Deploying as pipeline

README.md

@@ -1,6 +1,6 @@
 ---
 title: Question Generator
-emoji: 🐨
+emoji: 🐉
 colorFrom: gray
 colorTo: red
 sdk: streamlit

app.py

@@ -1,47 +1,37 @@
 import streamlit as st
 import nltk
-from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+from transformers import pipeline
 
 # Initialize NLTK
 nltk.download('punkt') 
 # Instantiate the model
-model = AutoModelForSeq2SeqLM.from_pretrained("dantedgp/question-generator")
-tokenizer = AutoTokenizer.from_pretrained("google-t5/t5-small")
-
-def generate_questions(txt):
-    input_text = f"ask: {txt}"
-    input_ids = tokenizer(input_text, return_tensors="pt").input_ids
-    outputs = model.generate(input_ids, max_new_tokens=100, do_sample=False)
-    question = tokenizer.decode(outputs[0], skip_special_tokens=True)
-
-    print('question: ', question)
-    return question
+model = pipeline(model="dantedgp/question-generator", tokenizer="google-t5/t5-small")
 
 st.set_page_config(page_title='AI Quiz Generator')
 st.title('AI Quiz Generator')
+
 text_input = st.text_area(
     'Enter your text', 
-    value="""
-      Algae cells can be used to synthesize lipids for the production of biofuels and foods. 
+    value="""Algae cells can be used to synthesize lipids for the production of biofuels and foods. 
       Integral transmembrane proteins shuttle nutrients in and out of the cell. 
       Liver cells can metabolize poisons and alcohol.
       Chocolate helps the brain stay focused.
       Cellular respiration is the inverse of photosynthesis.
       Caffeine blocks adenosine receptors.
       Adenosine Triphosphate (ATP) is the ernergy currency of the cell.
-      The mitochondria is the powerhouse of the cell.
-    """, 
+      The mitochondria is the powerhouse of the cell.""", 
     height=200
 )
 
 generate_button = st.button("Generate")
 
 if generate_button:
-    if text_input != '':
-        sentences = nltk.sent_tokenize(text_input)
-        for sentence in sentences:
-          question = generate_questions(sentence)
-          st.write(question)
-    else:
-      st.warning('Please enter some text.')
-    
+    sentences = nltk.sent_tokenize(text_input)
+    if not sentences:
+        st.warning('Please enter some text.')
+    for sentence in sentences:
+      try:
+        question = model(sentence)[0]['generated_text']
+        st.write(question)
+      except:
+        st.warning('An error occurred while generating the question.')