BERT for Patents Open-Source Model - Trained on Over 100 Million Patents and Optimized for Patent Analysis

Bert For Patents

Developed by anferico

Google's BERT large version model trained on over 100 million global patents, optimized for patent analysis

EnglishOpen Source License:Apache-2.0 #Patent Text Understanding #Multinational Patent Training #BERT Large Architecture

Downloads 54.25k

Release Time : 3/2/2022

Model Overview

A patent domain-specific model based on BERT large architecture, designed to enhance patent retrieval, classification, and technical analysis efficiency

Model Features

Patent Domain Pre-training

Trained on over 100 million patent documents, incorporating professional terminology and patent document structural features

Multinational Patent Coverage

Training data includes global patents, not limited to the US patent system

Technical Concept Understanding

Excels at capturing technical features and invention points in patents

Model Capabilities

Patent Semantic Search

Technical Field Classification

Patent Similarity Calculation

Patent Claim Analysis

Technology Trend Analysis

Use Cases

Intellectual Property Analysis

Enhanced Patent Retrieval

Improves patent search accuracy through semantic understanding

White paper shows significant improvement over keyword-based retrieval

Technology Landscape Mapping

Automatically identifies technical elements and their relationships in patents

Used in JRC projects for tracking climate change-related technologies

Research Support

Emerging Technology Monitoring

Detects patent application trends in specific fields

🚀 BERT for Patents

BERT for Patents is a model developed by Google, trained on over 100 million patents, not limited to those from the US. It is built upon the BERT_LARGE architecture. This model offers a powerful solution for patent - related analysis, leveraging the capabilities of the BERT framework to understand and process patent texts effectively.

🚀 Quick Start

If you're interested in learning more about the model, you can refer to the following resources:

[Blog post](https://cloud.google.com/blog/products/ai - machine - learning/how - ai - improves - patent - analysis)
White paper
[GitHub page](https://github.com/google/patents - public - data/blob/master/models/BERT%20for%20Patents.md) which contains the original TensorFlow checkpoint.

✨ Features

Large - scale Training: Trained on 100M+ patents, providing a comprehensive understanding of patent language.
Based on BERT_LARGE: Leveraging the power of the BERT_LARGE architecture for enhanced performance.

📚 Documentation

Projects using this model (or variants of it):

[Patents4IPPC](https://github.com/ec - jrc/Patents4IPPC) (carried out by [Pi School](https://picampus - school.com/) and commissioned by the Joint Research Centre (JRC) of the European Commission)

📄 License

This project is licensed under the [Apache 2.0 license](https://www.apache.org/licenses/LICENSE - 2.0).

Additional Information

Property	Details
Model Type	BERT for Patents
Training Data	100M+ patents
Pipeline Tag	fill - mask
Mask Token	[MASK]
Metrics	perplexity
Tags	masked - lm, pytorch

Widget Examples

"The present [MASK] provides a torque sensor that is small and highly rigid and for which high production efficiency is possible."
"The present invention relates to [MASK] accessories and pertains particularly to a brake light unit for bicycles."
"The present invention discloses a space - bound - free [MASK] and its coordinate determining circuit for determining a coordinate of a stylus pen."
"The illuminated [MASK] includes a substantially translucent canopy supported by a plurality of ribs pivotally swingable towards and away from a shaft."

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご