README.md

metadata

license: mit
language:
  - en
library_name: transformers
pipeline_tag: text-generation
tags:
  - banking
  - sms
  - json
  - parser
  - financial
  - india
datasets:
  - synthetic
widget:
  - example_title: Transaction SMS
    text: >-
      Sent Rs.1500.00 from HDFC Bank AC XX1234 to john@okicici on 15-Aug-25.UPI
      Ref 123456789012.
  - example_title: Credit SMS
    text: >-
      Rs.25000 credited to your SBI Bank a/c XX5678 via NEFT from beneficiary
      COMPANY LTD.

Banking SMS JSON Parser V8

Advanced AI model that converts Indian banking SMS messages into structured JSON format.

Features

• ✅ Detects transaction vs non-transaction messages
• ✅ Extracts amount, date, transaction type, last 4 digits
• ✅ Categorizes transactions into 32+ categories
• ✅ Handles unknown merchants with "Other" category
• ✅ Supports UPI, NEFT, RTGS, Card transactions
• ✅ 60,000+ training samples with realistic Indian banking patterns

Usage

Training Data

• 60,000 training samples
• 6,000 validation samples
• 75% transaction, 25% non-transaction messages
• Realistic Indian banking SMS patterns
• Major Indian banks: ICICI, HDFC, SBI, Kotak, Axis, BOB, YES, etc.

Performance

Optimized for high accuracy on real-world Indian banking SMS messages with proper category classification and transaction detection.