Skip to content

Choosing OpenAI Model for Translation

Choosing OpenAI Model for Translation

When using the TranslatorOpenAI class for translation tasks, you have two options to specify the model:

  1. Provide the model name as a string.
  2. Use the ModelForTranslator enum to select from predefined models.

Option 1: Using Model Name as a String

You can directly pass the model name as a string when creating an instance of TranslatorOpenAI.

from llmtranslate import TranslatorOpenAI

# Example with a model name as a string
translator = TranslatorOpenAI(
    api_key="YOUR_OPENAI_API_KEY", 
    model="gpt-4o-mini"
)

translated_text = translator.translate(
    text="Cześć jak się masz? Meu nome é Adam", 
    to_language="en"
)
print(translated_text)

Option 2: Using the ModelForTranslator Enum

Alternatively, you can use the ModelForTranslator enum to select a model. This enum contains predefined models that have been tested and come with specific versions.

from llmtranslate import TranslatorOpenAI
from llmtranslate import ModelForTranslator

# Example using the ModelForTranslator enum
translator = TranslatorOpenAI(
    api_key="YOUR_OPENAI_API_KEY", 
    model=ModelForTranslator.GPT_4o_mini
)

translated_text = translator.translate(
    text="Cześć jak się masz? Meu nome é Adam", 
    to_language="en"
)
print(translated_text)

Available Models in ModelForTranslator Enum

from enum import Enum

class ModelForTranslator(Enum):
    BEST_BIG_MODEL = "gpt-4o"
    BEST_SMALL_MODEL = "gpt-4o-mini"
    GPT_4o = "gpt-4o"
    GPT_4o_mini = "gpt-4o-mini"
  • BEST_BIG_MODEL: "gpt-4o" - Represents the best large model available.
  • BEST_SMALL_MODEL: "gpt-4o-mini" - Represents the best smaller model, optimized for speed.
  • GPT_4o: "gpt-4o" - Specific version of GPT-4o.
  • GPT_4o_mini: "gpt-4o-mini" - A smaller version of GPT-4o.

Summary

  • Flexibility: Choose between using a string or the enum based on your preference.
  • Predefined Models: The ModelForTranslator enum provides a convenient way to select tested models with known performance characteristics.