Published on October 21, 2025: Text-to-LoRA creates task-specific LoRA adapters for Mistral-7B-Instruct from text descriptions. Across 479 tasks, it reached 67.7% average accuracy.
On October 21, 2025, Text-to-LoRA was reported as a method for generating task-specific LoRA adapters for Mistral-7B-Instruct from natural-language descriptions. Trained on 479 tasks, it achieved 67.7% average accuracy: higher than the base model, but slightly below conventional adapters trained for each task.
The approach aims to generate adapters on demand, without training a new adapter for every task. To check the scope, results, and comparison conditions, consult the original publication and verify how it defines the tasks and accuracy metric; the available summary does not provide those criteria.