watsonx.ai

Hi there! im developing an automatic review analysis app for a client, and i'm using WatsonX for my backend and to serve the LLMs, specifically i'm using Granite 3.2 for that. My question is, does anybody has been able to use the "thinking" parameter with WatsonX api? I can't seem how make it work. Im doing certain experiments on Prompt lab, but when i want to do the same on my own scripts i'm getting different results. Maybe should i try adjusting the system prompt or something? Thanks for your time!