Base Model: Helsinki-NLP/opus-mt-en-es
Task: Machine Translation (English to Spanish)
Language: English → Spanish
Fine-tuning Data: Custom parallel corpus with educational and general domain text

Performance

The model achieves a BLEU score of 65.8 on the test set, indicating high-quality translation performance.

Sample Translations

English	Spanish (Reference)	Spanish (Model)
"The London and Rio de Janeiro buildings will be less than 540 meters and the Tokyo building will be taller than 540 meters."	"Los edificios de Londres y Río de Janeiro tendrán menos de 540 metros y el de Tokio será más alto de 540 metros."	"Los edificios de Londres y Río de Janeiro serán menos que 540 metros y el edificio de Tokio será más alto que 540 metros."
"What are some other ways to think about quantities in this problem?"	"¿De qué otras maneras puedes pensar en las cantidades de este problema?"	"¿Qué otras maneras de pensar en las cantidades de este problema?"

Usage

python

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("omkarhurix/fine-tuned-opus-mt-en-es")
model = AutoModelForSeq2SeqLM.from_pretrained("omkarhurix/fine-tuned-opus-mt-en-es")

# Translate text
text = "Hello, how are you?"
inputs = tokenizer(text, return_tensors="pt")
outputs = model.generate(**inputs)
translation = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(translation)  # "Hola, ¿cómo estás?"

Training Details

Training Framework: Hugging Face Transformers
Training Method: Fine-tuning with custom parallel data
Evaluation Metric: BLEU score
Hardware: GPU-accelerated training

Citation

If you use this model, please cite:

bibtex

@misc{fine-tuned-opus-mt-en-es,
  title={Fine-tuned English-Spanish Translation Model},
  author={omkarhurix},
  year={2025},
  howpublished={\url{https://huggingface.co/omkarhurix/fine-tuned-opus-mt-en-es}},
}

License

This model is released under the Apache 2.0 License, same as the base model.