Expand description
LoRA (Low-Rank Adaptation) support for ML models
This module provides comprehensive LoRA functionality including detection, loading, merging, and saving of LoRA adapters. LoRA enables efficient fine-tuning of large models by learning low-rank updates to weight matrices.
Modules§
- lora
- LoRA loading and saving utilities
Structs§
- LoRA
Adapter - Complete LoRA adapter with all modules
- LoRA
Config - LoRA adapter configuration
- LoRA
Model - LoRA model that combines base model with adapter
- LoRA
Weights - LoRA adapter weights for a single module