Powerful chat presets for enhanced responses, preview mode with draft system, advanced lorebook controls, and character wallpapers.
Sampling Parameters Explained
Parameter | Range | Description |
---|---|---|
Temperature | 0.0 – 2.0 | Controls randomness. Lower → more focused; higher → more creative and unpredictable. |
Top P | 0.0 – 1.0 | Nucleus sampling: keeps highest-probability tokens whose cumulative probability ≥ Top P. Lower → tighter vocabulary; higher → more diverse. |
Max Tokens | 1 – 700 | The cap on tokens in the model’s response. Higher → longer replies (and more cost). |
Frequency Penalty | -2.0 – 2.0 | Penalizes frequent tokens. Positive → less repetition of common words/phrases. |
Presence Penalty | -2.0 – 2.0 | Encourages novel content. Positive → more likely to introduce new topics. |
Repetition Penalty | 0.0 – 2.0 | Specifically discourages verbatim repeats. Higher → fewer exact phrase re-uses. |
Top K | 0 – 100 | Limits to the top K most probable tokens. Lower → even tighter focus than Top P. |
Min P | 0.0 – 1.0 | Minimum probability threshold for candidate tokens. Higher → more conservative choices. |