AI Parameter Profiles

This page is accessed via Configuration and Administration > Power Data > General > AI Parameter Profiles.

Use this page to customize the AI parameters using standard AI parameters. You can control the randomness, diversity, size, and frequency of the generated output by changing these parameters. You can add your own profiles to obtain tighter control over the generated text. AI parameter profiles are used in generative AI use cases.

Some default AI parameter profiles include:

• Default: Medium variability.
• Creative: High creativity.
• Constant: Low variability.

Adding an AI Parameter Profile

1. Enter an AI Parameter Profile ID.
2. Enter the Maximum Tokens. This sets the maximum number of tokens the model can generate in its response. Increasing this limit allows the model to provide longer answers, but it may also increase the likelihood of unhelpful or irrelevant content being generated.
3. Select the Summarize Format. This is the format for AI summarizations.
4. Select the Summarize Length. This is the length for AI summarizations.
5. Enter the Temperature of the model. Increasing the temperature will make the model answer more creatively.
6. Enter the Top P. Works together with Top K. A higher value, for example 0.95, will lead to more diverse text, while a lower value, such as 0.5, will generate more focused and conservative text.
7. Enter the Top K. This reduces the probability of generating nonsense. A higher value, such as 100, will give more diverse answers, while a lower value, such as 10, will be more conservative.
8. Enter the Frequency Penalty. This sets a scaling bias against tokens to penalize repetitions, based on how many times they have appeared. A higher value, such as 1.5, will penalize repetitions more strongly, while a lower value, such as 0.9, will be more lenient. At 0, it is disabled.
9. Enter the Presence Penalty. This sets a flat bias against tokens that have appeared at least once. A higher value, such as 1.5, will penalize repetitions more strongly, while a lower value, such as 0.9, will be more lenient. At 0, it is disabled.
10. Enter the Seed. This sets the random number seed to use for generation. Setting this to a specific number will make the model generate the same text for the same prompt.
11. Select the Echo Prompt checkbox if you want the AI to rephrase your prompt as part of the results.
12. Select the Stream Results checkbox if you want the results to appear as they are created, instead of all at once when the process is complete.
13. Select a domain from the Domain Name drop-down list.
14. Click Finished.