attention head AI jobs Alerts Saved
Jobs › Moonshot AI

大语言模型架构研究员(预训练方向)

Moonshot AI · 北京市
Apply on Moonshot AI's site
Tailor my resume
Posted
Feb 5, 2025
Last verified
19 min ago
Employment
Full-time
Level
Mid
Function
Research
AI roleCUDA

Open 21 months. Roles open this long are often filled, or kept open all year. Check the date on the employer's site before you apply.

职位描述:
1. 深度参与LLM的模型结构,包括MLP以及attention 模块改进,以及训练方法的设计,包括模型并行策略优化,优化算法改进,量化算法改进,目标损失函数改进等方向,目标是同时在模型训练以及大概规模推理效率上取得sota级别的领先水平
2. 深入调研如何设计科学的实验,利用scaling law验证模型设计,获取从头开始,端到端的大模型训练经验。
3. 良好的团队氛围,扁平化的沟通和管理,丰富的算力和计算资源。

职位要求:
1. 本科及以上学历,计算机、人工智能、自动化、数学专业优先
2. 扎实的编程基础,对数据结构、算法设计基础有深度了解,熟练使用pytorch或其他深度学习框架。
3. 有深度学习训练丰富实操以及参数调优经验。
4. 有LLM模型训练经验者优先
5. 有高性能计算经验者优先,有triton,cuda 等gpu kernel 编码经验优先
6. 有ACM/IOI/NOI/Top Coder等算法竞赛获奖经历优先
7. 在顶级学术会议上发表论文者优先,在机器学习、人工智能领域内获得过优异成绩者优先
8 责任心强,积极主动,有良好的沟通和合作能力

Apply on Moonshot AI's site

Attention Head lists this role as published on Moonshot AI's own job board. Applications go to the employer.

Not this one? Get roles like it weekly

A job alert: the week's best new AI roles from employers' own job boards, every Monday. Unsubscribe in one click.

Similar roles at other employers

All mid-level Research roles in China →

More roles at Moonshot AI

All roles at Moonshot AI →