Text-to-LoRA (T2L) is a method that uses hypernetworks to generate Low-Rank Adaptation (LoRA) weights for large language models based solely on a textual description of the target task. The system takes a task description as input and produces a LoRA adapter that can be applied to a base LLM to specialize it for that task. This enables rapid, no-code adaptation of foundation models to new benchmarks or domains without traditional fine-tuning.