Florence2-Sharp

Home

Build, query, and extend the graph + search + AI platform.

Roll out Curiosity to Docker, Kubernetes, or your cloud of choice.

Ship domain-specific AI apps with custom UIs, endpoints, and connectors.

Model your data as typed nodes and edges with stable keys.

Query with text, vector, and hybrid retrieval — permission-aware.

Ground LLMs, embeddings, and agents in your own graph.

Compose UIs in C# with our front-end component library.

Transpile C# to JavaScript and run it in the browser.

Test, ingest, sync, and promote workspaces from any shell.

Browse every open-source project Curiosity maintains.

Support

Core Concepts

What Florence-2 is, the family of tasks it supports, and how the C# wrapper maps onto them.

A unified vision-language model from Microsoft — one checkpoint that handles captioning, OCR, and detection.

The full TaskTypes enum: what each task does, which FlorenceResults field to read, and when to use it.