{"id":13535086,"url":"https://github.com/NoviScl/BERT-RACE","last_synced_at":"2025-04-02T00:32:22.554Z","repository":{"id":135906828,"uuid":"167188162","full_name":"NoviScl/BERT-RACE","owner":"NoviScl","description":null,"archived":false,"fork":false,"pushed_at":"2022-10-22T20:19:09.000Z","size":107,"stargazers_count":79,"open_issues_count":1,"forks_count":21,"subscribers_count":5,"default_branch":"master","last_synced_at":"2024-08-02T08:09:52.610Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/NoviScl.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null}},"created_at":"2019-01-23T13:29:33.000Z","updated_at":"2024-06-05T07:10:57.000Z","dependencies_parsed_at":"2023-07-12T01:00:46.501Z","dependency_job_id":null,"html_url":"https://github.com/NoviScl/BERT-RACE","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NoviScl%2FBERT-RACE","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NoviScl%2FBERT-RACE/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NoviScl%2FBERT-RACE/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NoviScl%2FBERT-RACE/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/NoviScl","download_url":"https://codeload.github.com/NoviScl/BERT-RACE/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":222788514,"owners_count":17037777,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-08-01T08:00:49.565Z","updated_at":"2024-11-02T23:30:28.093Z","avatar_url":"https://github.com/NoviScl.png","language":"Python","readme":"# BERT for RACE\n\nBy: Chenglei Si (River Valley High School)\n\n### Update:\nXLNet has achieved impressive gains on RACE recently. You may refer to my other repo: https://github.com/NoviScl/XLNet_DREAM to see how to use XLNet for multiple-choice machine comprehension problems. Huggingface has updated their work [pytorch_trainsformers](https://github.com/huggingface/pytorch-transformers), please refer to their repo for the documentation and more details of the new version. \n\n### Implementation\nThis work is based on Pytorch implementation of BERT (https://github.com/huggingface/pytorch-pretrained-BERT). I adapted the original BERT model to work on multiple choice machine comprehension.\n\n### Environment:\nThe code is tested with Python3.6 and Pytorch 1.0.0.\n\n### Usage\n1. Download the dataset and unzip it. The default dataset directory is ./RACE\n2. Run ```./run.sh```\n\n### Hyperparameters\nI did some tuning and find the following hyperparameters to work reasonally well:\n\nBERT_base: batch size: 32, learning rate: 5e-5, training epoch: 3\n\nBERT_large: batch size: 8, learning rate: 1e-5 (DO NOT SET IT TOO LARGE), training epoch: 2\n\n### Results\nModel | RACE | RACE-M | RACE-H \n--- | --- | --- | --- |\nBERT_base | 65.0 | 71.7 | 62.3 \nBERT_large | 67.9 | 75.6 | 64.7\n\nYou can compare them with other results on the [leaderboard](http://www.qizhexie.com/data/RACE_leaderboard).\n\nBERT large achieves the current (Jan 2019) best result. Looking forward to new models that can beat BERT!\n\n### More Details\nI have written a short [report](./BERT_RACE.pdf) in this repo describing the details.\n\n\n\n\n","funding_links":[],"categories":["BERT QA \u0026 RC task:"],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FNoviScl%2FBERT-RACE","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FNoviScl%2FBERT-RACE","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FNoviScl%2FBERT-RACE/lists"}