Improve provider-model match

L3Score.py

@@ -128,25 +128,30 @@ class L3Score(evaluate.Metric):
                 client = openai.OpenAI(api_key=api_key)
                 model_names = set([model.id for model in client.models.list()])
                 if model not in model_names:
-                    return ValueError(f"Model {model} not found for provider {provider}, available models: {model_names}")
+                    raise ValueError(f"Model {model} not found for provider {provider}, available models: {model_names}")
                 
             elif provider == "deepseek":
+                print("Checking DeepSeek model")
                 client = openai.OpenAI(api_key=api_key,base_url="https://api.deepseek.com")
                 model_names = [model.id for model in client.models.list()]
                 print(model_names)  
                 if model not in model_names:
-                    return ValueError(f"Model {model} not found for provider {provider}, available models: {model_names}")
+                    raise ValueError(f"Model {model} not found for provider {provider}, available models: {model_names}")
             
             elif provider == "xai":
                 client = openai.OpenAI(api_key=api_key, base_url="https://api.xai.com")
                 model_names = [model.id for model in client.models.list()]
                 print(model_names)
                 if model not in model_names:
-                    return ValueError(f"Model {model} not found for provider {provider}, available models: {model_names}")
-                
+                    raise ValueError(f"Model {model} not found for provider {provider}, available models: {model_names}")
+        
+        except ValueError as e:
+            return {"error": str(e)}
         except openai.AuthenticationError as e:
             message = e.body["message"]
             return {"error": f"Authentication failed: {message}"}
+        except Exception as e:
+            return {"error": f"An error occurred when verifying the provider/model match: {e}"}
         
         assert len(questions) == len(predictions) == len(references), "Questions, predictions and references must have the same length"
 

app.py

@@ -1,8 +1,8 @@
 import gradio as gr
 import evaluate
-from L3Score import L3Score
 
-l3score = L3Score()
+
+l3score = evaluate.load("nhop/L3Score")
 
 def compute_l3score(api_key, provider, model, questions, predictions, references):
     try: