AI.info

Soeun Kim

Explore Soeun Kim on AI.info.

• Preserve-Then-Quantize: Balancing Rank Budgets for Quantization Error Reconstruction in LLMs