🧰 LoRA

The items that name it 3

• SLDR: Defending Against Malicious Fine-tuning via Selective Layers Recovery and Dynamic Routing
  paperas classification.technologies2026-10-07
• DistillGuard: Malicious NPM Package Detection and API Attack Chain Analysis via Static Graph and LLM Distillation
  paperas classification.technologies2026-09-24
• How abliterated models can get you pwned
  writeupas toolsIt is written here as LoRA: small adapter used to implant the backdoor, about 43 million trainable parameters (roughly 0.6 percent of the base model), merged into base weights afterward.

  roughly 0.6 percent of the base model

In every workspace you can read 55