AI4BharatIndicNER
IndicNER
IndicNER fine-tunes bert-base-multilingual-uncased on entity data mined from Samanantar. It recognizes entities across 11 Indian languages.
Restricted access
The paper and repository do not necessarily contain the same checkpoint
A December 2022 card update says the Naamapadam paper reports a different model and promises a repository update. The card does not confirm that change. The published checkpoint should therefore not inherit the paper's results without verification.IndicNER official model card
Check entity boundaries in the released checkpoint
Practical guidance. Use names and places from the application domain. Correct entity type and correct span boundaries are separate questions, especially for unfamiliar names.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Restricted access |
| Access | Downloadable |
| Model identifiers | IndicNER |
| License | MIT; gated weights.IndicNER official model card |
| Languages | Assamese, Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil and Telugu.IndicNER official model card |
Related models
- IndicBERT v2 Masked Language Model
AI4Bharat ยท IndicBERT
Multilingual IndicBERT v2 encoder trained with the masked-language-model objective.
Launch date not confirmedDownloadable