Client LLM

Runtime

Model

System context

Contextual information to guide the model's behavior.

Repetition deterrence

Penalizes repeated content. 1 is neutral, >1 reduces, and <1 increases repetitions.

Output diversity

Controls output diversity. Closer to 0 for deterministic outputs.

Randomness level

Controls randomness. 0 for deterministic outputs, higher for more randomness.

Average output length

Desired average length of the generated output.

Shift fill factor

Determines the model's tendency to stick to or shift topics.

Frequency penalty

Determines the model's tendency to stick to or shift topics.