LPFQA is an authentic, structured, and interdisciplinary dataset with long-tail knowledge for evaluating LLMs’ ability in complex reasoning, providing a robust benchmark for assessing and advancing LLM performance in real-world professional contexts.
Innovated evaluation dimension design. LPFQA contains a set of fine-grained evaluation dimensions, including knowledge depth, reasoning ability, terminology comprehension, and… See the full description on the dataset page:
https://huggingface.co/datasets/m-a-p/LPFQA.
