metadata
license: mit
base_model:
- HuggingFaceTB/SmolLM-1.7B
tags:
- quantization
- smollm
- custom-kernel
SmolLM-1.7B (SpectralQ 6-bit) This model contains experimental 6-bit frequency-domain weights quantized using the Transformer Spectral Scaling Law. It utilizes 256-width DCT macroblocks and a 30-bit chunk-major packed layout.
⚠️ IMPORTANT: HOW TO USE ⚠️ This model cannot be loaded with standard Transformers or vLLM. It requires the custom SpectralQ inference engine, which includes the C++ SRAM bit-unpacking and IDCT projection kernels.
To run this model, please visit the GitHub repository: [👉 SpectralQ 👈]