Seungyeon Kim

A Little Help Goes a Long Way: Efficient LLM Training by Leveraging Small LMs

Hybrid Architectures for Language Models: Systematic Analysis and Design Insights

Relaxed Recursive Transformers: Effective Parameter Sharing with Layer-wise LoRA

Faster Cascades via Speculative Decoding

A Little Help Goes a Long Way: Efficient LLM Training by Leveraging Small LMs