solutions.txt

## 1a read the input field from args in input, return '' as default
input = args.get("input", "")

## 1b return a dictionary with key 'body' and value output
return {
    "body": output 
}

## 1c add the demo entry, an object with name Demo and url mastrogpt/demo
{ 
    "name": "Demo", 
    "url": "mastrogpt/demo",
},

## 2a connect ai with Azure OpenAI and return the api object
return  AzureOpenAI(
    api_version="2023-12-01-preview", 
    api_key=args.get("OPENAI_API_KEY"), 
    azure_endpoint=args.get("OPENAI_API_HOST"))

## 2b retrieve the model we use, check the status  and return 'Welcome.' if is 'succeeded'
global MODEL
model = ai.models.retrieve(MODEL)
if model.status == 'succeeded':
    output = "Welcome."

## 2c add the OpenAI entry, an object with name OpenAI and url openai/chat
{ 
    "name": "OpenAI", 
    "url": "openai/chat",
},

## 3a generate the role as shown in test3.test_request
system = {"role": "system", "content": role}
user = {"role": "user", "content": input}
return [system, user]    

## 3b invoke the chat completion API
comp = ai.chat.completions.create(
    model=MODEL, 
    messages=request(input, role)
)

## 3c read the first message content if any
if len(comp.choices) >0:
    res = comp.choices[0].message.content

##