QuIDE: Mastering the Quantized Intelligence Trade-off via Active Optimization
Fuente:
arXiv
Salvato in:
| Autore principale: | |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909034043408384 |
|---|---|
| author | Jiang, Xiantao |
| author_facet | Jiang, Xiantao |
| contents | There is currently no unified metric for evaluating the efficiency of quantized neural networks. We propose QuIDE, built around the Intelligence Index I = (C x P)/log_2(T+1), which collapses the compression-accuracy-latency trade-off into a single score. Experiments across six settings -- SimpleCNN (MNIST, CIFAR), ResNet-18 (ImageNet-1K), and Llama-3-8B -- show a task-dependent Pareto Knee. 4-bit quantization is optimal for MNIST and large LLMs, while 8-bit is the sweet spot for complex CNN tasks (ResNet-18 on ImageNet), where 4-bit PTQ collapses accuracy catastrophically. The accuracy-gated variant I' correctly flags these non-viable configurations that the raw I would reward. QuIDE provides a reproducible evaluation protocol and a ready-to-use fitness function for mixed-precision search. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_10959 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | QuIDE: Mastering the Quantized Intelligence Trade-off via Active Optimization Jiang, Xiantao Machine Learning Artificial Intelligence I.2.6; I.5.1 There is currently no unified metric for evaluating the efficiency of quantized neural networks. We propose QuIDE, built around the Intelligence Index I = (C x P)/log_2(T+1), which collapses the compression-accuracy-latency trade-off into a single score. Experiments across six settings -- SimpleCNN (MNIST, CIFAR), ResNet-18 (ImageNet-1K), and Llama-3-8B -- show a task-dependent Pareto Knee. 4-bit quantization is optimal for MNIST and large LLMs, while 8-bit is the sweet spot for complex CNN tasks (ResNet-18 on ImageNet), where 4-bit PTQ collapses accuracy catastrophically. The accuracy-gated variant I' correctly flags these non-viable configurations that the raw I would reward. QuIDE provides a reproducible evaluation protocol and a ready-to-use fitness function for mixed-precision search. |
| title | QuIDE: Mastering the Quantized Intelligence Trade-off via Active Optimization |
| topic | Machine Learning Artificial Intelligence I.2.6; I.5.1 |
| url | https://arxiv.org/abs/2605.10959 |