---
title: "BiSCo-LLM: Lookup-Free Binary Spherical Coding for Extreme Low-Bit LLM Compression"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.08643v1"
permalink: "https://feed.astralor.com/items/itm_48362c9ed9e9bb9f7cf7a716"
published: "2026-07-09T16:17:10.000Z"
score: 77
channel: "tech"
tags: ["arXiv","BiSCo-LLM","Binary Spherical Coding","Low-Bit Quantization","Large Language Models","Model Compression"]
---

# BiSCo-LLM: Lookup-Free Binary Spherical Coding for Extreme Low-Bit LLM Compression

## AI commentary

> This paper offers a creative attempt at extreme low-bit weight compression, using sign streams as the main payload to break away from traditional codebook thinking, with impressive results below 2 bits. It is especially valuable for engineers and researchers working on model deployment and edge AI, and could become a new benchmark for low-bit LLM compression.

## At a glance

This paper presents BiSCo-LLM, a codebook-free binary spherical coding framework for extreme low-bit weight compression of LLMs. Local weight chunks are mapped onto a unit hypersphere and binarized into compact spherical codes, with the main payload being a bit-packed sign stream. A residual BSQ stage encodes reconstruction errors without stored codebooks, and category-wise recovery distillation reduces mismatch between local reconstruction and assembled model behavior. An 8-bit protected-channel path stabilizes sensitive channels. Total storage includes binary codes, neural decoders, protected-channel payloads, LoRA adapters, and metadata. The method significantly outperforms existing quantization and VQ methods at very low bitrates.

## Key points

- Codebook-free binary spherical coding with main payload as sign bit stream.
- Residual BSQ stage provides rate-distortion path without stored codebooks.
- Category-wise recovery distillation reduces mismatch between local reconstruction and assembled model behavior.
- 8-bit protected-channel path as auxiliary stabilization for sensitive channels, counted separately in total budget.

## Full text

Large language models (LLMs) are increasingly constrained by memory capacity, weight bandwidth, and checkpoint storage during deployment. Existing low-bit compression methods mainly follow two directions. Scalar or group-wise quantization is simple and compatible with efficient low-precision kernels, but its representation capacity becomes limited when the target budget approaches 2 bits per weight. Vector-quantized weight compression provides a richer block-level representation, but usually introduces explicit codebooks, index lookup, and additional storage accounting. This paper presents BiSCo-LLM, a codebook-free binary spherical coding framework for extreme low-bit LLM weight compression. The core pipeline is built on three components. First, local weight chunks are mapped onto a unit hypersphere and binarized into compact spherical codes, so that the main payload is a bit-packed sign stream rather than explicit VQ centroids. Second, a residual BSQ stage encodes the reconstruction error left by the base spherical codec, providing an explicit rate-distortion path without stored codebooks. Third, category-wise recovery distillation is performed after replacing each Transformer module category, reducing the mismatch between local weight reconstruction and assembled model behavior. A small 8-bit protected-channel path is used as an auxiliary stabilization mechanism for sensitive channels and is counted separately from the BSQ payload. The reported storage budget includes binary codes, neural decoders, protected-channel payloads, LoRA adapters, and metadata.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_48362c9ed9e9bb9f7cf7a716