小度云平台部_大模型算法实习生(J38341)
Core
Designing, pre-training, training, and validating multimodal large models for Baidu's XiaoDu platform, covering cross-modal generation, retrieval, and dialogue applications.
Role type
Intern, Multimodal Large Model Algorithm Engineer
Builds
Multimodal models for XiaoDu business scenarios (visual, audio, NLP)
Domain
Artificial Intelligence, Multimodal Learning
Deliverable
production ML models
Required skills
Multimodal large model training, PyTorch/TensorFlow, Megatron/Deepspeed, cross-modal generation, cross-modal retrieval, conversational AI, large-scale parallel training
Preferred skills
Publications in ICLR, CVPR, ICCV, ECCV, ACL, NeurIPS, ICML, experience with ChatGPT/GPT-4/StableDiffusion
Technologies
PyTorch, TensorFlow, Megatron, Deepspeed
Responsibilities
Designing multimodal model architectures, conducting pre-training tasks, performing model training, validating inference, tracking research frontiers, productizing SOTA models
Seniority
Intern
