hf_search_models
Search HuggingFace models using semantic network and fuzzy matching, with filters for task, author, library, and popularity.
Instructions
Search HuggingFace models using navigational semantic network + fuzzy matching.
Combines three layers:
- Network: Bank-position constraints + anchor similarity (if models indexed)
- Fuzzy: String matching on model IDs, tags, descriptions
- Structured: HF API filters for task, author, library, popularity
Args:
query: Natural language search query (e.g. "small code model with tool-calling")
task: HuggingFace task filter (e.g. "text-generation", "text-classification")
author: Filter by model author/org (e.g. "meta-llama", "mistralai")
library: Filter by library (e.g. "transformers", "gguf", "diffusers")
min_likes: Minimum number of likes
min_downloads: Minimum number of downloads
limit: Number of results to return (default 20)
candidate_limit: How many candidates to fetch from HF API (default 500)
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| task | No | ||
| limit | No | ||
| query | Yes | ||
| author | No | ||
| library | No | ||
| min_likes | No | ||
| min_downloads | No | ||
| candidate_limit | No |
Output Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| result | Yes |