Large Language Model 연구 및 개발 Internship

1 month ago

Seoul, Korea, SouthIntern

Responsibilities

Research and develop technologies related to large language models.
Build pipelines for collecting and preprocessing training data.
Measure benchmark performance for in-context learning and fine-tuning.
Develop prompt-engineering methods for in-context learning.
Optimize inference engines.
Research and develop multi-task instruction tuning and performance-improvement methods.
Research and develop toxic classifiers and safety methodologies for large language models.
Develop methods to improve reinforcement-learning performance and distributed training for language models.
Research and develop model-compression methods for large language models.

Be an undergraduate student expected to graduate, a recent undergraduate graduate, or a current master's or doctoral student.
Have required experience developing deep learning or machine learning algorithms.
Have required proficiency with Linux, Python, PyTorch, and HuggingFace language models such as BERT, T5, or GPT.
Preferred: experience developing language-model pretraining.
Preferred: experience developing DeepSpeed or FastTransformer.
Preferred: high leaderboard placement, competition awards, or open-source activity.
Preferred: experience across the ML production cycle, including data collection, model development, model testing, and serving.
Preferred: experience developing with large-scale data-processing frameworks such as Apache Beam, Amazon EMR, or GCP Dataproc.
Preferred: publications related to language models.
Preferred: a major in computer engineering, computer science, artificial intelligence, or machine learning.

201-500 employees