CrowdCC is a benchmark that jointly evaluates corruption robustness and deployment quantization for crowd‑counting models on edge devices. It introduces 23 deployment‑motivated corruptions at three severity levels, tests 26 models across four datasets, and measures performance in FP32, FP16, and INT8 precision on A100 TensorRT engines and Jetson Orin NX hardware. The study finds that capture‑chain corruptions cause the greatest degradation, INT8 quantization alters robustness unpredictably, and that model selection must consider both corruption robustness and deployment cost together.
arXiv:2609.39223v2 Announce Type: new Abstract: Large language model (LLM) inference is increasingly moving toward lower precision to realize the throughput of hardware accelerators, but aggressive p...
arXiv:2606. 09864v1 Announce Type: cross Abstract: Key-value (KV) cache quantization is widely used to reduce Large Language Model (LLM) inference memory, yet existing evaluations solely focus on measuring perplexity and accuracy without assessing the safety impact.
The paper introduces RAMP, a method for robust adaptive mixed‑precision quantization of vision models on edge CPUs. It evaluates 13 sensitivity metrics across four neural networks, finding that Jensen‑Shannon Divergence consistently identifies layers that can be safely quantized. Using K‑Means clustering on these metrics, RAMP achieves near‑lossless accuracy with an average 1.81× speed‑up, while cautioning against excluding low‑speed‑up layers that can fragment the computational graph.
The paper introduces Sparsity-Adaptive Sharpness-Aware Minimization (SA‑SAM), a method that adjusts the perturbation radius in sharpness-aware training to remain consistent as model sparsity increases. It also evaluates a Magnitude‑Weighted Hessian (MWH) importance metric derived from second‑order analysis. Experiments on CIFAR‑10‑C, CIFAR‑100‑C, and ImageNet‑100‑C show that SA‑SAM improves corruption robustness at 80–90% sparsity while maintaining clean accuracy, and the study reports inference throughput at deployment‑relevant sparsity levels.
arXiv:2604. 10496v2 Announce Type: replace Abstract: Outliers have emerged as a fundamental bottleneck in preserving accuracy for low-precision large models, particularly within Mixture-of-Experts (MoE) architectures that are increasingly central to large-scale language modeling.
arXiv:2604. 18245v3 Announce Type: replace Abstract: Large language models operate in protocols containing multiple calls, yet added calls are usually evaluated only by their net effect.
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. 20128v1 Announce Type: cross Abstract: Benchmarks for LLM-generated GPU kernels (KernelBench, TritonBench, GEAK) score correctness through fixed-shape, small-sample allclose-style checks.
arXiv:2610.14745v1 Announce Type: new Abstract: Extreme low-bit weight-only quantization is an effective approach to reducing the deployment costs of large language models. Below 3 bits, scalar quant...
The paper introduces MuViS-C, a multi‑domain benchmark that evaluates the robustness of learning‑based virtual sensing models against ten common sensor failure modes, ranging from subtle drifts to catastrophic dropouts. It assesses models using average error, relative degradation, and worst‑case fragility across nine datasets from six domains, comparing six architectures (gradient‑boosted trees, convolution, recurrence, attention, and MLP‑mixing). The study finds that all models degrade under corruption, gradient‑boosted trees are most robust, and targeted robustification can improve attention models at the cost of nominal performance.
Deploying deep learning models on edge CPUs is bottlenecked by computational and memory constraints. Mixed-precision quantization promises to reduce inference latency while preserving accuracy. Howeve...
The paper investigates where post‑training quantization (PTQ) harms large language models (LLMs) and how to best allocate a limited precision budget. By causally raising each layer to 8‑bit precision across nine open‑weight models, the authors find that quantization damage is diffuse rather than concentrated in specific task circuits or weight statistics, and that globally refining quantization granularity outperforms selectively protecting the most recoverable layers. They also observe that the residual accuracy loss is budget‑limited and that peak recovery locations correlate with architecture within families but not across families.