Variable Bit-width Quantization: Learning Per-Group Precision for "Bigger-but-Smaller" Language Models

• URL: http://arxiv.org/abs/2607.02893v1
• Date: Fri, 03 Jul 2026 02:42:04 GMT
• Title: Variable Bit-width Quantization: Learning Per-Group Precision for "Bigger-but-Smaller" Language Models
• Abstract summary: We introduce Variable Bit-width Quantization (VBQ)<n>Training-time method in which each contiguous group of 64 weights learns its own resolution from 1,2,4,8 bits.<n>As quality-per-byte, VBQ is 3.9-8.4x more efficient than FP16.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Low-bit quantization shrinks language models but treats precision as a single global hyper-parameter: every weight uses the same bit-width. We introduce Variable Bit-width Quantization (VBQ), a training-time method in which each contiguous group of 64 weights learns its own resolution from {1,2,4,8} bits via a Gumbel-Softmax relaxation, trained jointly by an alternating optimization that gives the precision logits a clean, task-aligned signal. VBQ discovers a consistent, strongly heterogeneous allocation within individual projection types, not merely across layers, impossible to express with per-layer methods: 69% of groups collapse to 1 bit, the LM head averages 1.09 bits, while the first MLP block keeps ~2.5 bits. This pattern is stable enough to freeze into a fixed recipe and reuse without further search. The recipe yields a "bigger-but-smaller" regime: a 131M model at 1.82 mean bits reaches perplexity 4.2 on TinyStories, beating a 55M FP16 model (PPL 4.4) at 3.8x less storage, and lets a 1.46B model on FineWeb-Edu match a 593M FP16 control at ~3.7x less storage with 2.5x more parameters. As quality-per-byte, VBQ is 3.9-8.4x more efficient than FP16. The recipe maps directly to packed low-bit storage, so it also accelerates inference: with custom fused dequantize-and-multiply kernels, memory-bandwidth-bound autoregressive decode is faster at equal output, and the speedup grows with scale (parity at 131M, 1.9x at 1.0B, 4.7x at 9B on Apple silicon). A distributional analysis (KL divergence and argmax-flip rate) reveals a striking mechanism: deeper layers progressively self-heal the quantization error injected by early layers. The win is a from-scratch, train-time phenomenon; scaling the search economically beyond 1.5B parameters remains open. VBQ reframes precision as a learnable, non-uniform resource and shows that spending a fixed bit budget unevenly beats spending it uniformly.

Related papers

• Scaling Post-Training Ternarisation to Qwen3-8B Capability Retention, Reproduction, Lossless Packing, and Packed Execution [0.0]
  We study a scale-up of an aggressive post-training conversion pipeline from Qwen3-4B to Qwen3-8B.<n>The conversion uses KOTMS rotation, E2M-ATQ adaptive ternarisation, and GPTQ-style error compensation in a weight-only A16 configuration.
  arXiv  Detail & Related papers  (2026-09-08T06:21:44Z)
• XFP: Quality-Targeted Adaptive Codebook Quantization with Sparse Outlier Separation for LLM Inference [0.0]
  XFP determines codebook size, outlier budget, and packing per layer automatically.<n>XFP reaches 138 tok/s single-stream decode on workstation hardware.<n>For models that do not fit in the target memory envelope, we present the H-Process.
  arXiv  Detail & Related papers  (2026-05-14T13:52:31Z)
• StateSMix: Online Lossless Compression via Mamba State Space Models and Sparse N-gram Context Mixing [0.0]
  StateSMix is an online-trained Mamba-style State Space Model (SSM) with sparse n-gram context mixing and arithmetic coding.<n>The SSM provides a continuously-updated probability estimate over BPE tokens, while nine sparse n-gram hash tables add exact local and long-range pattern memorisation.<n>On the standard enwik8 benchmark, StateSMix achieves 2.123 bpb on 1 MB, 2.149 bpb on 3 MB, and 2.162 bpb on 10 MB, beating xz -9e (LZMA2) by 8.7%, 5.4%, and
  arXiv  Detail & Related papers  (2026-04-05T19:28:11Z)
• Scaling DoRA: High-Rank Adaptation via Factored Norms and Fused Kernels [83.99688944263843]
  Weight-De Low-Rank Adaptation (DoRA) extends LoRA by decoupling weight magnitude from direction.<n>At d_in = 8192 and rank r = 384, a single module's norm requires about 512 MB of transient working memory in bf16.<n>A factored norm decomposes the squared norm into base, cross, and Gram terms computable through O(d_out r + r2) intermediates, eliminating the dense product.
  arXiv  Detail & Related papers  (2026-03-23T17:57:24Z)
• BPDQ: Bit-Plane Decomposition Quantization on a Variable Grid for Large Language Models [56.504879072674015]
  We propose Bit-Plane Decomposition Quantization (BPDQ), which constructs a variable quantization grid via bit-planes and scalar coefficients.<n>BPDQ enables serving Qwen2.5-72B on a single GTX 3090 with 83.85% GSM8K accuracy (vs. 90.83% at 16-bit)
  arXiv  Detail & Related papers  (2026-02-04T02:54:37Z)
• Binary Quantization For LLMs Through Dynamic Grouping [13.578307208515819]
  Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of Natural Language Processing (NLP) tasks.<n> Binary quantization, which compresses model weights from 16-bit Brain Float to 1-bit representations in -1, 1, offers significant reductions in storage and inference costs.<n>We propose a novel optimization objective tailored for binary quantization, along with three algorithms designed to realize it effectively.
  arXiv  Detail & Related papers  (2025-09-03T06:36:21Z)
• ParetoQ: Improving Scaling Laws in Extremely Low-bit LLM Quantization [73.60493264901359]
  We present a unified framework for rigorous comparisons across 1-bit, 1.58-bit, 2-bit, 3-bit, and 4-bit quantization settings.<n>We show that ternary, 2-bit, and 3-bit quantization maintains comparable performance in the size-accuracy trade-off.<n>Considering hardware constraints, 2-bit quantization offers promising potential for memory reduction and speedup.
  arXiv  Detail & Related papers  (2025-02-04T18:59:26Z)
• SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models [61.474101404805545]
  Diffusion models can generate high-quality images, but as they scale, rising memory demands and higher latency pose deployment challenges.<n>We propose SVDQuant, a new 4-bit quantization paradigm to overcome this limitation.<n>We reduce the memory usage for the 12B FLUX.1 models by 3.5$times$, achieving 3.0$times$ speedup over the 4-bit weight-only quantization (W4A16) baseline.
  arXiv  Detail & Related papers  (2024-11-07T18:59:58Z)
• The case for 4-bit precision: k-bit Inference Scaling Laws [75.4335600212427]
  Quantization methods reduce the number of bits required to represent each parameter in a model. The final model size depends on both the number of parameters of the original model and the rate of compression. We run more than 35,000 zero-shot experiments with 16-bit inputs and k-bit parameters to examine which quantization methods improve scaling for 3 to 8-bit precision.
  arXiv  Detail & Related papers  (2022-12-19T18:48:33Z)
• 8-bit Optimizers via Block-wise Quantization [57.25800395197516]
  Statefuls maintain statistics over time, e.g., the exponentially smoothed sum (SGD with momentum) or squared sum (Adam) of past values. This state can be used to accelerate optimization compared to plain gradient descent but uses memory that might otherwise be allocated to model parameters. In this paper, we develop first gradients that use 8-bit statistics while maintaining the performance levels of using 32-bit gradient states.
  arXiv  Detail & Related papers  (2021-10-06T15:43:20Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.