Category: Model Efficiency

• Attention Is All You Need Author Returns: Can a 99% Sparse Transformer Be Even Faster?
• DeepSeek, GPT-5's Fast-Slow Thinking Switching Gets a Smarter, Multimodal Version