onejune2018/Awesome-LLM-Eval

Organizes evaluation resources around an **anthropomorphic and value-oriented taxonomy** that extends beyond traditional benchmarks to assess reasoning robustness, uncertainty quantification, and long-context capabilities. Integrates references to specialized evaluation tools (OpenCompass, DeepEval, AlpacaEval), domain-specific benchmarks (RAG, agents, coding, multimodal), and LLMOps frameworks, while maintaining structured categorization of pre-trained, instruction-tuned, and aligned models. Backed by a peer-reviewed survey paper that provides the methodological foundation for its continuously updated categorization scheme.

616 stars.