Advanced Topics¶
This section covers advanced usage patterns and customization options.
Topics¶
- Custom Chunkers - Implement your own chunking strategies
- Custom Embedders - Create custom embedding providers
- LLM Chunking - Configure and use LLM-based chunking
- Performance - Optimization and tuning tips
When to Use Advanced Features¶
Custom Chunkers¶
Use custom chunkers when:
- Built-in strategies don't fit your use case
- You need domain-specific chunking (e.g., code, legal text)
- You want to integrate with external NLP libraries
Custom Embedders¶
Use custom embedders when:
- You need a model not supported by FastEmbed
- You want to use external embedding APIs
- You need custom preprocessing or postprocessing
LLM Chunking¶
Use LLM chunking when:
- Semantic coherence is critical
- You have access to LLM infrastructure
- Quality matters more than speed/cost
Performance Tuning¶
Focus on performance when:
- Processing large volumes of data
- Running on resource-constrained systems
- Optimizing for production workloads