burakaydinofficial/Quantuzo
Quantuzo: KV Cache Quantization Benchmark Does KV cache quantization in llama.cpp hurt coding ability? Quantuzo measures the impact of KV cache quantization levels on real-world software engineering tasks using SWE-bench. Instead of synthetic benchmarks, models must actually browse repositories, understand code, write patches, and pass test suites. Motivation KV cache quantization (q8_0, q5_0, q4_0, etc.) significantly reduces VRAM usage during inference, making… See the full description on the dataset page: https://huggingface.co/datasets/burakaydinofficial/Quantuzo.
| Author / organization | burakaydinofficial |
|---|---|
| License (as declared) | mit |
| Downloads | 434 (as reported by Hugging Face) |
| Likes | 1 |
| Created | 27 February 2026 |
| Last updated at source | 2 August 2026 |
Popularity, last 2 days
434 now (+2 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.