LMCache Agentic Dataset Collection
- Type
- dataset
- Venue
- sammshen
- Year
- 2026
- Source
- huggingface
- Access
- free
- Added
- 2026-07-17T20:02:44.842792+00:00
- Verified
- 2026-07-17T20:02:44.842792+00:00
Summary
A curated dataset collection of **787 multi-turn agentic LLM sessions** (24,881 total LLM iterations) designed for benchmarking stateful LLM serving systems. Every session exhibits at least 5 turns with prefix growth and builds to at least 10K tokens of context — making it ideal for evaluating tiered KV Cache solutions like [LMCache](https://github.com/LMCache/LMCache).
Keywords
hf-dataset language-modeling parquet optimized-parquet tabular text datasets dask polars mlcroissant kv-cache llm-serving agentic multi-turn traces benchmark
Topics
Language Modeling
Research notes
- downloads=1355; likes=12