How to beginner · 3 min read

How to use translation pipeline Hugging Face

Q: How to use translation pipeline Hugging Face

Use the Hugging Face transformers library's pipeline function with the task set to translation or a specific language pair like translation_en_to_fr. Initialize the pipeline and pass the text to translate, which returns the translated output.

Quick answer

Use the Hugging Face transformers library's pipeline function with the task set to translation or a specific language pair like translation_en_to_fr. Initialize the pipeline and pass the text to translate, which returns the translated output.

PREREQUISITES

Python 3.8+
pip install transformers>=4.0.0
Internet connection to download model weights

Setup

Install the transformers library from Hugging Face and ensure Python 3.8 or higher is installed. No API key is required for local translation pipelines.

bash

pip install transformers

Step by step

Use the pipeline function from transformers to create a translation pipeline. Specify the translation task or model name for the language pair. Pass the input text to get the translated result.

python

from transformers import pipeline

# Create a translation pipeline from English to French
translator = pipeline("translation_en_to_fr")

# Input text to translate
text = "Hello, how are you?"

# Perform translation
result = translator(text)

# Output the translated text
print(result[0]['translation_text'])

output

Bonjour, comment ça va ?

Common variations

Use different language pairs by changing the pipeline task, e.g., translation_en_to_de for German.
Specify a particular model by passing model="Helsinki-NLP/opus-mt-en-fr" to pipeline.
Use the pipeline asynchronously with asyncio for batch translations.

python

from transformers import pipeline

# Specify a model explicitly
translator = pipeline("translation", model="Helsinki-NLP/opus-mt-en-fr")

text = "Good morning"
result = translator(text)
print(result[0]['translation_text'])

output

Bonjour

Troubleshooting

If you see OSError: Model name 'translation_en_to_fr' was not found, install the latest transformers and check your internet connection.
For slow performance, consider using a smaller model or running on GPU.
If translation output is empty or incorrect, verify the input text encoding and language pair.

✅

Key Takeaways

Use Hugging Face's pipeline with task names like translation_en_to_fr for quick translation.
You can specify any supported model for translation by passing the model parameter.
No API key is needed; models download automatically on first run.
For better performance, run on GPU or choose smaller models.
Handle errors by updating transformers and checking model availability.

Verified 2026-04 · Helsinki-NLP/opus-mt-en-fr, translation_en_to_fr

Verify ↗