TranslatorOpenSource Class Documentation

The TranslatorOpenSource class is designed for language detection and translation using open-source large language models (LLMs) deployed on a custom endpoint. This class enables interaction with these models to detect the language of text and translate between languages.

Initialization

from llmtranslate import TranslatorOpenSource

# Create the translator object
translator = TranslatorOpenSource(
    api_key="YOUR_API_KEY", 
    llm_endpoint="YOUR_LLM_ENDPOINT", 
    model="mistralai/Mistral-Nemo-Instruct-2407"
)

The constructor initializes the TranslatorOpenSource class by setting up the API key, endpoint, and large language model (LLM). It also configures the maximum text length limits depending on the model.

Parameters:

• api_key:
  A string representing the API key that authenticates access to the LLM service. This key ensures that the service can securely connect to your model deployment.

• llm_endpoint:
  A string representing the endpoint URL where the LLM model is deployed. This is usually a REST API endpoint provided by your deployment platform (e.g., Hugging Face or other open-source model deployment platforms).

• model:
  A string representing the name of the model used for translation. In this case, the mistralai/Mistral-Nemo-Instruct-2407 model is used. This model determines the capabilities and performance of the translation system.

• For models in MINI_MODELS:

  • max_length = 30: Limits the maximum length of text chunks that can be translated in a single request.
  • max_length_mini_text_chunk = 20: Limits the maximum length of mini text chunks for smaller translations.
• For larger models (those not in MINI_MODELS):
  • max_length = 100: Allows larger text chunks for translation.
  • max_length_mini_text_chunk = 50: Allows larger mini text chunks for more efficient translation handling.

Example Usage

get_text_language()

Detects the language of the provided text synchronously.

from llmtranslate import TranslatorOpenSource

# Create the translator object
translator = TranslatorOpenSource(
    api_key="YOUR_API_KEY", 
    llm_endpoint="YOUR_LLM_ENDPOINT", 
    model="mistralai/Mistral-Nemo-Instruct-2407"
)

# Detect the language of a given text
detected_language = translator.get_text_language("Bonjour tout le monde")

if detected_language is not None:
    print(detected_language.ISO_639_1_code)  # Output: 'fr'
    print(detected_language.ISO_639_2_code)  # Output: 'fra'
    print(detected_language.ISO_639_3_code)  # Output: 'fra'
    print(detected_language.language_name)   # Output: 'French'

async_get_text_language()

Detects the language of the provided text asynchronously.

import asyncio
from llmtranslate import TranslatorOpenSource

# Create the translator object
translator = TranslatorOpenSource(
    api_key="YOUR_API_KEY", 
    llm_endpoint="YOUR_LLM_ENDPOINT", 
    model="mistralai/Mistral-Nemo-Instruct-2407"
)

# Async function to detect language
async def detect_language_async():
    detected_language = await translator.async_get_text_language("Hola, ¿cómo estás?")
    if detected_language is not None:
        print(detected_language.ISO_639_1_code)  # Output: 'es'
        print(detected_language.language_name)   # Output: 'Spanish'

# Run the async function
asyncio.run(detect_language_async())

translate()

Translates text synchronously from one language to another.

from llmtranslate import TranslatorOpenSource

# Create the translator object
translator = TranslatorOpenSource(
    api_key="YOUR_API_KEY", 
    llm_endpoint="YOUR_LLM_ENDPOINT", 
    model="mistralai/Mistral-Nemo-Instruct-2407"
)

# Translate text from one language to another
translated_text = translator.translate(
    text="Hallo, wie geht's?", 
    to_language="en"  # Target language in ISO 639-1 format
)
print(translated_text)  # Output: "Hello, how are you?"

async_translate()

Translates text asynchronously from one language to another.

import asyncio
from llmtranslate import TranslatorOpenSource

# Create the translator object
translator = TranslatorOpenSource(
    api_key="YOUR_API_KEY",
    llm_endpoint="YOUR_LLM_ENDPOINT",
    model="mistralai/Mistral-Nemo-Instruct-2407"
)


# Async function to translate text
async def translate_text_async():
    translated_text = await translator.async_translate(
        text="Cześć, jak się masz?",
        to_language="en"  # Target language in ISO 639-1 format
    )
    print(translated_text)  # Output: "Hello, how are you?"


# Run the async function
asyncio.run(translate_text_async())