Spaces:
Sleeping
Sleeping
| import gradio as gr | |
| from transformers import pipeline | |
| # 1. Use a lighter model and GPU if available | |
| summarizer = pipeline( | |
| "summarization", | |
| model="sshleifer/distilbart-cnn-12-6", | |
| device=0 # set to -1 for CPU-only | |
| ) | |
| def chunked_summary(text, chunk_size=800): | |
| tokens = text.split() | |
| chunks = [" ".join(tokens[i:i+chunk_size]) for i in range(0, len(tokens), chunk_size)] | |
| summaries = [ | |
| summarizer(c, max_length=80, min_length=20, do_sample=False)[0]["summary_text"] | |
| for c in chunks | |
| ] | |
| return " ".join(summaries) | |
| def classify_action(email_text): | |
| email_lower = email_text.lower() | |
| if "meeting" in email_lower or "schedule" in email_lower: | |
| return "Schedule a meeting" | |
| elif "question" in email_lower or "reply" in email_lower or "can you" in email_lower: | |
| return "Reply" | |
| elif "unsubscribe" in email_lower or "spam" in email_lower: | |
| return "Delete or Mark as Spam" | |
| else: | |
| return "Read and Archive" | |
| def summarize_and_recommend(email_text): | |
| if not email_text.strip(): | |
| return "No content provided.", "No action" | |
| # 2. Decide whether to chunk | |
| word_count = len(email_text.split()) | |
| if word_count > 800: | |
| summary = chunked_summary(email_text) | |
| else: | |
| summary = summarizer( | |
| email_text, | |
| max_length=80, | |
| min_length=20, | |
| do_sample=False | |
| )[0]['summary_text'] | |
| action = classify_action(email_text) | |
| return summary, action | |
| iface = gr.Interface( | |
| fn=summarize_and_recommend, | |
| inputs=gr.Textbox(lines=15, placeholder="Paste your email here..."), | |
| outputs=[ | |
| gr.Textbox(label="Summary"), | |
| gr.Textbox(label="Suggested Action") | |
| ], | |
| title="📩 Smart Email Summarizer & Action Recommender", | |
| description="Faster summarization with a distilled model and length controls.", | |
| ) | |
| iface.launch(server_name="0.0.0.0", server_port=7860) | |