Datasetpublic

pbhappliedsystems/quant_eval_efficiency_and_footprint

quant_eval — Efficiency and footprint One row per published run: stored weight artifact bytes before and after quantization, compression ratio, observed evaluation wall-time ratio with an explicit direction label, the accelerator used on each lane, and token throughput. Part of the quant_eval public corpus: a per-case behavioral evaluation of full-weight and quantized large language models across eight agent-relevant task families, with paired statistical testing. Cite this… See the full description on the dataset page: https://huggingface.co/datasets/pbhappliedsystems/quant_eval_efficiency_and_footprint.

sourceHugging Facecc-by-4.0updated 8d agoView on Hugging Face

0likes65downloads

rootCITATION.cff

CITATION.cff26 linesDownload Raw Back to root

1cff-version: 1.2.02message: If you use this dataset, please cite it.3type: dataset4title: >-5  quant_eval Efficiency and footprint6authors:7  - family-names: Hill8    given-names: Patrick9    affiliation: PBH Applied Systems, LLC10license: CC-BY-4.011version: 1.0.112year: 202613doi: 10.5281/zenodo.2201072314identifiers:15  - type: doi16    value: 10.5281/zenodo.2201072317    description: Concept DOI, always resolves to the latest version18  - type: doi19    value: 10.5281/zenodo.2280612120    description: Version DOI, frozen to this exact deposit21  - type: doi22    value: 10.5281/zenodo.2201072423    description: >-24      Version DOI of 1.0.0, superseded. Its runtime figures omit the accelerator25      used on each lane of a pair.26