This is a fine-tuned BERT4patent model that is suitable for generating embeddings for patent texts. The resulting embeddings are suitable for a variety of downstream tasks, including classification, clustering, and retrieval. This model leverages domain-specific training following the mirror-bert approach to further capture the language in patent documents, providing high-quality representations for the analysis of patents.

Downloads last month
25
Safetensors
Model size
345M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported third-party Inference Providers, and HF Inference API was unable to determine this model's library.