Added Weights

Each synthetic user has a standard set of attributes, all of which are unique. One of these attributes is mission , which defines the user's ideal outcome from interacting with the agent. For each interaction, we track the task list (LLM-generated) that your agent executed and the outcome. We use this information to judge the outcome against the user's mission. If the agent achieves the ideal outcome and the user is satisfied, we will add positive weights to the execution graph, and vice versa if the outcome doesn't go as planned.

At the end of the fine-tuning session, you'll receive the weighted execution graph with all these weights added, ready for deployment to your private cloud :)