PydanticPrompt_Class - Structured prompt subclass of BasePrompt
SimplePrompt_Class - Independent simple prompt class
ImageTextPrompt_Class - Multi-modal prompt extending PydanticPrompt

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment