HF Search Booster

Improves Hugging Face model search precision with variant hover details and size filtering.

HF Search Booster makes Hugging Face's model search far more useful. FEATURES - Smart query matching: hides cards that don't actually match your query (e.g. searching "gemma 4 uncensored" filters out unrelated results). - Variant file sizes: fetches the real file sizes for each model variant (GGUF, SafeTensors, etc.) and shows a min–max range badge on every card. - Hover details: hover a card to see every model file with its exact size. - Size-based filter: dual-range slider + precise number inputs so you can narrow results to models that fit your GPU / disk. - Auto-scan on scroll: sizes populate automatically as you scroll — no manual clicks needed. - Persistent cache: once a model's files are scanned, the result is cached locally for 7 days so refreshes are instant. - Toggle everything: popup settings to disable smart matching or variant sizes without removing the extension. PRIVACY - No tracking, no analytics, no data sent to any third party. - Only network requests: Hugging Face's own public API (`/api/models/{id}/tree/main`), triggered by you browsing or hovering cards. - Settings and cache are stored locally in your browser (chrome.storage). HOW TO USE 1. Install, then visit https://huggingface.co/models. 2. Search as usual — the "Variant Sizes" panel appears in the sidebar. 3. Scroll to auto-scan; hover any card for full file details; use the range slider to filter by size.