Spaces:
Sleeping
Sleeping
File size: 2,627 Bytes
263b8fa 140cdae 263b8fa 140cdae 263b8fa 140cdae | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 | import gradio as gr
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
# Load the tokenizer and model from the model folder
tokenizer = AutoTokenizer.from_pretrained("./model")
model = AutoModelForSequenceClassification.from_pretrained("./model", trust_remote_code=True)
# Set the device (CPU or GPU)
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model = model.to(device)
# Define topic mapping
topic_mapping = {
0: 'Bank Account Services',
1: 'Credit Card or Prepaid Card',
2: 'Others',
3: 'Theft/Dispute Reporting',
4: 'Mortgage/Loan'
}
# Prediction function
def predict_complaint_topic(complaint_text):
if not complaint_text.strip(): # Check if the input is blank or only spaces
return "Please enter a valid complaint text."
encoding = tokenizer.encode_plus(
complaint_text,
add_special_tokens=True,
max_length=128,
return_token_type_ids=False,
padding='max_length',
truncation=True,
return_attention_mask=True,
return_tensors='pt'
)
input_ids = encoding['input_ids'].to(device)
attention_mask = encoding['attention_mask'].to(device)
with torch.no_grad():
outputs = model(input_ids=input_ids, attention_mask=attention_mask)
logits = outputs.logits
predicted_class_id = torch.argmax(logits, dim=1).item()
predicted_topic = topic_mapping[predicted_class_id]
return predicted_topic
# Create Gradio interface
iface = gr.Interface(
fn=predict_complaint_topic, # Function to call for prediction
inputs=gr.Textbox(label="Enter your complaint text"), # Input type (Textbox)
outputs=gr.Textbox(label="Predicted Complaint Topic"), # Output type (Textbox)
live=True, # Enable live prediction as the user types
title="Complaint Topic Classifier", # Title of the app
description=(
"This system leverages a machine learning model to automatically classify consumer complaints "
"into specific categories within the banking sector. The model categorizes complaints into "
"'Bank Account Services', 'Credit Card or Prepaid Card', 'Others', 'Theft/Dispute Reporting', "
"and 'Mortgage/Loan'. Built with the DistilBERT model, it enables efficient categorization of complaints, "
"saving time and improving response accuracy. This model is part of an M.Sc. Data Analytics project aimed "
"at improving operational efficiency in banking complaint management."
) # Modified description of the app
)
# Launch the interface
iface.launch() |