- use models

app.py

@@ -8,6 +8,16 @@ from constants import tweet_generator_prompt, absa_prompt
 # Adjust the layout for wider containers
 st.set_page_config(layout="wide")
 
+
+# @st.cache_resource
+def load_model():
+    classification_pipe = pipeline(
+        "text-classification", model="tweetpie/toxic-content-detector", top_k=None)
+    absa_pipe = pipeline("text2text-generation", model="tweetpie/stance-aware-absa")
+    tweet_generation_pipe = pipeline("text2text-generation", model="tweetpie/stance-directed-tweet-generator")
+    return classification_pipe, absa_pipe, tweet_generation_pipe
+
+
 # Set up the title
 st.title("Towards a Programmable Humanizing AI through Scalable Stance-Directed Architecture Dashboard")
 
@@ -40,23 +50,30 @@ with col12:
     generate_button = st.button("Generate tweet and classify toxicity")
 
 # Load the model (commented out, assuming model loading is handled elsewhere)
-# classifier, absa, generator = load_model()
+classifier, absa, generator = load_model()
 
 # Process the input text and generate output
 if generate_button:
     with col23:  # This block is for displaying outputs in the wider column
         with st.spinner('Generating the tweet...'):
-            # Example placeholders for the generated outputs
-            time.sleep(5)  # Simulating delay
-            generated_tweet = [{
-                                   "generated_text": "the agricultural sector is the single biggest recipient of migrants workers rights groups argue . nearly 90 % of those who come to the us are denied employment due to discriminatory employment laws and safety standards ."}]
+            prompt = tweet_generator_prompt.format(
+                ideology=model_selection.lower(),
+                pro_entities=pro_entities,
+                anti_entities=anti_entities,
+                neutral_entities=neutral_entities,
+                pro_aspects=pro_aspects,
+                anti_aspects=anti_aspects,
+                neutral_aspects=neutral_aspects
+            )
+            generated_tweet = generator(prompt, max_new_tokens=80, do_sample=True, num_return_sequences=3)
 
             # Displaying the input and model's output
-            st.write(f"Generated Tweet: {generated_tweet[0]['generated_text']}")
+            st.write(f"Generated Tweet-1: {generated_tweet[0]['generated_text']}")
+            st.write(f"Generated Tweet-2: {generated_tweet[1]['generated_text']}")
+            st.write(f"Generated Tweet-3: {generated_tweet[2]['generated_text']}")
 
         with st.spinner('Generating the Stance-Aware ABSA output...'):
-            time.sleep(3)  # Simulating delay
-            absa_output = [{'generated_text': 'migrants:positive, rights:positive, laws:positive, safety:positive'}]
+            absa_output = absa(absa_prompt.format(generated_tweet=generated_tweet[0]['generated_text']))
 
             stances = [x.strip() for x in absa_output[0]['generated_text'].split(',')]
             stances = [{
@@ -69,10 +86,7 @@ if generate_button:
             st.table(stances_df)
 
         with st.spinner('Classifying the toxicity...'):
-            time.sleep(2)  # Simulating delay
-            model_output = [[{'label': 'LABEL_0', 'score': 0.9999998807907104},
-                             {'label': 'LABEL_1', 'score': 1.1919785395889282e-07},
-                             {'label': 'LABEL_2', 'score': 1.1919785395889282e-07}]]
+            model_output = classifier(generated_tweet[0]['generated_text'])
             output = model_output[0]
 
             st.write("Toxicity Classifier Output:")