Update README.md
Browse files
README.md
CHANGED
|
@@ -1,10 +1,56 @@
|
|
| 1 |
---
|
| 2 |
tags:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
- model_hub_mixin
|
| 4 |
- pytorch_model_hub_mixin
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
---
|
| 6 |
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
+
- icos
|
| 4 |
+
- edge-ai
|
| 5 |
+
- energy-forecasting
|
| 6 |
+
- anomaly-detection
|
| 7 |
- model_hub_mixin
|
| 8 |
- pytorch_model_hub_mixin
|
| 9 |
+
- bento_model
|
| 10 |
+
- ai-coordination
|
| 11 |
+
- metrics-prediction
|
| 12 |
+
- cpu-utilization
|
| 13 |
+
- climate-tech
|
| 14 |
---
|
| 15 |
|
| 16 |
+
# ICOS-AI/icos_models
|
| 17 |
+
|
| 18 |
+
This repository serves as the **central model registry for the ICOS Intelligence Layer**, supporting the deployment and reuse of AI models developed across the ICOS ecosystem. These models power key system functionalities such as CPU utilization forecasting, anomaly detection, energy efficiency monitoring, and intelligent scheduling across edge-cloud nodes.
|
| 19 |
+
|
| 20 |
+
## 🔍 What’s Inside
|
| 21 |
+
|
| 22 |
+
The repository contains production-ready AI models developed using PyTorch, XGBoost, ARIMA, and other libraries, and prepared with **BentoML** for reproducible deployment. While active models operate within the live ICOS Intelligence Layer, this repository provides **cold storage** for historical and versioned models.
|
| 23 |
+
|
| 24 |
+
## Features
|
| 25 |
+
|
| 26 |
+
- Version-controlled AI models with BentoML tagging
|
| 27 |
+
- Compatible with `PytorchModelHubMixin` and Hugging Face CLI
|
| 28 |
+
- Structured model cards and metadata per ICOS standards
|
| 29 |
+
- Plug-and-play ready for integration with ICOS CLI and Export Metrics API
|
| 30 |
+
- Reusable across ICOS nodes for multivariate prediction and intelligent control
|
| 31 |
+
|
| 32 |
+
## Model Use Cases
|
| 33 |
+
|
| 34 |
+
- Forecasting system-level metrics (e.g., CPU, RAM, power consumption)
|
| 35 |
+
- Detecting anomalies in robotic and infrastructure data
|
| 36 |
+
- Supporting edge AI coordination and telemetry processing
|
| 37 |
+
- Training and evaluation workflows across heterogeneous environments
|
| 38 |
+
|
| 39 |
+
## Integration Details
|
| 40 |
+
|
| 41 |
+
- **Code**: [Coming Soon]
|
| 42 |
+
- **Paper/Docs**: Please refer to Deliverable [D4.3 – ICOS Dataset and AI Models Marketplace (M36)](https://icos-ai.eu)
|
| 43 |
+
- **Training & Evaluation Framework**: PyTorch, BentoML, XGBoost, Statsmodels, ICOS Export Metrics API
|
| 44 |
+
|
| 45 |
+
## Contribution Guidelines
|
| 46 |
+
|
| 47 |
+
Contributors must follow ICOS standards for:
|
| 48 |
+
- Naming conventions
|
| 49 |
+
- Metadata fields
|
| 50 |
+
- Documentation cards
|
| 51 |
+
- Token-based authentication (via Hugging Face)
|
| 52 |
+
|
| 53 |
+
For onboarding and access: [See Section 4 of D4.3]
|
| 54 |
+
|
| 55 |
+
## Access & Licensing
|
| 56 |
+
|