Here you can see an explanation of what which option does.
To see an example look at our template.config.json.
For your model configurations in models you can take the example_model_configurations.json as an example. The example model configurations also include up to date pricing (as of 7th February 2024), the correct context sizes and other settings.
{
"owner_ids": The ids of the owners discord account(s),
"staff_roles": The roles which your staff have. This will bypass filters and cooldowns (ARRAY OF ROLE IDS),
"staff_users": The staff users who don't have any of the staff roles. This will bypass filters and cooldowns (ARRAY OF USER IDS),
"blacklist_roles": Blacklist users based on their roles. Staff have full bypass (ARRAY OF ROLE IDS),
"default_dalle_model": The default dalle model to use in /create_image *1 *2,
"staff_can_bypass_feature_restrictions": When set to true staff won't be restricted by features turned off (BOOLEAN) *4,
"dev_config": {
"enabled": Whether this is a development instance or not (BOOLEAN) *3,
"debug_discord_messages": Show debug messages in discord,
"debug_logs": Show debug messages in the terminal,
},
"global_user_cooldown": The time until a user can send a new request in milliseconds (NUMBER),
"max_thread_folowup_length": The amount of followup prompts a user can send in a thread (NUMBER),
"allow_collaboration": When set to true anybody can use /chat thread in each others threads (BOOLEAN),
"hey_gpt": {
"enabled": Whether this is enabled or not (BOOLEAN),
"context_depth": The depth of previously messages which will be used as context (NUMBER) *17,
"model": The model to use for this action (STRING),
"processing_emoji": The emoji to use for showing the bot is working (unicode or emoji ID),
"system_instruction": The system instruction for this action (STRING),
"activation_phrases": Phrases which the message has to start with to be activated (ARRAY OF STRINGS) *13
},
"generate_image": {
"quality": "hd" or "standard" *1,
"default_size": "1024x1024" or "1024x1792" or "1792x1024" *1,
},
"assistants": {
"result_fetching_max_count": The number of times to fetch the response results from OpenAI before displaying an error (INTEGER),
"allow_collaboration": When enabled all users who have access to the thread can interact with the assistant (BOOLEAN),
"assistant_ids": Array of assistant IDs that can be used in the bot *1 (ARRAY OF STRINGS)
},
"generation_settings": {
"default_model": The default model to use. Model must support chat completion (STRING) *8,
"default_system_instruction": The system instruction for the chatbot (STRING) *2,
"selectable_models": A list of models the users can select from (ARRAY OF CHAT COMPLETION MODEL NAMES) *1 *4 *12 *14
},
"selectable_system_instructions": An array of selectable system instructions (ARRAY WITH OBJECTS WITH THE PROPERTIES name AND system_instruction) *4,
"logs": {
"enabled": Whether logging is enabled or not (BOOLEAN),
"directory": The directory where the logging files are saved to (STRING),
"plain": Whether to log in a plain text file (BOOLEAN),
"csv": Whether to log in csv format (BOOLEAN)
},
"features": {
"chat_single": Whether this feature is enabled or not (BOOLEAN) *4,
"chat_thread": Whether this feature is enabled or not (BOOLEAN) *4,
"assistants": Whether this feature is enabled or not (BOOLEAN) *4,
"image_in_prompt": Whether this feature is enabled or not (BOOLEAN) *4,
"create_image": Whether this feature is enabled or not (BOOLEAN) *4,
"regenerate_button": Whether this feature is enabled or not (BOOLEAN),
"delete_button": Whether this feature is enabled or not (BOOLEAN),
"view_system_instruction": Whether this feature is enabled or not (BOOLEAN) *4,
"user_stats": Whether to record user stats or not,
"user_leaderboard": Whether this feature is enabled or not (BOOLEAN) *4
},
"leaderboard_amount_users": How many users to display on the leaderboard (NUMBER),
"auto_create_commands": When set to true the commands will be automatically created (BOOLEAN) *11,
"message_context_actions": An Array of message contexts with system instructions (ARRAY WITH OBJECTS WITH THE PROPERTIES name, model (*15) AND system_instruction) *4,
"models": {
"MODEL NAME": {
"model": The configurations model name (STRING, REQUIRED),
"base_url": The base url of the API *1.
"images": {
"supported": Whether images are supported for this model *1,
"detail": "low", "high" or "auto" - The *1,
},
"moderation": {
"enabled": Whether to use OpenAI Moderation on the prompts,
},
"env_token_name": The key of the token in the .env file,
"max_completions_token": The max amount of tokens to generate *10,
"max_tokens": The max tokens for this model *1,
"cost": {
"prompt": The cost for prompts *8,
"completion": The cost for completions *8
},
"defaults": {
"frequency_penalty": The default value for this property *1,
"logit_bias": The default value for this property *1,
"logprobs": The default value for this property *1,
"top_logprobs": The default value for this property *1,
"presence_penalty": The default value for this property *1,
"response_format": The default value for this property *1,
"seed": The default value for this property *1,
"stop": The default value for this property *1,
"temperature": The default value for this property *1,
"top_p": The default value for this property *1
}
} *16
}
}
*1 See API Documentation for reference (https://platform.openai.com/docs/api-reference/chat)
*2 See Instructing chat models for reference (https://platform.openai.com/docs/guides/chat/instructing-chat-models)
*3 Developer mode will enable logging and will also show the generations ID in the embed in Discord
*4 This option changes how the command is created.
*5 It is not recommended to change this option.
*6 Configuration of context_action_instruction is advised.
*7 Prices for models to save money spent by users. Read more about pricing
*8 See API Documentation for compatibility. Read more
*9 The value can not exceed 6000
*10 It is recommended to set this value to -1 to avoid errors.
*11 When set to false changes in the config file will not be applied to commands on startup, only when using /reload_config THE DEFAULT IS TRUE
*12 For better configuration instead of the name the following can be given: {name: string, base_url?: string, supports_images?: boolean, env_token_name?: string}
*13 For better configuration instead of the activation phrase the following can be given: {phrase: string, system_instruction: string, model: string *15, allow_images?: boolean}
*14 When providing env_token_name you have to add a auth key to the .env with the exact name of the env_token_name similar to the OPENAI_TOKEN key
*15 This name has to be a key in models,
*16 Here you can set up external models as well by providing the model name, base url and token. The API is required to be compatible with OpenAI API
*17 The higher you set the number the slower the response will be generated. It is recommended to not go above a depth of 5