New
huggingface/lighteval v0.13.0 — An LLM Evaluation Toolkit That Runs on Multiple Backends
huggingface/lighteval is a Python toolkit that evaluates models served across multiple execution backends or loaded into memory, producing task-, metric- and sample-level results; the latest public release is v0.13.0.