lighteval

LLM evaluation results

Full reproduction, benchmark evaluation, and mechanistic audit of Tina (DeepSeek-R1-Distill-Qwen-1.5B LoRA-GRPO) on Dual Modified RTX 4090 48GB (96GB VRAM).

Rank-targeted nested sequential design for LLM evaluation: reach the same ranking conclusion for less, and see which comparisons the data never supported.

DPO implementation and evaluation for EPFL CS552 Modern NLP (Spring 2025), with scripts, Hugging Face models, and LightEval-ready datasets.

Evaluate and compare language models using lighteval for benchmarks and custom tasks, with tools for flexible, efficient analysis.

Wrapper of common LLM evaluation frameworks