IndoBERT NER Model
| Property | Value |
|---|---|
| Parameter Count | 110M |
| License | MIT |
| Base Model | indolem/indobert-base-uncased |
| Framework | PyTorch |
| Performance | F1: 0.838, Accuracy: 0.953 |
What is indobert-model-ner?
indobert-model-ner is a specialized Named Entity Recognition (NER) model built on top of the IndoBERT base architecture, specifically designed for processing Indonesian text. This model represents a fine-tuned version of the indolem/indobert-base-uncased foundation model, optimized for identifying and classifying named entities in Indonesian language content.
Implementation Details
The model utilizes a transformer-based architecture with 110M parameters, trained using the Adam optimizer with carefully tuned hyperparameters (learning rate: 2e-05, batch size: 16). The training process spanned 10 epochs with a linear learning rate scheduler, achieving impressive validation metrics including 83.07% precision and 84.54% recall.
- Transformer-based architecture with F32 tensor precision
- Implemented using PyTorch framework
- Supports TensorBoard integration for monitoring
- Uses Safetensors for efficient tensor storage
Core Capabilities
- Specialized in Indonesian Named Entity Recognition
- High accuracy (95.30%) in entity detection
- Balanced precision-recall trade-off (F1: 0.838)
- Supports inference endpoints for production deployment
Frequently Asked Questions
Q: What makes this model unique?
This model combines the power of BERT architecture with specific optimizations for Indonesian language processing, achieving high accuracy (95.3%) in NER tasks while maintaining balanced precision and recall metrics.
Q: What are the recommended use cases?
The model is ideal for applications requiring Named Entity Recognition in Indonesian text, such as information extraction, content analysis, and automated document processing systems focused on Indonesian language content.
