Update app.py

attempting cache

app.py

@@ -51,26 +51,40 @@ metrics = st.multiselect(
 st.markdown("### Please wait for the dataset and models to load (this can take some time if they are big!")
 
 ### Loading data
-data = datasets.load_dataset(dset, split=dset_split)
-st.text("Loaded the "+ str(dset_split)+ " split of dataset "+ str(dset))
+@st.cache 
+def loaddset(d, d_split):
+  data = datasets.load_dataset(d, split=d_split)
+  st.text("Loaded the "+ str(d_split)+ " split of dataset "+ str(d))
+  return(data)
+
+data = loaddset(dset,dset_split)
 
 ### Defining Evaluator
 eval = evaluator("text-classification")
 
 ### Loading models
-
-model_list=[]
-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])
-    model_list.append(models[i])
-  except:
-    print("Sorry, I can't load model "+ str(models[i]))
-
-for i in range (len(model_list)):
-  globals()[f"pipe_{i}"] = pipeline("text-classification", model = models[i], tokenizer = models[i], device=-1)
-  st.text("Loaded pipeline "+ str(models[i]))
+@st.cache 
+def load_models(mod_names):
+  model_list=[]
+  for i in range (len(mod_names)):
+    try:
+      globals()[f"tokenizer_{i}"] = AutoTokenizer.from_pretrained(mod_names[i])
+      globals()[f"model_{i}"] = AutoModelForSequenceClassification.from_pretrained(mod_names[i])
+      model_list.append(mod_names[i])
+    except:
+      print("Sorry, I can't load model "+ str(mod_names[i]))
+  return("Loaded "+ str(len(model_list))+ " models")
+
+load_models(models)
+
+@st.cache 
+def load_pipes(mod_list):
+  for i in range (len(mod_list)):
+    globals()[f"pipe_{i}"] = pipeline("text-classification", model = models[i], tokenizer = models[i], device=-1)
+    st.text("Loaded pipeline "+ str(models[i]))
+  return("Loaded "+ str(len(mod_list))+ " models") 
+
+load_pipes(model_list)
 
 ### Defining metrics 
 for i in range (len(metrics)):