Capability
Lora And Textual Inversion Adapter Composition
2 artifacts provide this capability.
Want a personalized recommendation?
Find the best match →Top Matches
1
stable-diffusion-webuiRepository58/100
Stable Diffusion web UI
Unique: Implements LoRA weight merging via low-rank matrix injection into UNet/text encoder layers with per-adapter strength scaling, and textual inversion via token replacement in CLIP tokenizer. Supports simultaneous composition of multiple LoRA adapters with independent strength control. Automatic discovery and caching of embeddings from directory structure.
vs others: Lighter-weight than full model fine-tuning (10-100MB vs 4-7GB) and more flexible than single-style checkpoints (compose multiple adapters, adjust strength dynamically)