← Latest products

We beat Gemini Embedding 2 by training only 16M params (open weights)

Feature Extraction model with 16.4M parameters. Download, explore files and run it on Hugging Face.

Visit huggingface.coexternal site

Reactions (1)

Similar products

by description similarity

  1. so I built sub-second video search

    Hacker News438 pts

  2. Needle

    github.com

    We Distilled Gemini Tool Calling into a 26M Model

    Hacker News776 pts

  3. Marlin-2B

    huggingface.co

    a tiny VLM to extract structured information from videos

    Hacker News7 pts

  4. Inspired by Simon Willison’s pelican-riding-a-bicycle benchmark, I used Claude, Claude Code, and OpenRouter to get SVGs from six models for thirty similar prompts.

    Hacker News16 pts

  5. one C file (no deps.)

    Hacker News7 pts