Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.19911825This dataset provides anonymized request-level traces for studying large language model (LLM) serving workloads. It is collected from a large-scale, multi-model production environment and has been fully desensitized to remove any user-identifiable or platform-specific information. Each record corresponds to a single request and includes:- input length (number of tokens)- output length (number of tokens)- end-to-end latency- relative timestamp- model architecture type (e.g., Dense or MoE)- model scale category (e.g., <=10B, 10–30B, 30–100B, >100B) To preserve privacy and anonymity:- All timestamps are converted to relative time- No user identifiers, prompts, or raw content are included- Model names and platform identifiers are abstracted This dataset represents a subset of a larger production trace and is intended for reproducibility and evaluation purposes in anonymous peer review.
Keyword(s):
LLM serving, workload trace.
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
No Associated Publication Data Available
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)