Spaces:
Sleeping
Sleeping
metadata
title: Questro Recommender Model API
emoji: 🎬
colorFrom: indigo
colorTo: purple
sdk: docker
app_port: 7749
pinned: false
Questro Recommender Model API (CL-EPIDTN, improved_8)
FastAPI service that scores and re-ranks movie/game candidates for the Questro
RAG pipeline. It exposes /recommend, /recommend/rerank, /catalog/add,
/genres, and /health on port 7749.
The full catalog (189,753 items) is baked into the artifacts, so the model
never hot-adds items at runtime. The ~1.2 GB artifacts are not stored in
this Space — the Dockerfile downloads them from the public model repo
zeyadgamal00/CL-EPIDTN at
build time (MODEL_REPO env). See
RECOMMENDER_API_IMPROVED8_DOCS.md for the
API reference.