r0b0bench
Specification for a reproducible, provenance-bound multi-lane LLM benchmark suite.
- 11
- Python
- MIT
- 5 wk ago
Specification for a reproducible, provenance-bound multi-lane LLM benchmark suite.
Self-hosted LLM evaluation lab: MMLU-Pro / BFCL / IFEval / LiveCodeBench / HumanEval+ runs on Tesla V100 and RTX 3090, llama.cpp serving (incl. Volta sm_70), patched API benchmark clients