Spaces:
Sleeping
Sleeping
| title: ConceptVector | |
| emoji: 🧠 | |
| colorFrom: blue | |
| colorTo: indigo | |
| sdk: docker | |
| app_port: 7860 | |
| pinned: false | |
| license: mit | |
| # ConceptVector - Conceptual Understanding Analytics | |
| This Space hosts the FastAPI backend for ConceptVector, an advanced AI system that evaluates conceptual understanding through multi-dimensional explanation analysis. | |
| ## Features | |
| - Multi-signal understanding inference | |
| - Logical consistency checking (NLI models) | |
| - Concept dependency coverage analysis | |
| - Explanation stability testing | |
| - Assumption completeness detection | |
| ## API Endpoints | |
| - `POST /analyze` - Analyze user explanation | |
| - `GET /concepts` - Get sample concepts | |
| - `GET /health` - Health check | |
| - `GET /docs` - Interactive API documentation | |
| ## Environment Variables | |
| Required: | |
| - `HUGGINGFACE_API_KEY` - Your Hugging Face API token for LLM access | |
| Set this in Space Settings → Repository secrets | |
| ## Tech Stack | |
| - FastAPI | |
| - PyTorch + Transformers | |
| - Sentence Transformers | |
| - Microsoft DeBERTa (NLI) | |
| - Mistral-7B-Instruct (LLM) | |