v0.1.0 - Initial Release
Initial release of DPBench - A benchmark framework for evaluating LLM multi-agent coordination under resource contention.
Features
- Simultaneous and sequential decision modes
- Configurable agent counts (N philosophers)
- Optional inter-agent communication
- Comprehensive metrics: deadlock rate, throughput, fairness, starvation
- JSONL logging and human-readable transcripts
- Works with any LLM via simple function protocol