Update app.py

app.py

@@ -1,30 +1,36 @@
-from transformers import pipeline
+from transformers import pipeline, AutoTokenizer
 import gradio as gr
 
-# Use the pipeline with optimized settings (no sampling, smaller batch)
+# Load tokenizer for truncation
+tokenizer = AutoTokenizer.from_pretrained("sshleifer/distilbart-cnn-6-6")
+
+# Load summarization pipeline
 summarizer = pipeline(
     "summarization",
     model="sshleifer/distilbart-cnn-6-6",
-     # force CPU (if not using GPU)
 )
 
-# Function with higher max length and lower min length for longer summaries
+# Truncation helper 🫶
+def safe_truncate(text, max_tokens=1024):
+    tokens = tokenizer.encode(text, truncation=True, max_length=max_tokens)
+    return tokenizer.decode(tokens, skip_special_tokens=True)
+
+# Summary function with truncation applied
 def summarize_article(text):
+    short_text = safe_truncate(text)  # prevent model from breaking!
     summary = summarizer(
-        text,
-        max_length=250,  # ✨ allow longer output
-        min_length=100,  # 🚨 ensure decent length
-        do_sample=False,  # ⚡ makes it deterministic and faster
+        short_text,
+        max_length=250,
+        min_length=100,
+        do_sample=False,
     )
     return summary[0]['summary_text']
 
-# Sample input
-default_article = """New York (CNN)When Liana Barrientos was 23 years old, she got married..."""  # [TRIMMED for brevity]
-
-# Generate summary once to display as default
+# Default example
+default_article = """New York (CNN)When Liana Barrientos was 23 years old, she got married..."""  # [shortened for demo]
 default_summary = summarize_article(default_article)
 
-# Gradio interface (read-only)
+# Gradio Interface
 iface = gr.Interface(
     fn=summarize_article,
     inputs=gr.Textbox(lines=20, label="Article (Read Only)", value=default_article, interactive=False),