Three production-ready applications demonstrating SAP-RPT-1-OSS's In-Context Learning for SAP enterprise data
| # | App | Description | Live Demo |
|---|---|---|---|
| 1 | Finance Dashboard | SAP financial statement analysis & profitability prediction | 🤗 HuggingFace |
| 2 | Forecast Showdown | RPT-1 vs LLM comparison on numeric forecasting | 🤗 HuggingFace |
| 3 | Predictive Integrity | Proactive SAP operations risk prediction (Jobs, Transports, Interfaces) | 🤗 HuggingFace |
| 4 | SAP Local RPT-1 Workspace | SAP-RPT-1-OSS on Microsoft Foundry Hosted | ☁️ Azure |
SAP-RPT-1-OSS-App/
│
├── 📁 apps/ # Application packages
│ ├── 01-finance-dashboard/ # App 1: Financial Analysis
│ │ ├── app.py
│ │ ├── requirements.txt
│ │ ├── Dockerfile
│ │ └── README.md
│ │
│ ├── 02-forecast-showdown/ # App 2: RPT-1 vs LLM
│ │ ├── app.py
│ │ ├── api.py
│ │ ├── requirements.txt
│ │ ├── Dockerfile
│ │ └── README.md
│ │
│ └── 03-predictive-integrity/ # App 3: Operations Risk
│ ├── app.py
│ ├── utils/
│ ├── requirements.txt
│ ├── Dockerfile
│ └── README.md
│
├── 📁 deploy/ # Deployment configurations
│ ├── azure/ # Azure Container Apps
│ │ ├── Dockerfile
│ │ ├── deploy-azure.ps1
│ │ ├── supervisord.conf
│ │ └── README.md
│ │
│ └── huggingface/ # HuggingFace Spaces
│ └── README.md
│
├── 📁 shared/ # Shared utilities
│ ├── data/ # Sample datasets
│ ├── models/ # Model utilities
│ └── utils/ # Common helpers
│
├── 📁 docs/ # Documentation
│ ├── ARCHITECTURE.md
│ ├── API_REFERENCE.md
│ └── DEPLOYMENT_GUIDE.md
│
└── README.md # This file
| Component | Technology |
|---|---|
| ML Model | RPT-1 - Foundation Model for Tabular Data |
| Frontend | Streamlit |
| Backend | FastAPI |
| Cloud | Azure Container Apps, HuggingFace Spaces |
| Container | Docker |
# Python 3.10+
python --version
# Get TabPFN token from https://tabpfn.com
export TABPFN_ACCESS_TOKEN="your_token_here"# Clone the repo
git clone https://github.com/amitlals/SAP-RPT-1-OSS-App-playground.git
cd SAP-RPT-1-OSS-App-playground
# Install dependencies
pip install -r requirements.txt
# Run any app
streamlit run apps/01-finance-dashboard/app.py
streamlit run apps/02-forecast-showdown/app.py
streamlit run apps/03-predictive-integrity/app.py- Create a new Space at huggingface.co/new-space
- Select Docker as SDK
- Copy the contents of the desired app folder
- Add secrets:
TABPFN_ACCESS_TOKEN- From tabpfn.comSAP_RPT1_TOKEN(optional) - For SAP-RPT-1 Closed API
See deploy/huggingface/README.md for details.
# From repository root
cd deploy/azure
./deploy-azure.ps1See deploy/azure/README.md for details.
All apps use synthetic SAP-like datasets:
| Dataset | Description | SAP Tables Referenced |
|---|---|---|
synthetic_financial_statements.csv |
Balance sheet & income data | BSEG, BKPF, T001 |
synthetic_gl_accounts.csv |
General ledger accounts | SKA1, SKAT |
synthetic_sales_orders.csv |
Sales order history | VBAK, VBAP, LIKP |
| Scenario | SAP Tables | Key Fields |
|---|---|---|
| Job Failure | TBTCO, TBTCP | JOBNAME, STATUS, SDLUNAME |
| Transport Failure | E070, E071 | TRKORR, TRSTATUS, AS4USER |
| Interface Failure | EDIDC, EDIDS | DOCNUM, STATUS, MESTYP |
See Claude Skills
MIT License - See LICENSE for details.

