Update components/LLM.py

components/LLM.py

@@ -1,32 +1,34 @@
 import openai
 import time
+import tiktoken
+from components.configs import LLMtemp, gptmodel
 
 #Call to Openai Chat Model
 def chatopenai(finalprompt):
     response = openai.ChatCompletion.create(
-        model="gpt-3.5-turbo",
+        model=gptmodel,
         messages=finalprompt,
-    )
-    print (f"got API response from openAI:\n{response}\n")
-    API_response = response
+        temperature = LLMtemp,
+        )
+    # API_response = response should I want to store the full API returned content...
     assistant_response = response.choices[0].message['content']
     usedtokens = response["usage"]["total_tokens"]
-    return API_response, assistant_response, usedtokens
+    return assistant_response, usedtokens
 
-def chatopenaiandretry(finalprompt, max_retries=3, retry_delay=1):
-    default_response = None
+#Call to Openai Chat Model 3 times, if it fails will return default messages back
+def chatopenairetry(finalprompt, max_retries=3, retry_delay=1):
     default_assistant_response = "Sorry, I'm having issue with OpenAI API, it happens... could you retry?"
     default_used_tokens = 0
-
     for _ in range(max_retries):
         try:
             response = openai.ChatCompletion.create(
-                model="gpt-3.5-turbo",
+                model=gptmodel,
                 messages=finalprompt,
-            )
+                temperature = LLMtemp,
+                )
             assistant_response = response.choices[0].message['content']
             used_tokens = response["usage"]["total_tokens"]
-            return response, assistant_response, used_tokens
+            return assistant_response, used_tokens
         except Exception as e:
             print(f"Encountered an error: {e}")
             print("Retrying the API call...")
@@ -34,4 +36,26 @@ def chatopenaiandretry(finalprompt, max_retries=3, retry_delay=1):
 
     # If all retries fail, return dummy values
     print("Maximum retries reached. Unable to get a response from OpenAI.")
-    return default_response, default_assistant_response, default_used_tokens
+    return default_assistant_response, default_used_tokens
+
+
+# Call to Openai with the stream of content
+def chatopenaistream(finalprompt, callback, LLMtemp=LLMtemp):
+    assistant_response = ""
+    for chunk in openai.ChatCompletion.create(
+        model=gptmodel,
+        messages=finalprompt,
+        temperature = LLMtemp,
+        stream=True,
+        ):
+        content = chunk["choices"][0].get("delta", {}).get("content")
+        if content is not None:
+            assistant_response += content
+            yield callback(content)  # Yield the result of the callback function
+
+#token calculator to be run when using the streaming. Will calculate an estimate after the stream ended
+def calculatetokens(finalprompt,assistant_response):
+    encoding = tiktoken.encoding_for_model(gptmodel)
+    usedtokens = len(encoding.encode(assistant_response + f"{finalprompt}"))
+    usedtokens=0 
+    return usedtokens