AI4BharatIndicBART
IndicBART Separate Scripts
IndicBART Separate Scripts is a pretrained sequence-to-sequence model based on the mBART architecture. It covers English and 11 Indian languages.
Separate scripts change the preprocessing contract
This separately trained variant keeps the same language-generation role as IndicBART while removing its shared-script preprocessing requirement.IndicBART Separate Scripts official model card
Compare preprocessing together with model quality
Practical guidance. This variant is useful when your application must keep text in its original script throughout the pipeline. Compare quality per language rather than assuming that simpler preprocessing always improves output.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | IndicBARTSS |
| Training | 452 million sentences and about 9 billion tokens from IndicCorp.IndicBART Separate Scripts official model card |
| Script design | Each language keeps its own script; no Devanagari conversion is required.IndicBART Separate Scripts official model card |
| License | MIT.IndicBART model release, authors and license |
Related models
Multilingual generation model that maps Indian-language text into shared Devanagari script.
Launch date not confirmedDownloadable