Behaviors
Behavior settings can be defined in dict_base_settings.yaml, or in any character card file. Any settings in the character file will override ones in dict_base_settings.yaml.
See example character M1nty for usage.
The following settings have been part of the bot for a long time (from parent project). These are pretty basic, some of which are for very edge cases.
| Setting | Value Type | Description | Default |
|---|---|---|---|
| reply_to_itself | float | 0.0 = never happens / 1.0 = always happens | 0.0 |
| chance_to_reply_to_other_bots | float | Chance for bot to reply when other bots speak in main channel | 0.0 |
| reply_to_bots_when_addressed | float | Chance for bot to reply when other bots mention it by name | 0.0 |
| only_speak_when_spoken_to | boolean | This value gets ignored if you're talking in the bot's main channel | true |
| ignore_parentheses | boolean | Bot ignores you if you write like this | true |
| go_wild_in_channel | boolean | Whether or not the bot will always reply in the main channel | true |
| conversation_recency | int | How long (in seconds) a user is determined to be "in active conversation". Only applicable if 'go_wild_in_channel' is false | 600 |
These recent behaviors can allow your character to behave more humanistic (or like computer program, by default), by causing the character to go idle, pause to read messages before writing, and type at different speeds.
This DOES NOT actually slow down task management / performance at all! This only parks and schedules message response sends, while simulating all the delays!
| Setting | Value Type | Description | Default |
|---|---|---|---|
| maximum_typing_speed | int | Words per Minute. 80 = 80 WPM (good typist) / -1 = no restriction (default). Basically a shortcut to the 'max_tokens_second' param (scales it to WPM). | -1 |
| responsiveness | float | Controls "how much the character is paying attention". 0.0 = Distracted human / 1.0 = Computer program. | 1.0 |
| msg_size_affects_delay | boolean | Will make the bot respond faster for shorter messages, and slower for longer messages | false |
| max_reply_delay | float | Maximum delay before bot begins to respond (-seconds-) | 30.0 |
Allows your character to send messages "unprompted", by picking a random prompt from a list. Dynamic Prompting and Variables can make this very effective.
These settings can be used to make an automatic image prompting generator character, leveraging Tags and Dynamic Prompting in the prompt(s).
| Setting | Value Type | Description | Default |
|---|---|---|---|
| spontaneous_msg_chance | float | 0.0 = never happens / 1.0 = always happens (guaranteed to happen prior to or at 'max_delay') | 0.0 |
| spontaneous_msg_max_consecutive | int | Number of allowed spontaneous messages without User activity. -1 = Unlimited / other value like 10 would be 10 messages | 1 |
| spontaneous_msg_min_wait | float | Minimum wait before bot may spontaneously message (-minutes-) | 10.0 |
| spontaneous_msg_max_wait | float | Maximum wait before bot may spontaneously message (-minutes-) | 60.0 |
| spontaneous_msg_prompts | list | Random prompt will be picked from this list if triggered | ['[SYSTEM] The conversation has been inactive for {time_since_last_msg}, so you should say something.'] |