Classify

classify returns one of the labels you supply. Set multi_label=True to return a list of labels.

import marvin

sentiment = marvin.classify("This product is excellent", ["positive", "negative"])
topics = marvin.classify(
    "A romantic comedy set aboard a spaceship",
    ["comedy", "romance", "science fiction", "history"],
    multi_label=True,
)

An enum gives the result a named Python type. Use descriptive enum values: those values, rather than member names, are shown to the model.

from enum import Enum
import marvin

class Priority(Enum):
    LOW = "low priority"
    HIGH = "high priority"

priority = marvin.classify("Production is down", Priority)

Use instructions to define decision criteria, such as what distinguishes high from low priority. classify_async is the async counterpart. See the API reference for optional arguments.

Get Started

Concepts

Functions

Patterns

Guides

Community