The paper introduces a method for allocating bit-widths to individual layers in Gemma-3-1B to maximize performance (latency reduction) while staying within a specified quality-degradation budget. Using a layer sensitivity profile from SA-PTQ and TensorRT-LLM, the authors evaluate 13 W8A8 variants on an RTX 5090, finding that FFN 5+5 with lm_head yields an 11.0% latency reduction with negligible quality loss. They also discuss trade-offs for Attention layers and propose further optimizations such as fused INT8 attention kernels and FP8 usage.
arXiv:2607. 14618v1 Announce Type: new Abstract: CPUs are the most universal target for on-device LLM inference, but existing low-bit quantization methods offer either coarse operating points or fine-grained mixed precision that is difficult to execute efficiently on CPUs.
arXiv:2605. 06675v2 Announce Type: replace Abstract: Large language models cache all previously computed key-value (KV) pairs during generation, and this KV cache grows linearly with sequence length, making it a primary memory bottleneck for serving.
arXiv:2606. 24033v1 Announce Type: new Abstract: Existing low-bit KV-cache quantizers often treat each cached key as a flat vector.
arXiv:2607. 23047v1 Announce Type: cross Abstract: Mixed-precision quantization improves the accuracy of post-training quantization by allocating higher bitwidths to sensitive layers, but existing methods solve the allocation for a single fixed memory budget.
arXiv:2606. 00079v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) large language models reduce per-token computation through sparse expert activation, but their deployment remains memory-intensive because all expert weights must be kept resident in memory.
arXiv:2607. 02893v1 Announce Type: new Abstract: Low-bit quantization shrinks language models but treats precision as a single global hyper-parameter: every weight uses the same bit-width.
arXiv:2410. 13056v4 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have demonstrated remarkable success across a wide range of language tasks, but their deployment on edge devices remains challenging due to the substantial memory requirements imposed by their large parameter sizes.
arXiv:2607. 28699v1 Announce Type: cross Abstract: KV-cache quantization is validated today by offline benchmark averages; a deployed system cannot tell whether compression is damaging the request it is serving right now.
HBQ: Hierarchical Scaling Block Quantization with Hardware‑Efficiency‑Aware Design for Accurate LLM Inference proposes a new block‑quantization scheme that uses large blocks and low‑overhead significand scaling to balance hardware efficiency and accuracy. The authors demonstrate that larger blocks improve efficiency by amortizing dequantization and accumulation costs, while their SIG scaling compensates for the resulting accuracy loss. Experiments on a 28 nm ASIC accelerator show that HBQ achieves up to 4.6× higher area/energy efficiency than state‑of‑the‑art weight‑only quantization, with 1.5–3.0× speedup and 1.6–3.3× system energy reduction over existing BQ methods.
arXiv:2608. 08081v1 Announce Type: cross Abstract: Large mixture-of-experts (MoE) language models with 26--120 billion parameters exceed the memory capacity of consumer devices through three simultaneous pressures: resident weight matrices, key-value (KV) cache state that grows linearly with context, and dozens of expert sublayers that must be paged on demand.
arXiv:2608.23816v1 Announce Type: new Abstract: Quantized fine-tuning (QLoRA) saves memory but not time. It dequantizes every 4-bit weight on the fly, so it trains more slowly than fp16 LoRA. We pres...
The paper introduces a composite metric for quantizing small language models that balances information retention and throughput gains, using a normalized SQNR-based coefficient and roofline-based latency analysis. Profiling Gemma 3 1B shows that Feed‑Forward Network blocks and the embedding matrix are prime candidates for acceleration, with the metric enabling tuning of speed‑quality trade‑offs without actual execution. The authors demonstrate that their estimates predict accelerated speedup within about 4% error and allocate resources more effectively than evolutionary search or Shapley‑value methods.