Publish StockForge optimizer configuration

README.md

@@ -0,0 +1,43 @@
+---
+license: mit
+tags:
+  - inventory-optimization
+  - predict-then-optimize
+  - operations-research
+  - chronos-2
+library_name: stockforge
+---
+
+# StockForge Optimizer Config
+
+Configuration manifest for the StockForge retail decision intelligence engine.
+
+## Forecast Model
+
+- **Base model:** [amazon/chronos-2](https://huggingface.co/amazon/chronos-2)
+- **Quantile levels:** P10, P50, P90
+- **Horizon:** 14 days
+
+## Optimizers
+
+| Optimizer | Description |
+|-----------|-------------|
+| `deterministic_inventory_milp` | Point forecast (P50) → reorder/safety stock MILP |
+| `stochastic_quantile_milp` | Expected cost over quantile scenarios |
+| `robust_quantile_milp` | Worst-case cost minimization |
+
+## Solvers
+
+Pyomo + HiGHS, scipy.stats fallback
+
+## Simulation
+
+SimPy discrete-event inventory validation
+
+## No Training Required
+
+This repository contains optimizer configuration and benchmark metadata only. Forecasting uses the pretrained Chronos-2 foundation model.
+
+## License
+
+MIT

config.json

@@ -0,0 +1,29 @@
+{
+  "model_type": "optimizer_config",
+  "version": "1.0.0",
+  "product": "StockForge",
+  "forecast_model": "amazon/chronos-2",
+  "forecast_model_url": "https://huggingface.co/amazon/chronos-2",
+  "source_dataset": "t4tiana/store-sales-time-series-forecasting",
+  "supplement_dataset": "Jacoblian/RetailOpt-190",
+  "no_training_required": true,
+  "optimizers": [
+    "deterministic_inventory_milp",
+    "stochastic_quantile_milp",
+    "robust_quantile_milp"
+  ],
+  "solvers": [
+    "Pyomo",
+    "HiGHS",
+    "scipy.stats"
+  ],
+  "simulation": "SimPy",
+  "quantile_levels": [
+    0.1,
+    0.5,
+    0.9
+  ],
+  "decision_horizon_days": 14,
+  "best_decision_policy": "deterministic_p50",
+  "forecast_decoupling_rate": 0.6
+}