#Transformer

2026

• 09-25 EN

  FACTS and CoRS: Which Curvature Should Low-Rank Compression Preserve?

• 09-25 中

  FACTS 与 CoRS 阅读笔记：低秩压缩究竟该保留哪一种曲率？

• 09-24 EN

  Deep Delta Learning: Editing the Residual Stream, and Accounting for the Cost

• 09-24 中

  Deep Delta Learning 阅读笔记：怎样改写残差流，以及这次改写要付出什么代价

• 09-20 EN

  mHC: Stable Residual Mixing, Its Guarantees, and Its Costs

• 09-20 中

  mHC 阅读笔记：残差混合如何稳定，保证又止于哪里

• 09-19 EN

  Attention Residuals: Learning What to Retrieve Across Network Depth

• 09-19 中

  Attention Residuals 阅读笔记：让每一层选择自己需要的历史表示

• 08-03 EN

  BLADE: Boundary-Expanded and Layer-Adaptive Dynamic Exit for Efficient LLM Reasoning

• 08-03 中

  BLADE 阅读笔记：把「早退」这件事从自我怀疑词扩展到普通句子边界

• 06-12 EN

  SliceGPT: Post-Training LLM Compression via Computational Invariance

• 06-12 中

  SliceGPT 阅读笔记：用计算不变性删除 Transformer 的行与列

• 06-11 EN

  MegaScale: Engineering 55% MFU at 12,288 GPUs for LLM Training

• 06-11 中

  MegaScale：ByteDance 如何在 12,288 块 GPU 上实现 55% MFU 的大规模 LLM 训练

• 04-22 EN

  SAGE: Training-Free Semantic Evidence Composition for Edge-Cloud Inference Under Hard Uplink Budgets

• 03-28 EN

  Mamba: Linear-Time Sequence Modeling with Selective State Spaces — In-Depth Technical Review

• 03-22 EN

  Attention Is All You Need: The Transformer — In-Depth Technical Review

• 02-18 EN

  DeepSeek-V2: Multi-head Latent Attention and DeepSeekMoE — Technical Review

• 02-18 EN

  GLM-5 Technical Review: From Vibe Coding to Agentic Engineering