adding Ollama.py

local_llm/LocalLM.py → Messaging_system/Ollama.py

@@ -18,23 +18,97 @@ class LocalLM:
     #     return response.response
 
     def preprocess_and_parse_json(self, response):
+
         # Remove any leading/trailing whitespace and newlines
         if response.startswith('```json') and response.endswith('```'):
-            cleaned_response = response[len('```json'):-len('```')].strip()
+            response = response[len('```json'):-len('```')].strip()
 
         # Parse the cleaned response into a JSON object
         try:
-            json_object = json.loads(cleaned_response)
+            json_object = json.loads(response)
             return json_object
         except json.JSONDecodeError as e:
             print(f"Failed to parse JSON: {e}")
             return None
 
+
+    def get_llm_response(self, prompt, max_retries=4):
+        """
+        sending the prompt to the LLM and get back the response
+        """
+
+        openai.api_key = self.Core.api_key
+        instructions = self.llm_instructions()
+        client = OpenAI(api_key=self.Core.api_key)
+
+        for attempt in range(max_retries):
+            try:
+                response = client.chat.completions.create(
+                    model=self.Core.model,
+                    response_format={"type": "json_object"},
+                    messages=[
+                        {"role": "system", "content": instructions},
+                        {"role": "user", "content": prompt}
+                    ],
+                    max_tokens=500,
+                    n=1,
+                    # temperature=self.Core.temperature,
+                    temperature=0.7,
+                )
+
+                tokens = {
+                    'prompt_tokens': response.usage.prompt_tokens,
+                    'completion_tokens': response.usage.completion_tokens,
+                    'total_tokens': response.usage.total_tokens
+                }
+
+                try:
+                    content = response.choices[0].message.content
+
+                    # Extract JSON code block
+
+                    output = json.loads(content)
+                    # output = json.loads(response.choices[0].message.content)
+
+                    if 'message' not in output or 'header' not in output:
+                        print(f"'message' or 'header' is missing in response on attempt {attempt + 1}. Retrying...")
+                        continue  # Continue to next attempt
+
+                    else:
+                        if len(output["header"].strip()) > self.Core.config_file["header_limit"] or len(
+                                output["message"].strip()) > self.Core.config_file["message_limit"]:
+                            print(
+                                f"'header' or 'message' is more than specified characters in response on attempt {attempt + 1}. Retrying...")
+                            continue
+
+                    # validating the JSON
+                    self.Core.total_tokens['prompt_tokens'] += tokens['prompt_tokens']
+                    self.Core.total_tokens['completion_tokens'] += tokens['completion_tokens']
+                    self.Core.temp_token_counter += tokens['total_tokens']
+                    return output
+
+                except json.JSONDecodeError:
+                    print(f"Invalid JSON from LLM on attempt {attempt + 1}. Retrying...")
+
+            except openai.APIConnectionError as e:
+                print("The server could not be reached")
+                print(e.__cause__)  # an underlying Exception, likely raised within httpx.
+            except openai.RateLimitError as e:
+                print("A 429 status code was received; we should back off a bit.")
+            except openai.APIStatusError as e:
+                print("Another non-200-range status code was received")
+                print(e.status_code)
+                print(e.response)
+
+        print("Max retries exceeded. Returning empty response.")
+        return None
+
     def get_llm_response(self, prompt, mode, max_retries=10):
         """
         Send the prompt to the LLM and get back the response.
         Includes handling for GPU memory issues by clearing cache and waiting before retry.
         """
+
         for attempt in range(max_retries):
             try:
                 # Try generating the response

messaging_main_test.py

@@ -131,7 +131,7 @@ if __name__ == "__main__":
     number_of_samples = 3
 
     # number of messages to generate
-    number_of_messages = 5
+    number_of_messages = 3
     instructionset = {
         1: "Be highly motivational positive and kind",
         2: "Be highly motivational positive and kind",