Team Ai
Datasetpublic

hoangphu7122002ai/llm-serving-bench-data

llm-serving-bench data Derived data for the LLM serving capacity study in https://github.com/hoangphu7122002/llm-serving-bench, built by src/workloads/corpus.py (sources, versions and licenses: src/workloads/README.md). Each file keeps the license of its source: corpus/chat.jsonl — user turns sampled from ShareGPT V3 (anon8231489123/ShareGPT_Vicuna_unfiltered, Apache-2.0 on the Hub; conversations with ChatGPT — research use). corpus/rag.jsonl — English Wikipedia 20231101… See the full description on the dataset page: https://huggingface.co/datasets/hoangphu7122002ai/llm-serving-bench-data.

sourceHugging Faceotherupdated 3d agoView on Hugging Face
0likes74downloads
Dataset Card

llm-serving-bench data

Derived data for the LLM serving capacity study in https://github.com/hoangphu7122002/llm-serving-bench, built by src/workloads/corpus.py (sources, versions and licenses: src/workloads/README.md). Each file keeps the license of its source:

  • —corpus/chat.jsonl — user turns sampled from ShareGPT V3 (anon8231489123/ShareGPTVicunaunfiltered, Apache-2.0 on the Hub; conversations with ChatGPT — research use).
  • —corpus/rag.jsonl — English Wikipedia 20231101 articles. CC BY-SA 3.0, attribution: Wikipedia contributors.
  • —corpus/code.jsonl — codeparrot-clean-valid Python files, only MIT / Apache-2.0 / BSD / ISC licensed files.
  • —corpus/code_instructions.jsonl — CodeAlpaca-20k instructions (sahil2801/CodeAlpaca-20k), CC BY 4.0.
  • —traces/requests.parquet (from tag v2) — request lengths and arrival times only, no text. Schema ts, source, kind, in_tok, out_tok, session_id, elapsed_s, built by workloads/adapters/traces.py. CC BY 4.0:
  • —Azure LLM inference trace 2024 (Azure/AzurePublicDataset, conv + code, 1 week). Cite: Stojkovic et al., "DynamoLLM: Designing LLM Inference Clusters for Performance and Energy Efficiency", HPCA 2025.
  • —BurstGPT v2.0 BurstGPT_3.csv (HPMLL/BurstGPT). Cite: Wang et al., "BurstGPT: A Real-world Workload Dataset to Optimize LLM Serving Systems".

Versions: v1 = corpus only · v2 = v1 corpus unchanged + traces.