SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2204.12632 · NAACL · 2022

Testing the Ability of Language Models to Interpret Figurative Language

Graham Neubig, Emmy Liu, Chenxuan Cui, Kenneth Zheng

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 14 functions out of this paper's own repositories and ran 5 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
nightingal3/fig-qa canonical 1 of 2
simran-khanuja/multilingual-fig-qa pwc_unofficial 4 of 12
FunctionStatusWhere it lives
eval_model Ran simran-khanuja/multilingual-fig-qa/run_baselines.py
code served (permissive licence) · get_code("31e5d4061c0165ee")
evaluate_model Ran nightingal3/fig-qa/src/models/gpt_score.py
code served (permissive licence) · get_code("2a5615f598f3af2b")
is_subarray Ran simran-khanuja/multilingual-fig-qa/get_syntax_chunks.py
code served (permissive licence) · get_code("21007e567ee7726b")
main_train_loop Ran simran-khanuja/multilingual-fig-qa/run_baselines.py
code served (permissive licence) · get_code("454507be28be36dd")
substring_between Ran simran-khanuja/multilingual-fig-qa/get_syntax_chunks.py
code served (permissive licence) · get_code("62a67c92331edd10")
convert_examples_to_features Not yet run simran-khanuja/multilingual-fig-qa/utils.py
code served (permissive licence) · get_code("4f74b4458496f1dc")
convert_multiple_choice_examples_to_features Not yet run simran-khanuja/multilingual-fig-qa/utils.py
code served (permissive licence) · get_code("a7ffc645f7b1c747")
get_hypernyms Not yet run simran-khanuja/multilingual-fig-qa/plot_parsed_data.py
code served (permissive licence) · get_code("10b84b6e2e00afe1")
get_pos_tags Not yet run simran-khanuja/multilingual-fig-qa/plot_parsed_data.py
code served (permissive licence) · get_code("16090c8ed2178c61")
lemmatize_words Not yet run simran-khanuja/multilingual-fig-qa/plot_parsed_data.py
code served (permissive licence) · get_code("792128008f28fb12")
remove_substrings Not yet run simran-khanuja/multilingual-fig-qa/get_syntax_chunks.py
code served (permissive licence) · get_code("581a4523d21cb660")
sent_scoring Not yet run nightingal3/fig-qa/src/models/gpt_score.py
code served (permissive licence) · get_code("39cf438241a8d1dd")
simple_accuracy Not yet run simran-khanuja/multilingual-fig-qa/utils.py
code served (permissive licence) · get_code("3c241ecfe3749a6d")
train_model Not yet run simran-khanuja/multilingual-fig-qa/run_baselines.py
code served (permissive licence) · get_code("e7b3e41b3b961fb4")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Figurative and metaphorical language are commonplace in discourse, and figurative expressions play an important role in communication and cognition. However, figurative language has been a relatively under-studied area in NLP, and it remains an open question to what extent modern language models can interpret nonliteral phrases. To address this question, we introduce Fig-QA, a Winograd-style nonliteral language understanding task consisting of correctly interpreting paired figurative phrases with divergent meanings. We evaluate the performance of several state-of-the-art language models on this task, and find that although language models achieve performance significantly over chance, they still fall short of human performance, particularly in zero-or few-shot settings. This suggests that further work is needed to improve the nonliteral reasoning capabilities of language models. 1

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2204.12632")
get_code_for_paper("2204.12632")
have("2204.12632")

Connect an agent — have() is free.