Update app.py

app.py

@@ -1,3 +1,6 @@
+from transformers import AutoTokenizer
+import transformers
+from transformers import pipeline
 import en_ner_bc5cdr_md
 import spacy
 from spacy import displacy
@@ -13,6 +16,23 @@ def listToString(s):
     str1 = " ; " 
     return (str1.join(s))
 
+
+@st.cache(allow_output_mutation=True)
+def flair_model(model_name):
+    return SequenceTagger.load(model_name)
+
+
+@st.cache(allow_output_mutation=True)
+def scispacy_model():
+    return en_ner_bc5cdr_md.load()
+
+
+@st.cache(allow_output_mutation=True)
+def questionna(model_nm):
+    return pipeline(model_nm)
+     
+
+
 ### Defining List
 lst_chk = ["PHI Masking","Disease & Drug Extraction","Question Answering"]
 
@@ -31,16 +51,10 @@ hom =st.sidebar.radio("Text Predictions",["Home","Text Prediction Models","Feedb
 
 ### Main Code
 if hom == "Home":
-    #colm1 , colm2, colm3, colm4, colm5 = st.columns(5)
-    #with colm5:
-    #    st.image("data//Goofy.jpg", width = 100)
         
     col1 , col2 = st.columns([0.8,3])
     with col2:
         st.title("GOOFY THE ASSIST :male-doctor:")
-        # :male-doctor:,  :person_doing_cartwheel:
-    #with col3:
-        #st.write("[Gavs](https://www.gavstech.com/)")
     st.caption("")
     st.video("https://www.youtube.com/watch?v=q_hfdvmS8As")
     
@@ -76,7 +90,10 @@ if hom == "Text Prediction Models":
               lst_person = []
               lst_gpe = []
 
-              tagger = SequenceTagger.load("flair/ner-english-ontonotes-large")
+              # tagger = SequenceTagger.load("flair/ner-english-ontonotes-large")
+              # sentence = Sentence(txt_input1)
+              # tagger.predict(sentence)
+              tagger = flair_model('flair/ner-english-ontonotes-large')
               sentence = Sentence(txt_input1)
               tagger.predict(sentence)
 
@@ -128,8 +145,8 @@ if hom == "Text Prediction Models":
                     lst_person = []
                     lst_gpe = []
 
-                    tagger = SequenceTagger.load("flair/ner-english-ontonotes-large")
-                    sentence = Sentence(str(row.PHI_DESCRIPTION))
+                    tagger = flair_model('flair/ner-english-ontonotes-large')
+                    sentence = Sentence(txt_input1)
                     tagger.predict(sentence)
 
                     out_txt = txt_input1
@@ -188,7 +205,8 @@ if hom == "Text Prediction Models":
             if st.button("Submit"):
 
 ##### Scispacy Model
-              nlp_bc5cdr = en_ner_bc5cdr_md.load()
+              # nlp_bc5cdr = en_ner_bc5cdr_md.load()
+              nlp_bc5cdr = scispacy_model()
               doc_nlp_bc5cdr = nlp_bc5cdr(txt_input1)
               #displacy_image = displacy.render(doc_nlp_bc5cdr, jupyter=True,style='ent')
               lst_disease = []
@@ -225,7 +243,8 @@ if hom == "Text Prediction Models":
 #### Scispacy Model full dataframe
                   phi_df_full = pd.DataFrame()
                   for row in df.itertuples():
-                    nlp_bc5cdr = en_ner_bc5cdr_md.load()
+                    # nlp_bc5cdr = en_ner_bc5cdr_md.load()
+                    nlp_bc5cdr = scispacy_model()
                     doc_nlp_bc5cdr = nlp_bc5cdr(str(row.Medical_Record))
                     #displacy_image = displacy.render(doc_nlp_bc5cdr, jupyter=True,style='ent')
                     lst_disease = []
@@ -264,19 +283,14 @@ if hom == "Text Prediction Models":
         qstn = st.selectbox("Select Question",questions,index = 0)
 
         if st.button("Submit"):
-            from transformers import AutoTokenizer
-            import transformers
-            from transformers import pipeline
-            qa_model = pipeline("question-answering")
+            qa_model = questionna('question-answering')
             qa_op = qa_model(question = qstn, context = context)
-
+            
             cnt = 0
             for i in qa_op.values():
               cnt += 1
               if cnt == 4:
                 op_fin = str(i)
-
-            #st.write("Output",op_fin)
             st.success(op_fin)