张跃伟
Employers on the program
- 2024阿里巴巴1 talk
Events1
Conferences1
Talks1
Years active2024
Appearances by year
- AICon 全球人工智能开发与应用大会 (AICon Global AI Development and Application Conference)
Talks
| Date | Event | Conference | Role | Title | Links |
|---|---|---|---|---|---|
| Dec 13, 2024 | AICon 全球人工智能开发与应用大会 ｜北京 | AICon 全球人工智能开发与应用大会 (AICon Global AI Development and Application Conference) | speaker | 大模型超低比特无损量化技术实践 as 阿里巴巴 模型量化通过降低权重精度减少计算量与内存占用，加速推理过程并降低能耗，对于提升大模型在资源受限设备上的部署效率至关重要。以往 PTQ 量化方法忽略不同层权重对量化精度的敏感性，且在超低比特量化下（2-3位）精度损失严重，论文针对这两个问题，提出 AMLQ( Adaptive Mixed precision and…Show less模型量化通过降低权重精度减少计算量与内存占用，加速推理过程并降低能耗，对于提升大模型在资源受限设备上的部署效率至关重要。以往 PTQ 量化方法忽略不同层权重对量化精度的敏感性，且在超低比特量化下（2-3位）精度损失严重，论文针对这两个问题，提出 AMLQ( Adaptive Mixed precision and Low-rank Quantization error reconstruction) 量化方法，极大减少模型量化的精度损失，并提高推理速度。量化方案详情可参考：https://openreview.net/pdf?id=T5pGDydMkS … |
Frequent co-speakers
No one else spoke at two or more of the same events.
Speakers with the same name are merged into one entry, so this page may combine more than one person. Affiliations are shown as each conference listed them.