Instructions to use mohsenvand/savant-4b-preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers.js
How to use mohsenvand/savant-4b-preview with Transformers.js:
// npm i @huggingface/transformers import { pipeline } from '@huggingface/transformers'; // Allocate pipeline const pipe = await pipeline('text-generation', 'mohsenvand/savant-4b-preview');
Savant-4B
A small language model that computes exactly.
Savant writes a small program for the question in front of it, then runs that program inside its own forward passes, one checked step at a time. No tools. No interpreter. No internet.
It runs entirely in your browser: try it.
Research preview: not a general assistant, not safety-tuned. Paper forthcoming.
Files. ONNX, 4-bit weights, for WebGPU with transformers.js. License. Proprietary (see LICENSE). A modified derivative of Qwen3-4B, Apache 2.0 (see NOTICE).
- Downloads last month
- 591