|
|
--- |
|
|
language: |
|
|
- hau |
|
|
- eng |
|
|
license: apache-2.0 |
|
|
library_name: transformers |
|
|
tags: |
|
|
- translation |
|
|
- hausa-to-english |
|
|
- opus-mt |
|
|
- seq2seq |
|
|
- hf-aware |
|
|
- africa |
|
|
- low-resource |
|
|
pipeline_tag: translation |
|
|
--- |
|
|
|
|
|
# Hausa → English Translation Model (`localenlp-hau-eng-0.01`) |
|
|
|
|
|
This model is a fine-tuned version of [`Helsinki-NLP/opus-mt-mul-en`](https://huggingface.co/Helsinki-NLP/opus-mt-mul-en) for translating **Hausa (hau) to English (eng)**. It was trained on **16,000 parallel sentence pairs** and achieves strong performance on held-out test sets. |
|
|
|
|
|
Developed and maintained by **LocaleNLP** — advancing African language technologies. |
|
|
|
|
|
--- |
|
|
|
|
|
## Metrics |
|
|
|
|
|
BLEU Score: 49.62 |
|
|
|
|
|
chrF Score: 76.49 |
|
|
|
|
|
|
|
|
## Limitations and Bias |
|
|
|
|
|
Data Limitations: Training data of 15k sentences may not cover all Hausa dialects and domains |
|
|
|
|
|
Bias Potential: May reflect biases present in the training data |
|
|
|
|
|
Format Sensitivity: Performance may vary with input formatting and punctuation |
|
|
|
|
|
Domain Adaptation: Best performance on text similar to training data |
|
|
|
|
|
|
|
|
## Model Card Authors |
|
|
|
|
|
[Mgolo/LocaleNLP] |
|
|
|
|
|
## Framework Versions |
|
|
|
|
|
Transformers 4.30.0 |
|
|
|
|
|
PyTorch 2.0.0+ |
|
|
|
|
|
Tokenizers 0.13.3 |
