All posts on technology, daily life, and thoughts.
Branching on Compressed Memory: Adding Tree Reasoning to DeepSeek-V4 TL;DR Adding tree-structured speculative decoding to DeepSeek-V4-Flash yields longer acceptance…
Flash-dLLM: I/O-First Diffusion LLM Acceleration, Unlocking 210.6 tokens/s with Fused Cache and Self-Verification One-Line Summary (TL;DR) Diffusion LLMs (dLLMs) support…
Parallel Inference without Drafter-Side KV Cache: H-Spec's Hybrid Speculative Decoding TL;DR shows that the bottleneck in concurrent serving is that a block-diffusion…
Analytical PD-inference provisioning uses instance counts and KV-cache reservations to find the lowest-power deployment for a target capacity.
VC-Attention is a training-free kernel that removes value-quantization and softmax bottlenecks in low-bit video-DiT attention, speeding it up 1.59× on B200.
ActionPiece: Rethinking Action Tokenization Beyond Reconstruction Error, via Physical Ranking TL;DR — In autoregressive VLA (vision-language-action) models, the action…