Category: AI Model Optimization

• ICML 2025 | Fast and Powerful Liger! Transformer Instantly Switches to Linear RNN with only 20M Token Fine-tuning