Unverified50% confidenceFactExact time
The GQA technique was initially proposed by Google Research in 2023 and was subsequently widely adopted by mainstream models such as Llama 2 and Mistral.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/5/2026
First Seen
Valid until: 12/4/2026
Sources
Related Entities
Related Claims
Unverified可溯源生成（attributable generation）是 NLP 领域的研究方向，最早由 Google Research 等团队在 2022-2023 年间系统化提出69% similarUnverifiedMLP Mixer 最初由 Google Research 于 2021 年提出，用于图像识别任务，论文发表于 NeurIPS 202168% similarUnverified智谱GLM系列模型自2022年起持续迭代，与Meta的LLaMA、Mistral等西方开源模型竞争66% similarUnverifiedGLM 5.2引入了多令牌预测（MTP）技术，该技术由Meta在2024年研究中提出，可通过投机解码提升推理速度66% similarVerifiedThe ReAct framework was proposed by Google Research in 2022 and published at ICLR 2023 by Shunyu Yao et al.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/860212API
curl https://kongchang.com/api/v1/knowledge/claims/860212MCP
get_claim(id=860212)