Update app.py

app.py

@@ -44,30 +44,24 @@ st.markdown("### What two metrics do you want to compare?")
 
 metrics = st.multiselect(
      'Choose the metrics for the comparison', 
-     options=['f1', 'accuracy', 'precision', 'recall'])
+     options=['f1', 'accuracy', 'precision', 'recall'],
+     default=["f1", "accuracy"])
      
      
 st.markdown("### Please wait for the dataset and models to load (this can take some time if they are big!")
 
 ### Loading data
-try:
-  data = datasets.load_dataset(dset, split=dset_split)
-  st.text("Loaded the validation split of dataset "+ str(dset))
-except:
-  data = datasets.load_dataset(dset, split="test") 
-  st.text("Loaded the test split of dataset "+ str(dset))
-                   
-  st.text("Sorry, I can't load this dataset... try another one!")
-     
+data = datasets.load_dataset(dset, split=dset_split)
+st.text("Loaded the "+ str(dset_split)+ " split of dataset "+ str(dset))
+
+### Defining Evaluator
+eval = evaluator("text-classification")
+
 ### Loading models
 
 for i in range (len(models)):
-  try:
-    globals()[f"tokenizer_{i}"] = AutoTokenizer.from_pretrained(models[i])
-    globals()[f"model_{i}"] = AutoModelForSequenceClassification.from_pretrained(models[i])
-    st.text("Loaded model "+ str(models[i]))
-  except:
-    st.text("Sorry, I can't load model "+ str(models[i]))
+  globals()[f"pipe_{i}"] = pipeline("text-classification", model = models[i], tokenizer = models[i], device=0)
+  st.text("Loaded model "+ str(models[i]))
 
 ### Defining metrics 
 for i in range (len(metrics)):
@@ -77,8 +71,9 @@ for i in range (len(metrics)):
      st.text("Sorry, I can't load metric "+ str(metrics[i]) +"... Try another one!")
 
 
-### Defining Evaluator
-eval = evaluator("text-classification")
+
+
+
 
 ### Defining pipelines
 
@@ -100,7 +95,7 @@ for i in range (len(models)):
   except:
      st.text("Sorry, I can't load model "+ str(models[i]))
 
-pipe1 = pipeline("text-classification", model=model1, tokenizer= tokenizer1, device=0)
+
 res_accuracy1 = eval.compute(model_or_pipeline=pipe1, data=data, metric=accuracy,
                        label_mapping={"NEGATIVE": 0, "POSITIVE": 1},)
 res_f11 = eval.compute(model_or_pipeline=pipe1, data=data, metric=f1,