ner-english-large
| Property | Value |
|---|---|
| Research Paper | FLERT: Document-Level Features for Named Entity Recognition |
| Downloads | 596,836 |
| Framework | PyTorch / Flair |
| Training Data | CoNLL-2003 |
What is ner-english-large?
The ner-english-large is a sophisticated Named Entity Recognition (NER) model developed using the Flair framework, achieving an impressive F1-score of 94.36% on the CoNLL-03 dataset. It's designed to identify and classify four types of entities: Person (PER), Location (LOC), Organization (ORG), and Miscellaneous (MISC) names in English text.
Implementation Details
This model leverages document-level XLM-R embeddings and implements the FLERT architecture. It's built without CRF or RNN layers, instead utilizing a transformer-based approach with XLM-RoBERTa large as its backbone. The model employs fine-tunable transformer embeddings with document context and is trained using the AdamW optimizer.
- Uses XLM-RoBERTa large embeddings with document context
- Implements first subtoken pooling strategy
- Trained for 20 epochs with a small learning rate of 5.0e-6
- Utilizes a hidden size of 256 dimensions
Core Capabilities
- High-accuracy entity detection across four categories
- Context-aware named entity recognition
- Efficient processing of document-level features
- Seamless integration with the Flair framework
Frequently Asked Questions
Q: What makes this model unique?
This model stands out due to its implementation of the FLERT architecture, which incorporates document-level features for enhanced NER performance. Its high F1-score of 94.36% on the CoNLL-03 dataset demonstrates its superior accuracy in entity recognition tasks.
Q: What are the recommended use cases?
The model is ideal for applications requiring accurate named entity recognition in English text, such as information extraction, document analysis, and automated content categorization. It's particularly useful when context-aware entity recognition is crucial for accuracy.
