BWLA: Breaking the Barrier of W1AX Post-Training Quantization for LLMs
For practitioners deploying large language models, BWLA enables extreme compression (1-bit weights + low-bit activations) with minimal accuracy loss, significantly reducing memory and compute costs.
BWLA is the first post-training quantization framework that achieves 1-bit weight quantization with low-bit activations (e.g., 6 bits) for LLMs, overcoming the activation heavy-tail problem. On Qwen3-32B, it achieves a Wikitext2 perplexity of 11.92 (vs. 38 from SOTA), improves zero-shot tasks by >70%, and delivers 3.26× speedup.
Large language models (LLMs) have driven major progress in NLP, yet their substantial memory and compute demands still hinder practical deployment. Binarization can compress weights to 1 bit, fundamentally lowering compute and bandwidth cost. However, existing methods cannot address activation heavy tails and thus must keep activations in high precision, preventing true end-to-end acceleration. To overcome this limitation, we propose BWLA (Binarized Weights and Low-bit Activations), the first post-training quantization framework that preserves high accuracy while achieving 1-bit weight quantization together with low-bit activations (e.g., 6 bits). The Orthogonal-Kronecker Transformation (OKT) learns an orthogonal mapping via EM minimization, converting unimodal weights into symmetric bimodal forms while suppressing activation tails and incoherence. The Proximal SVD Projection (PSP) then performs lightweight low-rank refinement through proximal SVD projection, further enhancing quantizability with minimal overhead. On Qwen3-32B, BWLA reaches a Wikitext2 perplexity of 11.92 under 6-bit activations (vs. 38 from SOTA), improves five zero-shot tasks by more than 70%, and delivers 3.26 times inference speedup, demonstrating strong potential for real-world LLM compression and acceleration.