AILLM inference dataset announced with metadata on 6.12 billion requests
The team behind the dataset says it spans a year and hopes it can support research into real-world workloads, system design and infrastructure optimization.
TLDR
The announcement describes one year of metadata traces covering 6.12 billion LLM inference requests—requests to run large language models. The team hopes researchers can use the dataset to better understand real-world serving workloads and study system design and infrastructure optimization. It points readers to data.agentic-system.org to explore the data.
Combined views
27
2 Sources, first seen 11h ago
LLM inference dataset announced with metadata on 6.12 billion requests
The team behind the dataset says it spans a year and hopes it can support research into real-world workloads, system design and infrastructure optimization.