Unverified50% confidenceFactExact time
GPTQ was proposed by a research team at IST Austria in 2023 and uses an approximate second-order optimization method based on OBS to perform layer-by-layer quantization calibration of weights.
1
Sources
50%
Confidence
Long-term
Relevance
9/10/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGPTQ was proposed in 2022 and improves INT4 quantization quality through layer-wise second-order information calibration.73% similarUnverifiedGPTQ introduced layer-wise calibration datasets in 2022, and AWQ in 2023 reduced precision loss by protecting weight channels corresponding to salient activations.68% similarUnverifiedGenerative Engine Optimization (GEO) was formally proposed around 2023 by researchers from Princeton University and other institutions.64% similarUnverified2022-2023 年间「大模型+工具调用」的可行性被相继验证63% similarUnverifiedMeta's research team proposed the SpinQuant method in 2024, attempting to maintain model capabilities at lower precision.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/888052API
curl https://kongchang.com/api/v1/knowledge/claims/888052MCP
get_claim(id=888052)