Hello 
This is pretty low priority.
A few description missing in the settings.
ai hugging face api key
ai hugging face token limit
ai hugging face model display name
ai hugging face tei endpoint
ai hugging face tei api key
ai gemini api key
ai vllm endpoint
ai vllm api key
post ai helper allowed groups
ai embeddings per post enabled
Thank you 
Running the AI Bot (setting persona) for a test drive here
Saif:
Explain the following site settings
ai hugging face api key
ai hugging face token limit
ai hugging face model display name
ai hugging face tei endpoint
ai hugging face tei api key
ai gemini api key
ai vllm endpoint
ai vllm api key
post ai helper allowed groups
ai embeddings per post enabled
gpt4_bot:
Reading context for: ai_hugging_face_api_key
The ai_hugging_face_api_key is a site setting for Discourse forums that have the Discourse AI plugin installed. This setting is used to store the API key that authenticates requests to the Hugging Face API. The key is an essential component for making calls to Hugging Face’s models for various AI-based features such as text generation, language modeling, etc.
Here’s a breakdown of the context provided for this setting:
settings.yml with the default value as an empty string, indicating no key is set by default. The secret: true parameter indicates that this setting is sensitive and should be treated as confidential information.hugging_face.rb file that prepares an HTTP request, where it adds an Authorization header if the ai_hugging_face_api_key is present. This is standard practice when dealing with API requests that require authentication.ai_hugging_face_token_limit, which suggests that it is related to limiting the number of tokens (pieces of text) generated by an AI model.This setting would need to be provided by a user with the API key from Hugging Face to enable the Discourse forum to communicate with Hugging Face services.
Let’s find out the contexts for the rest of the settings you asked about.
Reading context for: ai_hugging_face_token_limit
Reading context for: ai_hugging_face_model_display_name
Reading context for: ai_hugging_face_tei_endpoint
Reading context for: ai_hugging_face_tei_api_key
Here’s an explanation of each site setting you asked about, arranged by their purpose:
ai_hugging_face_api_key: This setting holds the API key used to authenticate with Hugging Face services for AI functionality within Discourse. It’s a secret setting, meaning the actual key value is not displayed in plain text to ensure security.
ai_hugging_face_token_limit: This setting determines the maximum number of tokens that the Hugging Face model is allowed to generate in a single API request. A token can be thought of as a piece of information in a sentence (like a word), and limiting the number allows you to control the length and computational cost of responses from the model.
ai_hugging_face_model_display_name: The display name for the Hugging Face model being used. This name is used in the user interface to refer to the specific AI model configured for Hugging Face services.
ai_hugging_face_tei_endpoint: This setting is the endpoint URL for Hugging Face’s Text Embeddings Interface (TEI). It’s used when making requests for text embeddings from the service.
ai_hugging_face_tei_api_key: Similar to ai_hugging_face_api_key, this setting specifically stores the API key for accessing the Hugging Face Text Embeddings Interface.
ai_gemini_api_key: The API key for the Gemini AI service, which would be used for AI operations in Discourse related to whatever services Gemini provides.
ai_vllm_endpoint: The endpoint URL for accessing a Very Large Language Model (VLLM) service, this setting would inform Discourse where to send requests for language model-related tasks.
ai_vllm_api_key: The API key used for authenticating with the VLLM service endpoint.
post_ai_helper_allowed_groups: This setting specifies which groups of users are allowed to use AI helper functionality in their posts. It could be a comma-separated list or a specific group name.
ai_embeddings_per_post_enabled: When enabled, this allows for the embedding of AI-generated content per post. Essentially, this allows the AI to generate responses or additional content based on each individual post when configured correctly.
Each setting plays a role in configuring how AI services interact with your Discourse instance. The specific settings for API keys and endpoints will depend on the external AI services you have subscribed to and intend to integrate with your forum for enhanced functionalities such as content generation, summarization, or other intelligent operations.
Saif:
describe each site setting in one sentence
gpt4_bot:
ai_hugging_face_api_key: Holds the secret API key for authenticating with Hugging Face AI services within Discourse.ai_hugging_face_token_limit: Defines the maximum number of tokens that Hugging Face AI can generate for a single request for content.ai_hugging_face_model_display_name: Specifies a name for the Hugging Face AI model for identification in user interfaces.ai_hugging_face_tei_endpoint: The endpoint URL for Hugging Face’s Text Embeddings Interface, used for text analysis features.ai_hugging_face_tei_api_key: The secret API key needed to access Hugging Face’s Text Embeddings Interface.ai_gemini_api_key: The API key required to integrate Gemini AI services with the Discourse platform.ai_vllm_endpoint: The service endpoint for interacting with an external Very Large Language Model.ai_vllm_api_key: A secret key for authentication with the Very Large Language Model (VLLM) service.post_ai_helper_allowed_groups: Identifies which user groups have permission to use AI helper features in their posts.ai_embeddings_per_post_enabled: Enables or disables the addition of AI-generated content embeddings for each post.Saif:
Try shorter and simpler explanations
gpt4_bot:
ai_hugging_face_api_key: API key for Hugging Face AI integrations.ai_hugging_face_token_limit: Max tokens Hugging Face AI can use per request.ai_hugging_face_model_display_name: Name displayed for the Hugging Face AI model.ai_hugging_face_tei_endpoint: URL endpoint for Hugging Face Text Embeddings.ai_hugging_face_tei_api_key: API key for Hugging Face Text Embeddings.ai_gemini_api_key: API key for Gemini AI services.ai_vllm_endpoint: Endpoint URL for a Very Large Language Model service.ai_vllm_api_key: API key for authenticating with VLLM services.post_ai_helper_allowed_groups: User groups allowed to access AI helper features.ai_embeddings_per_post_enabled: Toggle for AI-generated content per post.We’ll try to add these to the site settings soon
A couple need refining but overall this is good
Edited them a bit @Falco
ai_hugging_face_api_key: API key for Hugging Face APIai_hugging_face_token_limit: Max tokens Hugging Face API can use per requestai_hugging_face_model_display_name: Name displayed for the Hugging Face AI modelai_hugging_face_tei_endpoint: URL where the API is running for the Hugging Face text embeddingsai_hugging_face_tei_api_key: API key for Hugging Face text embeddingsai_gemini_api_key: API key for Google Gemini APIai_vllm_endpoint: URL where the API is running for a very large language modelai_vllm_api_key: API key for very large language model APIpost_ai_helper_allowed_groups: User groups allowed to access AI Helper featuresai_embeddings_per_post_enabled: Enabling embeddings per post within topicsFixed in