1  from mlflow.deployments import get_deploy_client
 2  
 3  
 4  def main():
 5      client = get_deploy_client("http://localhost:7000")
 6  
 7      print(f"Mistral endpoints: {client.list_endpoints()}\n")
 8      print(f"Mistral completions endpoint info: {client.get_endpoint(endpoint='completions')}\n")
 9  
10      # Completions request
11      response_completions = client.predict(
12          endpoint="completions",
13          inputs={
14              "prompt": "How many average size European ferrets can fit inside a standard olympic?",
15              "temperature": 0.1,
16          },
17      )
18      print(f"Mistral response for completions: {response_completions}")
19  
20      # Embeddings request
21      response_embeddings = client.predict(
22          endpoint="embeddings",
23          inputs={
24              "input": [
25                  "How does your culture celebrate the New Year, and how does it differ from other countries' "
26                  "celebrations?"
27              ]
28          },
29      )
30      print(f"Mistral response for embeddings: {response_embeddings}")
31  
32  
33  if __name__ == "__main__":
34      main()