add ModelzClient.create_completion classmethod

modelz/client.py

@@ -1,5 +1,5 @@
 from __future__ import annotations
-from typing import Any, Generator
+from typing import Any, Dict, Generator
 from http import HTTPStatus
 from urllib.parse import urljoin
 
@@ -131,3 +131,45 @@ def build(self, repo: str):
 
         ModelzResponse(resp)
         console.print(f"created the build job for repo [bold cyan]{repo}[/bold cyan]")
+
+    @classmethod
+    def create_completion(
+        cls,
+        deployment: str,
+        model: str,
+        prompt: str,
+        params: Dict[str, Any] | None = None,
+        serde: str = "json",
+    ):
+        """Create a completion using the model.
+
+        Args:
+            deployment: deployment ID
+            prompt: The prompt to use for the completion.
+            params: additional request params, will be serialized by `serde`
+            serde: serialize/deserialize method, choose from ("json", "msg", "raw")
+        """
+        # Create an instance of the class.
+        client = cls(deployment=deployment)
+
+        try:
+            from llmspec import LLMSpec
+
+            # Instantiate LLMSpec and transform the prompt
+            llmspec = LLMSpec(prompt)
+            transformed_prompt = llmspec.to_model(model)
+        except ImportError as err:
+            raise ImportError(
+                "llmspec is required for LLM models"
+                "\nPlease install it with the command `pip install llmspec"
+            ) from err
+
+        # Prepare request params
+        request_params = {"prompt": transformed_prompt}
+        if params:
+            request_params.update(params)
+
+        # Get the inference result
+        response = client.inference(request_params, deployment, serde)
+
+        return response