SAP Support Ticket Classifier

Fine-tuned bert-base-uncased that classifies SAP enterprise support tickets into the correct SAP functional module.

Live Demo: https://karthionlupenix-sap-ticket-classifier-app.hf.space

Model ID: KarthionLupenix/sap-ticket-classifier-bert

Labels

ID Label SAP Module
0 FI Finance (GL, AP, AR, Asset Accounting, Bank)
1 MM Materials Management (Purchasing, Inventory, MRP)
2 SD Sales & Distribution (Orders, Delivery, Billing)
3 HR Human Resources (Payroll, Time, Org Mgmt)

Pipeline Overview

SAP Ticket Classification Pipeline

Usage

from transformers import pipeline

clf = pipeline("text-classification", model="KarthionLupenix/sap-ticket-classifier-bert")
result = clf("Vendor invoice not posted due to GR/IR account mismatch")
# [{'label': 'FI', 'score': 0.97}]

Or with all scores:

clf = pipeline(
    "text-classification",
    model="KarthionLupenix/sap-ticket-classifier-bert",
    return_all_scores=True
)
clf("MRP not generating planned orders for raw material")
# [{'label': 'FI', 'score': 0.01}, {'label': 'MM', 'score': 0.96}, ...]

Training Details

  • Base model: bert-base-uncased
  • Architecture: BertForSequenceClassification ([CLS] pooled โ†’ Linear)
  • Max sequence length: 128
  • Epochs: 4 (early stopping on weighted F1)
  • Batch size: 16
  • Learning rate: 2e-5 with linear warmup (10%)
  • Optimizer: AdamW, weight decay 0.01

Intended Use

Route incoming SAP support tickets automatically to the correct functional team (Finance, MM, SD, HR) without manual triaging.

Limitations

  • Trained on a small seed dataset โ€” production accuracy improves significantly with more labeled examples per class (500+ recommended).
  • Covers 4 SAP modules; extend LABELS in config.py and retrain to add PP, PM, PS, QM, etc.
  • English only.
Downloads last month
154
Safetensors
Model size
0.1B params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support