Pruned mDeBERTa
Collection
A collection of vocabulary-pruned mDeBERTa-v3 models (30k/20k) optimized for efficient Indonesian NLP deployment.
•
3 items
•
Updated
mdeberta-hybrid-30k
This model is a vocabulary-pruned version of microsoft/mdeberta-v3-base, specifically optimized for the Indonesian language.
Vocabulary: 30k tokens (Hybrid Indonesian-English)
Note: This model is part of an ongoing research project on efficient Transformer deployment. Full paper and benchmarks will be linked upon publication.
- Downloads last month
- 51
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support