Question Answering On Drop Test

Results

Performance results of various models on this benchmark

		Paper Title
QDGAT (ensemble)	88.38	Question Directed Graph Attention Network for Numerical Reasoning over Text
POET	87.6	Reasoning Like Program Executors
PaLM 2 (few-shot)	85.0	PaLM 2 Technical Report
BERT+Calculator (ensemble)	81.78	Giving BERT a Calculator: Finding Operations and Arguments with Reading Comprehension
NeRd	81.71	Neural Symbolic Reader: Scalable Integration of Distributed and Symbolic Representations for Reading Comprehension
GPT-4 (few-shot, k=3)	80.9	GPT-4 Technical Report
TASE-BERT	80.7	A Simple and Effective Model for Answering Multi-span Questions
MTMSN Large	79.88	A Multi-Type Multi-Span Network for Reading Comprehension that Requires Discrete Reasoning
GenBERT (+ND+TD)	72.4	Injecting Numerical Reasoning Skills into Language Models
NumNet	67.97	NumNet: Machine Reading Comprehension with Numerical Reasoning
GPT 3.5 (few-shot, k=3)	64.1	GPT-4 Technical Report
Orca 2-7B	60.26	Orca 2: Teaching Small Language Models How to Reason
Orca 2-13B	57.97	Orca 2: Teaching Small Language Models How to Reason
NAQA Net	47.01	DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphs
GPT-3 175B (few-shot, k=32)	36.5	Language Models are Few-Shot Learners
BERT	32.7	DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphs

0 of 16 row(s) selected.