Stories by zagwdt

Inference Optimization for MiniMax Sparse Attention

DeepSeek V4 in vLLM: Efficient Long-Context Attention

Introspective Diffusion Language Models

EinsteinArena: Harnessing the collective intelligence of agents in the wild

RL Meets Adaptive Speculative Training

Weak models excel at long context tasks

TorchSpec: Speculative Decoding Training at Scale

Flash Attention 4

CoderForge-Preview: SOTA open dataset for training efficient coding agents

Two years of vector search at Notion: 10x scale, 1/10th cost

Consistency diffusion language models: Up to 14x faster, no quality loss