https://github.com/brucewlee/h-test

[ACL 2024] Language Models Don't Learn the Physical Manifestation of Language

https://github.com/brucewlee/h-test

Science Score: 33.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
  • codemeta.json file
    Found codemeta.json file
  • .zenodo.json file
  • DOI references
  • Academic publication links
    Links to: arxiv.org
  • Committers with academic emails
    1 of 2 committers (50.0%) from academic institutions
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (12.2%) to scientific vocabulary

Keywords

benchmark evaluation language-model
Last synced: 5 months ago · JSON representation

Repository

[ACL 2024] Language Models Don't Learn the Physical Manifestation of Language

Basic Info
Statistics
  • Stars: 1
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Topics
benchmark evaluation language-model
Created almost 2 years ago · Last pushed over 1 year ago
Metadata Files
Readme License

readme.md

H-TEST: Language Models Don’t Learn the Physical Manifestation of Language


Paper Twitter Thread


This repository contains the code and data for the paper "Language Models Don't Learn the Physical Manifestation of Language". The paper introduces H-TEST, a series of tasks designed to assess the ability of language models to understand and utilize the visual and auditory properties of language.

drawing

Overview

The key components of this repository are:

  • data/: Contains scripts for generating test instances for each task in H-TEST.
  • generatetraindata.py: Script for generating training data for fine-tuning experiments.
  • run_test.py: Main script for running H-TEST on various language models.
  • runlettergeometry_test.py: Script for running the Letter Geometry challenge on language models.
  • utils.py: Utility functions for interacting with language model APIs and processing responses.

Usage

Running H-TEST

To run H-TEST on a set of language models, modify the modellist and other setups in runtest.py to include the models you wish to test, and then run the script:

python run_test.py

The script will generate test instances, run the tests on the specified models, and print the results.

Running the Letter Geometry Challenge

To run the Letter Geometry challenge on a set of language models, modify the modellist in runlettergeometrytest.py to include the models you wish to test, and then run the script:

python run_letter_geometry_test.py

The script will generate challenge instances, run the tests on the specified models, and print the results.

Generating Training Data

To generate training data for fine-tuning experiments, run the generatetraindata.py script:

python generate_train_data.py

The script will generate training instances for each task in H-TEST and save them in the specified format in the train/ directory.

Citation

If you use this code or the H-TEST benchmark in your research, please cite our paper:

@misc{lee2024language, title={Language Models Don't Learn the Physical Manifestation of Language}, author={Bruce W. Lee and JaeHyuk Lim}, year={2024}, eprint={2402.11349}, archivePrefix={arXiv}, primaryClass={cs.CL} }

License

This project is licensed under the MIT License. See the LICENSE file for more information.

Owner

  • Name: Bruce W. Lee (이웅성)
  • Login: brucewlee
  • Kind: user
  • Location: Philadelphia, PA
  • Company: University of Pennsylvania

Research Scientist - NLP

GitHub Events

Total
Last Year

Committers

Last synced: over 1 year ago

All Time
  • Total Commits: 3
  • Total Committers: 2
  • Avg Commits per committer: 1.5
  • Development Distribution Score (DDS): 0.333
Past Year
  • Commits: 3
  • Committers: 2
  • Avg Commits per committer: 1.5
  • Development Distribution Score (DDS): 0.333
Top Committers
Name Email Commits
Bruce Lee w****e@g****m 2
Bruce Lee b****s@s****u 1
Committer Domains (Top 20 + Academic)

Issues and Pull Requests

Last synced: 9 months ago

All Time
  • Total issues: 0
  • Total pull requests: 0
  • Average time to close issues: N/A
  • Average time to close pull requests: N/A
  • Total issue authors: 0
  • Total pull request authors: 0
  • Average comments per issue: 0
  • Average comments per pull request: 0
  • Merged pull requests: 0
  • Bot issues: 0
  • Bot pull requests: 0
Past Year
  • Issues: 0
  • Pull requests: 0
  • Average time to close issues: N/A
  • Average time to close pull requests: N/A
  • Issue authors: 0
  • Pull request authors: 0
  • Average comments per issue: 0
  • Average comments per pull request: 0
  • Merged pull requests: 0
  • Bot issues: 0
  • Bot pull requests: 0
Top Authors
Issue Authors
Pull Request Authors
Top Labels
Issue Labels
Pull Request Labels