#skip connections
2 related articles
Deep Dives·3 min
DeepSeek V4 Deep Technical Breakdown: Million-Token Context and Extreme Cost Efficiency
Deep analysis of DeepSeek V4's core architecture: Hybrid Compressed Attention, Manifold-Constrained Hyperconnection, and MUON optimizer—how they cut inference costs by 10x and enable million-token context processing.
Read more →
Tutorials
Efficient PyTorch Learning: A Source C…
·3 min
Efficient PyTorch Learning: A Source Code-Driven Methodology
A proven PyTorch learning method: spend 2-3 days on basics, then advance rapidly by reading U-Net and ViT source code line by line. Master PyTorch through source code-driven learning.
Read more →