4x faster LLM inference (Flash Attention guy's company)
https://www.together.ai/blog/adaptive-learning-speculator-system-atlas
#HackerNews #4xFasterInference #FlashAttention #LLMTechnology #AIInnovation #AdaptiveLearning
AdapTive-LeArning Speculator System (ATLAS): A New Paradigm in LLM Inference via Runtime-Learning Accelerators
LLM inference that gets faster as you use it. Our runtime-learning accelerator adapts continuously to your workload, delivering 500 TPS on DeepSeek-V3.1, a 4x speedup over baseline performance without manual tuning.