Datasets:
Science-and-Technology_v2: Multilingual STEM Translation
Science-and-Technology_v2 (SAT_v2) is a specialized parallel corpus covering popular science explainers, research announcements, and digital service instructions.
📂 Folder Structure & Quality
Following the <SRC>-<TGT> convention, this dataset includes manually verified translations:
- Path:
Science-and-Technology_v2 / <SRC>-<TGT> / source_reviewed / SAT / *.txt - Quality: All files in this version are source-reviewed (manually translated and verified).
📄 Data Format (TSV)
Tab-separated files with the following headerless structure:
id | src_<lang> | tgt_<lang> | domain="science_and_technology"
Example:
1 प्रौद्योगिकी जीवन को सरल बनाती है। <translated_sentence> science_and_technology
🌐 Supported Language Pairs
- Source: Hindi (hi) → Targets: Assamese (as), Punjabi (pa)
- Source: Tamil (ta) → Targets: Kannada (kn), Malayalam (ml), Telugu (te)
🔬 Data Content
- Popular science magazines and technology adoption guides.
- STEM education snippets and research announcements.
- Digital services instructions and technical FAQs.
🚀 Intended Use
- NMT Training: Fine-tuning models for technical, scientific, and digital domain accuracy.
- Digital Literacy: Powering tools that translate complex technology concepts into regional Indian languages.
- Downloads last month
- 21