AI4BharatIndicNLG
MultiIndic Headline Generation
This headline-generation model fine-tunes IndicBART on 1.316 million paragraphs. It covers 11 Indian languages.
The input is a passage, not just a title prompt
The model's training pairs connect paragraphs to short outputs. Non-Devanagari languages need conversion before inference and conversion back afterward.MultiIndic Headline Generation official model card
Check whether a headline changes the claim
Practical guidance. A title often drops conditions to become shorter. Review names, dates and qualifications against the source passage before publishing the headline.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | MultiIndicHeadlineGeneration |
| Scripts | Indic text is represented in Devanagari.MultiIndic Headline Generation official model card |
| Training scope | Headline generation and related summarization tasks.MultiIndic Headline Generation official model card |
Related models
Separate-script IndicBART model trained on 1.316 million paragraphs for headline generation in 11 Indian languages.
Launch date not confirmedDownloadable