GPT2 Local Model isn't working

## **Bug #1**

When using the **query** command, I found the LLM is giving an error for the KV Caché (it isn't properly set). I need to set the cache to improve exponencially the model performance on every device, no matter the hardware.