Create app.py

app.py

@@ -0,0 +1,33 @@
+import streamlit as st
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+# Load the merged model
+model_name = "EmTpro01/merged-gemma"  # Replace with your merged model path
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)  # Default device is CPU
+
+# Streamlit UI
+st.title("Text Paraphrasing ")
+st.write("Provide a paragraph, and this AI will paraphrase it for you.")
+
+# Input paragraph
+paragraph = st.text_area("Enter a paragraph to paraphrase:", height=200)
+
+if st.button("Paraphrase"):
+    if paragraph.strip():
+        with st.spinner("Paraphrasing..."):
+            # Prepare the prompt
+            alpaca_prompt = f"Below is a paragraph, paraphrase it.\n### paragraph: {paragraph}\n### paraphrased:"
+            
+            # Tokenize input and move to CPU
+            inputs = tokenizer(alpaca_prompt, return_tensors="pt")
+            
+            # Generate paraphrased text
+            output = model.generate(**inputs, max_new_tokens=200)
+            paraphrased = tokenizer.decode(output[0], skip_special_tokens=True)
+            
+            # Extract the paraphrased portion
+            result = paraphrased.split("### paraphrased:")[1].strip()
+            st.text_area("Paraphrased Output:", result, height=200)
+    else:
+        st.warning("Please enter a paragraph to paraphrase.")