AbsPyramid: Benchmarking the Abstraction Ability of Language Models with a Unified Entailment Graph
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2023
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929299457572864 |
|---|---|
| author | Wang, Zhaowei Shi, Haochen Wang, Weiqi Fang, Tianqing Zhang, Hongming Choi, Sehyun Liu, Xin Song, Yangqiu |
| author_facet | Wang, Zhaowei Shi, Haochen Wang, Weiqi Fang, Tianqing Zhang, Hongming Choi, Sehyun Liu, Xin Song, Yangqiu |
| contents | Cognitive research indicates that abstraction ability is essential in human intelligence, which remains under-explored in language models. In this paper, we present AbsPyramid, a unified entailment graph of 221K textual descriptions of abstraction knowledge. While existing resources only touch nouns or verbs within simplified events or specific domains, AbsPyramid collects abstract knowledge for three components of diverse events to comprehensively evaluate the abstraction ability of language models in the open domain. Experimental results demonstrate that current LLMs face challenges comprehending abstraction knowledge in zero-shot and few-shot settings. By training on our rich abstraction knowledge, we find LLMs can acquire basic abstraction abilities and generalize to unseen events. In the meantime, we empirically show that our benchmark is comprehensive to enhance LLMs across two previous abstraction tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2311_09174 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | AbsPyramid: Benchmarking the Abstraction Ability of Language Models with a Unified Entailment Graph Wang, Zhaowei Shi, Haochen Wang, Weiqi Fang, Tianqing Zhang, Hongming Choi, Sehyun Liu, Xin Song, Yangqiu Computation and Language Artificial Intelligence Cognitive research indicates that abstraction ability is essential in human intelligence, which remains under-explored in language models. In this paper, we present AbsPyramid, a unified entailment graph of 221K textual descriptions of abstraction knowledge. While existing resources only touch nouns or verbs within simplified events or specific domains, AbsPyramid collects abstract knowledge for three components of diverse events to comprehensively evaluate the abstraction ability of language models in the open domain. Experimental results demonstrate that current LLMs face challenges comprehending abstraction knowledge in zero-shot and few-shot settings. By training on our rich abstraction knowledge, we find LLMs can acquire basic abstraction abilities and generalize to unseen events. In the meantime, we empirically show that our benchmark is comprehensive to enhance LLMs across two previous abstraction tasks. |
| title | AbsPyramid: Benchmarking the Abstraction Ability of Language Models with a Unified Entailment Graph |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2311.09174 |