Using prompt engineering instead of fine tuning

app.py

@@ -31,14 +31,40 @@ I Fine-Tuned the OpenAi **ada** model on Car Seats Command.
 """
 
 
+# def get_command(command, model, id2label):
+#     """
+#     This function get the classification outputs from openai API
+#     """
+#     completion = openai.Completion.create(
+#         model=model, prompt=f"{command}->", max_tokens=1, temperature=0
+#     )
+#     id = int(completion["choices"][0]["text"].strip())
+#     result = id2label[id] if id in id2label else "unknown"
+#     return result
+
+
 def get_command(command, model, id2label):
     """
     This function get the classification outputs from openai API
     """
+    prompt = f"""
+    We want to control the seats of a car which has features to cool, heat, or massage a seat. The user said "{command}", Which feature we should use to ensure user comfort? Give just the number of the feature.
+    Mapping:
+    1: "massage_seats_on"
+    2: "massage_seats_off"
+    3: "heated_seats_on"
+    4: "heated_seats_off"
+    5: "cooled_seats_on"
+    6: "cooled_seats_off"
+
+    Command_Code:
+    """
+
     completion = openai.Completion.create(
-        model=model, prompt=f"{command}->", max_tokens=1, temperature=0
+        model="text-davinci-003", prompt=prompt, max_tokens=2, temperature=0
     )
     id = int(completion["choices"][0]["text"].strip())
+
     result = id2label[id] if id in id2label else "unknown"
     return result