Frantar and colleagues: GPTQ post-training weight quantization

What changed

Used approximate second-order information for one-shot post-training weight quantization of large pretrained language models.

Scientific scope

Used approximate second-order information for one-shot post-training weight quantization of large pretrained language models. Year follows the 2022 preprint; conference publication was 2023. Compression and accuracy results depend on the model, calibration data, bit width and kernels.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Year follows the 2022 preprint; conference publication was 2023. Compression and accuracy results depend on the model, calibration data, bit width and kernels.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction