张杰
Employers on the program
- 2021阿里巴巴1 talk
Events1
Conferences1
Talks1
Years active2021
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 |
|---|---|---|---|---|---|
| Nov 25, 2021 | AICon北京2021 | AICon 全球人工智能开发与应用大会 (AICon Global AI Development and Application Conference) | speaker | 大规模预训练模型高效训练的架构实践 as 阿里巴巴 最近,阿里巴巴并联合清华大学等共同开发了全球最大规模的中文多模态预训练模型 M6(MultiModality-to-MultiModality Multitask Mega-transformer)。模型参数超1万亿规模,借助 PAI 自研的分布式训练框架 Whale,首次2天内在496…Show less最近,阿里巴巴并联合清华大学等共同开发了全球最大规模的中文多模态预训练模型 M6(MultiModality-to-MultiModality Multitask Mega-transformer)。模型参数超1万亿规模,借助 PAI 自研的分布式训练框架 Whale,首次2天内在496 GPU卡上完成M6模型1亿图文样本的预训练。 相比于非纯文本或者纯图像预训练模型,多模态预训练模型数据挑战更大、模型训练难度更高、下游覆盖的应用场景更广泛。从学术研究的角度来讲,跨模态的理解和生成和人类认知也有着千丝万缕的联系(语义对齐、语言/视觉控制),对其更加深入的… |
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.