This is a three-way classifier built on top of DeBERTaV3. It classifies text as

  • out-of-domain, similar to sentences from news articles
  • spoken in-domain, similar to sentences from DailyDialog and BOLT SMS/Chat
  • written in-domain, similar to sentences from mobile forum posts

See our EMNLP 2025 paper for details.

Downloads last month
63
Safetensors
Model size
0.2B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for figmtu/deberta-v3-aac-classifier

Finetuned
(605)
this model