QAM-W: Joint 2D Codebook Quantization for LLM Weights via Hadamard Rotation and Activation-Aware Scaling
For LLM practitioners needing efficient weight compression, QAM-W provides a high-quality 5-6 bpw quantization that preserves perplexity, though it is outperformed at strict 4 bpw by QTIP.
QAM-W introduces a joint 2D codebook quantization method for LLM weights that recovers pairwise coordinate structure, achieving at ~5.5 bpw within ±0.4% of BF16 perplexity across five LLMs, matching SmoothQuant W8A8 quality with 32% fewer weight bits.
Scalar post-training quantizers discard pairwise coordinate structure within weight rows. We introduce QAM-W (Quadrature Amplitude Modulation for Weights), a codec that recovers this structure: each row is L2-normalized, block-Hadamard rotated, paired into 2D coordinates, and quantized against a single Lloyd-Max codebook trained on the unit circular Gaussian, with activation-aware per-channel scaling. In a cross-model study spanning five LLMs from four families (1.1B--13B parameters) and eight quantized configurations, the activation-aware variant at $\approx 5.5$ bpw stays within $\pm 0.4\%$ of BF16 WikiText-2 perplexity on every model, matching the SmoothQuant W8A8 quality envelope at $32\%$ fewer weight bits. Joint 2D coding outperforms polar (amplitude $\times$ phase) coding by 2--15~pp $Δ$PPL at equal bitrate, and paired KL against BF16 tracks $Δ$PPL\% at Spearman $ρ= 0.99$ across 37 (method, model) rows, consistent with a monotone composite bound from codec distortion to KL divergence. A 3.5~bpw variant is competitive on quantization-tolerant architectures. At strict 4~bpw, the rotated-codebook frontier method QTIP outperforms QAM-W; the contribution is the quality-preserving 5--6~bpw band.