Training room 6 | Coachvox AI

Fine-tuning

Here’s how to test your AI model in real conversations after training it

Fine-tuning is the most fun part of training your AI because you actually get to interact with it, as if you were the client or audience member.

We recommend that you only do fine-tuning once you’ve completed the previous training rooms. This is so the responses are based on your data and style preferences.

Fine-tuned prompt and completion pairs will show in your fine-tuning history that you can toggle “on” to view. However, because these are not editable, it’s always better to add important content directly into training data.

Troubleshooting responses:

If you see responses from your AI that don’t use your training data – you might need to go back to the training data room and click “save”. Then try again.

If you still think the responses aren’t right, head to training data and search key words or phrases similar to the prompts used in fine-tuning. You can do this using the search bar in the table. See if there is any conflicting information that could be causing confusion for the AI. Also search your fine-tuning history.

If you have the right training data in place, have saved the training data, but are still getting odd responses in fine-tuning, contact support.

Here’s Ben with more info:

I show you how to use the fine-tuning and feedback training room in Coachvox. This is where you can interact with your AI, fine-tune its responses and see how the conversations flow.

I walk you through the functionality as we go, including how to rate responses, edit completions, and avoid duplicate data. This is a quick and efficient way to build the training data for your AI. If you have any questions, just get in touch.

Go back to the training overview page for other training resources.

Fine-tuning

Create an AI version of you

Links

Info

Articles

Legals

Get your FIRST 30 leads

in 14 days: CHALLENGE