Datasets:

recursal
/

longbench-v2

_id string	domain string	difficulty string	length string	question string	choices list	answer class label	context string
66ec0c4c821e116aacb1994a	Multi-Document QA	easy	medium	Which of the following statements is correct?	["Both contractor data and data crawled from the Internet are used to train VPT agents to model stat(...TRUNCATED)	3D	"Video PreTraining (VPT): Learning to Act by\nWatching Unlabeled Online Videos\nAbstract\nPretrainin(...TRUNCATED)
66ed875e821e116aacb2023e	Multi-Document QA	hard	medium	On what issue do Wang Hui and Badiou have similar views?	["Fixed identities should be transformed into dynamic identities","The problem with contemporary pol(...TRUNCATED)	0A	"SAINT PAUL \n\n\nSAINT PAUL \nThe Foundation of \nUniversalism \nAlain Badiou \nTranslated by Ray B(...TRUNCATED)
66f599ef821e116aacb34099	Multi-Document QA	hard	short	Which of the following descriptions is correct?	["Both StereoSet and CrowS-Pairs used word-filling testing methods to detect the anti-stereotype abi(...TRUNCATED)	1B	"CrowS-Pairs: A Challenge Dataset for Measuring Social Biases\nin Masked Language Models\nAbstract\n(...TRUNCATED)
66f167c3821e116aacb274f4	Multi-Document QA	hard	medium	"Both ‘Sport and Secessionism’ and ‘Discrimination, Sport, and Nation Building among Indonesia(...TRUNCATED)	["The use of sports to promote national unity often leads to failure when minority groups are expect(...TRUNCATED)	2C	"1\n2\n3\n4\n5\nContents\nNotes on contributors\nIntroduction\nMARIANN VACZI AND ALAN BAIRNER\nSport(...TRUNCATED)
66eefe85821e116aacb228dc	Multi-Document QA	hard	short	"These are two articles about grassland simulation. The first article is \"Responsive Real Time Gras(...TRUNCATED)	["In the first article, some unimportant leaves were removed to save performance, and the second art(...TRUNCATED)	0A	"Responsive Real-Time Grass Rendering for General 3D Scenes\nKlemens Jahrmann∗\nMichael Wimmer†\(...TRUNCATED)
66f95126bb02136c067c5070	Multi-Document QA	easy	short	"Compared with lets verify step by step, which of the following points is not included in the improv(...TRUNCATED)	["math-shepherd uses reinforcement learning to improve model capabilities and provides specific trai(...TRUNCATED)	2C	"Let’s Verify Step by Step\nOpenAI\nAbstract\nIn recent years, large language models have greatly (...TRUNCATED)
66f2adaf821e116aacb2aca9	Multi-Document QA	hard	long	"In both The Varieties of Grounded Theory and Advancing Grounded Theory with Mixed Methods, the auth(...TRUNCATED)	["The combination of grounded theory and mixed methods often requires researchers to compromise on t(...TRUNCATED)	1B	"THE VARIETIES \nOF GROUNDED \nTHEORY\n\n\nIn 1976 SAGE published a series of short ‘university pa(...TRUNCATED)
66f14c70821e116aacb271ee	Multi-Document QA	hard	short	What is the difference between the datasets of the two papers?	["The data set in DAIR-V2X includes actual measured data of V2V and V2I, while the data set in V2X-S(...TRUNCATED)	2C	"DAIR-V2X: A Large-Scale Dataset for Vehicle-Infrastructure Cooperative\n3D Object Detection\nHaibao(...TRUNCATED)
66ebee0a5a08c7b9b35e1d05	Multi-Document QA	hard	short	"In the Phidias model, the loss function for reference-augmented multi-view diffusion is expressed a(...TRUNCATED)	["Meta-ControlNet introduces alignment-weighted gradients where the similarity between the 3D refere(...TRUNCATED)	3D	"PHIDIAS: A GENERATIVE MODEL FOR CREATING 3D\nCONTENT FROM TEXT, IMAGE, AND 3D CONDITIONS\nWITH REFE(...TRUNCATED)
66f2cacb821e116aacb2ba50	Multi-Document QA	hard	medium	According to the two articles above, which of the following statements is incorrect?	["Using the following prompt to generate a specific molecular will get a better performance on molT5(...TRUNCATED)	2C	"The Impact of Large Language Models on Scientific Discovery:\na Preliminary Study using GPT-4\nMicr(...TRUNCATED)

End of preview.

Citation

@article{bai2024longbench2,
  title={LongBench v2: Towards Deeper Understanding and Reasoning on Realistic Long-context Multitasks},
  author={Yushi Bai and Shangqing Tu and Jiajie Zhang and Hao Peng and Xiaozhi Wang and Xin Lv and Shulin Cao and Jiazheng Xu and Lei Hou and Yuxiao Dong and Jie Tang and Juanzi Li},
  journal={arXiv preprint arXiv:2412.15204},
  year={2024}
}

Downloads last month: 79