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"Togetherai endpoints: {client.list_endpoints()}\n")
 8      print(f"Togetherai completions endpoint info: {client.get_endpoint(endpoint='completions')}\n")
 9      print(f"Togetherai chat endpoint info: {client.get_endpoint(endpoint='chat')}\n")
10      print(f"Togetherai embeddings endpoint info: {client.get_endpoint(endpoint='embeddings')}\n")
11  
12      response_completions = client.predict(
13          endpoint="completions",
14          inputs={
15              "prompt": "Who is the protagonist in Witcher 3 Wild Hunt?",
16              "max_tokens": 200,
17              "temperature": 0.1,
18          },
19      )
20  
21      print(f"Togetherai response for completions: {response_completions}")
22  
23      response_embeddings = client.predict(
24          endpoint="embeddings",
25          inputs={
26              "input": ["Who is Wes Montgomery?"],
27          },
28      )
29  
30      print(f"Togetherai response for embeddings: {response_embeddings}")
31  
32      response_chat = client.predict(
33          endpoint="chat",
34          inputs={
35              "messages": [{"role": "user", "content": "Get out of the sunlight's way Alexander!"}],
36          },
37      )
38  
39      print(f"Togetherai response for chat: {response_chat}")
40  
41  
42  if __name__ == "__main__":
43      main()