Prompt¶

The Prompt class is a simple singleton template that allows you to create prompts with schema validation using Pydantic models. It provides a straightforward way to generate structured outputs from language models.

Basic Usage¶

1. Import Required Dependencies¶

from dria import Prompt, DatasetGenerator, DriaDataset, Model
from pydantic import BaseModel, Field

2. Define Output Schema¶

Create a Pydantic model that defines the structure of your expected output:

class OutputSchema(BaseModel):
    field1: str = Field(..., title="Field1")
    field2: str = Field(..., title="Field2")

3. Create Dataset¶

Initialize a DriaDataset with your schema:

dataset = DriaDataset(
    name="my_dataset",
    description="Dataset description",
    schema=OutputSchema
)

4. Initialize Prompt¶

Create a Prompt instance with your prompt text and schema. Add variables in the prompt text using double curly braces {{}}:

prompter = Prompt(prompt="Do something with {{field1}}", schema=OutputSchema)

5. Generate Data¶

Use DatasetGenerator to execute the prompt:

generator = DatasetGenerator(dataset=dataset)
instructions = [{"field1": "value1"}, {"field2": "value2"}]

asyncio.run(
    generator.generate(
        instructions=instructions,
        singletons=prompter,
        models=Model.LLAMA_3_1_8B_OR
    )
)

Complete Example¶

Here's a complete example that generates tweets about different topics:

import asyncio
from dria import Prompt, DatasetGenerator, DriaDataset, Model
from pydantic import BaseModel, Field

# Define output schema
class Tweet(BaseModel):
    topic: str = Field(..., title="Topic")
    tweet: str = Field(..., title="tweet")

# Create dataset
dataset = DriaDataset(
    name="tweet_test",
    description="A dataset of tweets!",
    schema=Tweet
)

# Define instructions
instructions = [
    {"topic": "BadBadNotGood"},
    {"topic": "Decentralized synthetic data"}
]

# Create prompt
prompter = Prompt(prompt="Write a tweet about {{topic}}", schema=Tweet)

# Initialize generator
generator = DatasetGenerator(dataset=dataset)

# Generate data
asyncio.run(
    generator.generate(
        instructions=instructions,
        singletons=prompter,
        models=Model.GPT4O
    )
)

# View results
dataset.to_pandas()

Prompt is useful for straightforward tasks. For more complex tasks, consider creating a custom singleton using the SingletonTemplate class.