Update app.py

app.py

@@ -1,54 +1,44 @@
 import gradio as gr
-import torch
 from transformers import BartTokenizer, BartForConditionalGeneration
 
-# Load the model and tokenizer from Hugging Face hub
+# Load model and tokenizer from Hugging Face hub using the provided model name
 model_name = "iimran/SAM-TheSummariserV2"
 tokenizer = BartTokenizer.from_pretrained(model_name)
 model = BartForConditionalGeneration.from_pretrained(model_name)
-model.eval()  # Set the model to evaluation mode
 
-# Function to summarize the input text
+# Define the summarization function
 def summarize(input_text):
-    # Tokenize the input text with truncation (adjust max_length as needed)
-    inputs = tokenizer(input_text, return_tensors="pt", truncation=True, max_length=1024)
-
-    # Create global attention mask: assign global attention to the first token (required by LED)
-    global_attention_mask = torch.zeros(inputs["input_ids"].shape, dtype=torch.long)
-    global_attention_mask[:, 0] = 1
-
-    # Generate the summary using beam search (you can adjust parameters as needed)
+    # Tokenize the input text with truncation
+    inputs = tokenizer(input_text, max_length=1024, truncation=True, return_tensors="pt")
+    
+    # Generate the summary using beam search
     summary_ids = model.generate(
         inputs["input_ids"],
-        attention_mask=inputs["attention_mask"],
-        global_attention_mask=global_attention_mask,
-        max_length=512,
-        num_beams=4,
-        early_stopping=True,
+        num_beams=4,          # Use beam search with 4 beams for quality summaries
+        max_length=128,       # Set maximum length for the generated summary
+        early_stopping=True   # Enable early stopping if all beams finish
     )
-
-    # Decode the generated ids to a summary string, skipping special tokens
+    
+    # Decode the generated summary tokens to a string
     summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
     return summary
 
-# Create a Gradio interface with a title, description, submit button, and larger input text area
+# Create a Gradio interface
 iface = gr.Interface(
-    fn=summarize,  # Function that handles the summarization
+    fn=summarize,
     inputs=gr.Textbox(
         label="Enter Text to Summarize",
-        lines=10,  # Make the input area larger by increasing the number of lines
-        placeholder="Paste or type the text you want to summarize here...",
+        lines=10,
+        placeholder="Paste or type the text you want to summarize here..."
     ),
     outputs=gr.Textbox(
-        label="Summary", 
-        lines=5,  # Adjust output area size (number of lines)
+        label="Summary",
+        lines=5,
         placeholder="Summary will appear here..."
     ),
-    live=False,  # Disable live updates, use the submit button instead
-    allow_flagging="never",  # Optionally disable flagging
-    title="SAM - The Summariser",  # Title of the page
-    description="SAM is a model which will help summarize large knowledge base articles into small summaries.",  # Description of the model
+    title="SAM - The Summariser",
+    description="SAM is a model that summarizes large texts into concise summaries."
 )
 
-# Launch the interface
-iface.launch()
+# Launch the Gradio interface
+iface.launch()