About this role
Contribute to research and implementation of pre-training and alignment algorithms for ultra-large-scale multilingual models (including Mixture-of-Experts, Instruction Pretraining, SFT, and RLHF), improve explanation and safety of trustworthy large language models, conduct experiments to evaluate model performance, and collaborate using statistics, scripting, and programming languages such as Python/PyTorch.
Skills for this role
Pre-training algorithmsAlignment algorithmsUltra-large-scale multilingual pre-trainingMixture-of-Experts model trainingInstruction PretrainingSFTRLHFModel explainabilityModel safetyStatisticsScriptingPythonPyTorchData structuresAlgorithmsText generationDialogue systemsLarge model training