Infrastructure Layer

LLM (Large Language Model)

Example:

infrastructure {
   llm "My LLM" {
        uid = aiurn:model:id:geepeetee
        provider = aiurn:model:provider:azure 
        model = "gpt-4o"
        endpoint = "https://any.openai.azure.com/"
        version = "2025-01-01-preview"
        aiurn:global:var:temperature = 0.7
        aiurn:global:var:costid = "ewe3949" 
        aiurn:global:var:hello = True 
    }
}

Specs

Give the LLM a name by assign a name to tool %MY_NAME% { ... }

Attribute	Assignment	Rule	Cardinality	Examples
uid	aiurn:model:%MODEL_ID%	append %MODEL_ID% to aiurn:model: the identifier of your llm	1..1	uid = aiurn:model:id:geepeetee
provider	aiurn:model:provider:azure or aiurn:model:provider:openai	specify the provider of the Model [^1]: More Provider to follow.	1..1	provider = aiurn:model:provider:azure
model	%MODEL_NAME%	specify the given name of the model, recognize this with the selected provider. I.e Azure Foundry uses a custom name for this	1..1	model = "gpt-4o"
endpoint	%ENDPOINT%	The endpoint the provider ensures access	1..1	endpoint = "https://any.openai.azure.com/"
version	%VERSION%	a version how to talk to the provider	1..1	version = "2025-01-01-preview"
aiurn:global:var	aiurn::globalvar:%VARNAME% = %MY_VALUE"	add a property to the llm. The parameter names can also overide model options if name is aligned	0..N	aiurn:var:temperature = 0.7