Model Card for Fine-tuned Helsinki-NLP/opus-mt-en-hi on IITB English-Hindi Dataset

Model Details

Model Description

This model is a fine-tuned version of Helsinki-NLP/opus-mt-en-hi for English-to-Hindi translation using the IIT Bombay English-Hindi Parallel Corpus. The model is designed to translate English sentences into Hindi with improved accuracy and fluency.

  • Developed by: shogun-the-great
  • Model type: Seq2Seq (Sequence-to-Sequence) for Translation
  • Language(s): English to Hindi
  • License: Apache-2.0 (or specify your license)
  • Finetuned from model: Helsinki-NLP/opus-mt-en-hi

Model Sources

Uses

Direct Use

This model can be directly used for English-to-Hindi translation tasks, such as:

  • Translating text-based content (e.g., documents, articles) from English to Hindi.
  • Assisting in bilingual applications requiring English-Hindi translation.
  • Language learning and cross-lingual understanding.

Out-of-Scope Use

This model may not perform well on:

  • Specialized domains like medical, legal, or technical text.
  • Translation of highly idiomatic, ambiguous, or informal sentences.

Bias, Risks, and Limitations

Bias

The model may inherit biases from the IIT Bombay English-Hindi dataset, such as:

  • Translation bias in cultural, gender, or regional contexts.
  • Limited coverage of less frequent phrases or idioms.

Risks

  • Inaccurate translations in critical scenarios (e.g., medical or legal use cases).
  • Possible loss of nuance or meaning in complex sentences.

Recommendations

  • Validate translations for critical use cases.
  • Fine-tune further on domain-specific datasets if required.

How to Get Started with the Model

You can load and use the fine-tuned model directly from the Hugging Face Hub:

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# Load the tokenizer and model
model_name = "YourUsername/finetuned-opus-mt-en-hi"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

# Example usage for translation
text = "Hello, how are you?"
inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=128)
translation_ids = model.generate(inputs['input_ids'], max_length=128, num_beams=4, early_stopping=True)

# Decode the translated text
translation = tokenizer.decode(translation_ids[0], skip_special_tokens=True)
print("Translation:", translation)
Downloads last month
38
Safetensors
Model size
75.9M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.