Bank User Query Intent Classification Model

This model classifies queries into twelve distinct buckets:

Sample Label: Category, Type, Subtype
Label 0: DEPOSIT, SB/ CA/TERM DEPOSIT ACCOUNTS, Dispute in charges deducted
Label 1: DEPOSIT, SB/ CA/TERM DEPOSIT ACCOUNTS, Minimum Balance Charges related
Label 2: DIGITAL BANKING, ATM RELATED, Dispute in ATM AMC Charges
Label 3: DIGITAL BANKING,FUND REMITTANCE: NEFT/ RTGS/ IMPS through Branch,Dispute in charges deducted
Label 4: LOANS & ADVANCES, Education Loans, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
Label 5: LOANS & ADVANCES, Govt Scheme loans, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
Label 6: LOANS & ADVANCES, Home loans, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
Label 7: LOANS & ADVANCES, OTHER ADVANCES, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
Label 8: LOANS & ADVANCES, Personal loans, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
Label 9: LOANS & ADVANCES, SME ADVANCES, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
Label 10: LOANS & ADVANCES, VEHICLE LOANS, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
Label 11: Others

Training Metrics

The following training metrics were observed over 10 epochs:

Epoch	Loss	Accuracy	F1 Score
1	0.4103	0.9182	0.9184
2	0.0672	0.9827	0.9828
3	0.0351	0.9917	0.9917
4	0.0221	0.9948	0.9948
5	0.0171	0.9942	0.9943
6	0.0107	0.9966	0.9966
7	0.0056	0.9989	0.9989
8	0.0037	0.9986	0.9986
9	0.0152	0.9955	0.9955
10	0.0061	0.9982	0.9982