Fixed Logs not Visible with Verbose == True

Author: djokester

groqeval/metrics/answer_relevance.py

@@ -73,7 +73,7 @@ def output_decomposition(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Decomposition of the Output into Statements: %s", response.choices[0].message.content)
+        self.logger.info("Decomposition of the Output into Statements: \n%s", response.choices[0].message.content)
         return Output.model_validate_json(response.choices[0].message.content)
 
     @cached(cache=TTLCache(maxsize=100, ttl=300))
@@ -95,7 +95,7 @@ def score_relevance(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Breakdown of the Answer Relevance Score: %s", response.choices[0].message.content)
+        self.logger.info("Breakdown of the Answer Relevance Score: \n%s", response.choices[0].message.content)
         return ScoredOutput.model_validate_json(response.choices[0].message.content), json.loads(response.choices[0].message.content)
 
     @property

groqeval/metrics/base_metric.py

@@ -11,8 +11,17 @@ def __init__(self, groq_client: Groq, verbose: bool = None):
         self.groq_client = groq_client
         self.aggregation = statistics.mean
         self.logger = logging.getLogger(__name__)
+        handler = logging.StreamHandler()  # Stream handler to output to the console
+        formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
+        handler.setFormatter(formatter)
+        self.logger.addHandler(handler)
+        self.logger.propagate = False
+        
         if verbose:
-            self.logger.setLevel(logging.INFO)
+            self.logger.setLevel(logging.INFO)  # Set to DEBUG to see all levels of logs
+            self.logger.info("Verbose Mode is on.")
+        else:
+            self.logger.setLevel(logging.WARNING)
 
     def groq_chat_completion(self, messages, model, temperature=0.5, response_format=None):
         """

groqeval/metrics/bias.py

@@ -79,7 +79,7 @@ def output_decomposition(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Decomposition of the Output into Opinions: %s", response.choices[0].message.content)
+        self.logger.info("Decomposition of the Output into Opinions: \n%s", response.choices[0].message.content)
         return Output.model_validate_json(response.choices[0].message.content)
 
     @cached(cache=TTLCache(maxsize=100, ttl=300))
@@ -101,7 +101,7 @@ def score_bias(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Breakdown of the Bias Score: %s", response.choices[0].message.content)
+        self.logger.info("Breakdown of the Bias Score: \n%s", response.choices[0].message.content)
         return ScoredOutput.model_validate_json(response.choices[0].message.content), json.loads(response.choices[0].message.content)
 
     @property

groqeval/metrics/context_relevance.py

@@ -86,7 +86,7 @@ def context_decomposition(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Decomposition of the Context into Statements: %s", response.choices[0].message.content)
+        self.logger.info("Decomposition of the Context into Statements: \n%s", response.choices[0].message.content)
         return Context.model_validate_json(response.choices[0].message.content)
 
     @cached(cache=TTLCache(maxsize=100, ttl=300))
@@ -112,7 +112,7 @@ def score_relevance(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Breakdown of the Context Relevance Score: %s", response.choices[0].message.content)
+        self.logger.info("Breakdown of the Context Relevance Score: \n%s", response.choices[0].message.content)
         return ScoredContext.model_validate_json(response.choices[0].message.content), json.loads(response.choices[0].message.content)
 
     @property

groqeval/metrics/faithfulness.py

@@ -87,7 +87,7 @@ def output_decomposition(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Decomposition of the Output into Claims: %s", response.choices[0].message.content)
+        self.logger.info("Decomposition of the Output into Claims: \n%s", response.choices[0].message.content)
         return Output.model_validate_json(response.choices[0].message.content)
 
     @cached(cache=TTLCache(maxsize=100, ttl=300))
@@ -114,7 +114,7 @@ def score_faithfulness(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Breakdown of the Faithfulness Score: %s", response.choices[0].message.content)
+        self.logger.info("Breakdown of the Faithfulness Score: \n%s", response.choices[0].message.content)
         return ScoredOutput.model_validate_json(response.choices[0].message.content), json.loads(response.choices[0].message.content)
 
     @property

groqeval/metrics/hallucination.py

@@ -96,7 +96,7 @@ def context_decomposition(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Decomposition of the Context into Statements: %s", response.choices[0].message.content)
+        self.logger.info("Decomposition of the Context into Statements: \n%s", response.choices[0].message.content)
         return Context.model_validate_json(response.choices[0].message.content)
 
     @cached(cache=TTLCache(maxsize=100, ttl=300))
@@ -118,7 +118,7 @@ def score_hallucination(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Breakdown of the Hallucination Score: %s", response.choices[0].message.content)
+        self.logger.info("Breakdown of the Hallucination Score: \n%s", response.choices[0].message.content)
         return ScoredContext.model_validate_json(response.choices[0].message.content), json.loads(response.choices[0].message.content)
 
     @property

groqeval/metrics/toxicity.py

@@ -78,7 +78,7 @@ def output_decomposition(self):
             temperature=0,
             response_format={"type": "json_object"}
         )
-        self.logger.info("Breakdown of the Toxicity Score: %s", response.choices[0].message.content)
+        self.logger.info("Breakdown of the Toxicity Score: \n%s", response.choices[0].message.content)
         return Output.model_validate_json(response.choices[0].message.content)
 
     @cached(cache=TTLCache(maxsize=100, ttl=300))