AI4EOSC – Dashboard (User Platform)

In the Marketplace view. you can find:

• AI Modules Catalog: This is a library of ready-to-run AI/ML models (“modules”), each packaged with its code and (often) pre-trained weights. Modules cover various tasks – e.g. image classification, object detection, NLP tasks, etc. The catalog includes models developed within AI4EOSC as well as models from external sources that have been integrated for one-click deployment. In fact, AI4EOSC has integrated external model repositories (like the DEEP-Hybrid-DataCloud Open Catalog), so you’ll see a wide range of options beyond just in-house models. You can filter modules by ML library (TensorFlow, PyTorch, etc.), task type, data type, and other tags to quickly find what you need. Each module has a detail page describing what it does and what resources it requires.

• Tools Catalog: Besides AI models, the platform offers various auxiliary tools to support the AI workflow. For example, there’s an “AI4OS Development Environment” (essentially a preconfigured JupyterLab or VS Code workspace) for developing or tweaking models. There are also Federated Learning servers (e.g. Flower, NVFLARE) if you want to coordinate distributed training across sites, and a CVAT annotation tool for labeling images. These tools can also be deployed via the same interface, helping manage the end-to-end cycle of AI development (from data labeling to training to inference) on EOSC.

• LLM Catalog: As mentioned above, the dashboard includes a section for Large Language Models as well, where you can choose an LLM to deploy for your own use. This is integrated similarly to modules and tools – you select a model card and deploy your chatbot or API instance.