Model Summary: This is a multitask transformer-based model designed to jointly learn semantic representations for search queries and web pages, aligned to a shared IAB Content Taxonomy goal space.
It uses a shared RoBERTa encoder with separate task-specific heads and incorporates domain context for web pages. The output embeddings are trained to align with pretrained IAB goal embeddings, allowing both classification and nearest-neighbor retrieval.
- Downloads last month
- 14
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support
Model tree for Mozilla/iab-multitask-inference
Base model
distilbert/distilroberta-base