Datasets:

KMayanja
/

augmented_medical_29

dataset_info:
  features:
    - name: english
      dtype: string
    - name: luganda
      dtype: string
    - name: back_english
      dtype: string
    - name: bleu
      dtype: float64
    - name: topic
      dtype: string
  splits:
    - name: train
      num_bytes: 2558692
      num_examples: 7217
  download_size: 181093
  dataset_size: 2558692
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Synthetic Translation Dataset

This is a dataset created through translation of a monolingual English dataset generated using an LLM, then backtranslated for evaluation on quality.

Domain: Medical

Translation Model used: Sunbird/translate-nllb-1.3b-salt-8bit

LLM for Generation: Groq models

Purpose: Dataset augmentation(Not recommendation for validation and testing.)