{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/improving-the-training-of-graph-neural","title":"SCR: Training Graph Neural Networks with Consistency Regularization","arxiv_id":"2112.04319","date":"2021-12-08","proceeding":null,"authors":["Chenhui Zhang","Yufei He","Yukuo Cen","Zhenyu Hou","Wenzheng Feng","Yuxiao Dong","Xu Cheng","Hongyun Cai","Feng He","Jie Tang"],"abstract":"We present the SCR framework for enhancing the training of graph neural networks (GNNs) with consistency regularization. Regularization is a set of strategies used in Machine Learning to reduce overfitting and improve the generalization ability. However, it is unclear how to best design the generalization strategies in GNNs, as it works in a semi-supervised setting for graph data. The major challenge lies in how to efficiently balance the trade-off between the error from the labeled data and that from the unlabeled data. SCR is a simple yet general framework in which we introduce two strategies of consistency regularization to address the challenge above. One is to minimize the disagreements among the perturbed predictions by different versions of a GNN model. The other is to leverage the Mean Teacher paradigm to estimate a consistency loss between teacher and student models instead of the disagreement of the predictions. We conducted experiments on three large-scale node classification datasets in the Open Graph Benchmark (OGB). Experimental results demonstrate that the proposed SCR framework is a general one that can enhance various GNNs to achieve better performance. Finally, SCR has been the top-1 entry on all three OGB leaderboards as of this submission.","url_abs":"https://arxiv.org/abs/2112.04319v2","url_pdf":"https://arxiv.org/pdf/2112.04319v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"improving-the-training-of-graph-neural","repo_url":"https://github.com/thudm/crgnn","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"improving-the-training-of-graph-neural","repo_url":"https://github.com/thudm/scr","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"improving-the-training-of-graph-neural","repo_url":"https://github.com/THUDM/CRGNN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"improving-the-training-of-graph-neural","repo_url":"https://github.com/THUDM/SCR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"node-classification","task_name":"Node Classification"},{"task_slug":"node-property-prediction","task_name":"Node Property Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/node-property-prediction-on-ogbn-mag","task":"Node Property Prediction","dataset":"ogbn-mag","model":"NARS-GAMLP+RLU+SCR","rank_in_archive_order":12,"of":39,"metrics":{"Ext. data":"No","Number of params":"6734882","Test Accuracy":"0.5631 ± 0.0021","Validation Accuracy":"0.5734 ± 0.0035"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-mag","task":"Node Property Prediction","dataset":"ogbn-mag","model":"NARS-GAMLP+SCR-m","rank_in_archive_order":15,"of":39,"metrics":{"Ext. data":"No","Number of params":"6734882","Test Accuracy":"0.5451 ± 0.0019","Validation Accuracy":"0.5590 ± 0.0028"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-mag","task":"Node Property Prediction","dataset":"ogbn-mag","model":"NARS-GAMLP+SCR","rank_in_archive_order":17,"of":39,"metrics":{"Ext. data":"No","Number of params":"6734882","Test Accuracy":"0.5432 ± 0.0018","Validation Accuracy":"0.5654 ± 0.0021"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-papers100m","task":"Node Property Prediction","dataset":"ogbn-papers100M","model":"GAMLP+RLU+SCR","rank_in_archive_order":3,"of":20,"metrics":{"Ext. data":"No","Number of params":"67560875","Test Accuracy":"0.6842 ± 0.0015","Validation Accuracy":"0.7188 ± 0.0007"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-papers100m","task":"Node Property Prediction","dataset":"ogbn-papers100M","model":"GAMLP+SCR-m","rank_in_archive_order":6,"of":20,"metrics":{"Ext. data":"No","Number of params":"67560875","Test Accuracy":"0.6816 ± 0.0012","Validation Accuracy":"0.7186 ± 0.0008"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-papers100m","task":"Node Property Prediction","dataset":"ogbn-papers100M","model":"GAMLP+SCR","rank_in_archive_order":7,"of":20,"metrics":{"Ext. data":"No","Number of params":"67560875","Test Accuracy":"0.6814 ± 0.0008","Validation Accuracy":"0.7190 ± 0.0007"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GIANT-XRT+SAGN+SCR+C&S","rank_in_archive_order":7,"of":64,"metrics":{"Ext. data":"Yes","Number of params":"1154654","Test Accuracy":"0.8680 ± 0.0007","Validation Accuracy":"0.9357 ± 0.0004"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GIANT-XRT+SAGN+MCR+C&S","rank_in_archive_order":8,"of":64,"metrics":{"Ext. data":"Yes","Number of params":"1154654","Test Accuracy":"0.8673 ± 0.0008","Validation Accuracy":"0.9387 ± 0.0002"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GIANT-XRT+SAGN+SCR","rank_in_archive_order":9,"of":64,"metrics":{"Ext. data":"Yes","Number of params":"1154654","Test Accuracy":"0.8667 ± 0.0009","Validation Accuracy":"0.9364 ± 0.0005"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GIANT-XRT+SAGN+MCR","rank_in_archive_order":10,"of":64,"metrics":{"Ext. data":"Yes","Number of params":"1154654","Test Accuracy":"0.8651 ± 0.0009","Validation Accuracy":"0.9389 ± 0.0002"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GIANT-XRT+GAMLP+MCR","rank_in_archive_order":14,"of":64,"metrics":{"Ext. data":"Yes","Number of params":"2144151","Test Accuracy":"0.8591 ± 0.0008","Validation Accuracy":"0.9402 ± 0.0004"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GAMLP+RLU+SCR+C&S","rank_in_archive_order":15,"of":64,"metrics":{"Ext. data":"No","Number of params":"3335831","Test Accuracy":"0.8520 ± 0.0008","Validation Accuracy":"0.9304 ± 0.0005"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GAMLP+RLU+SCR","rank_in_archive_order":16,"of":64,"metrics":{"Ext. data":"No","Number of params":"3335831","Test Accuracy":"0.8505 ± 0.0009","Validation Accuracy":"0.9292 ± 0.0005"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GAMLP+MCR","rank_in_archive_order":19,"of":64,"metrics":{"Ext. data":"No","Number of params":"3335831","Test Accuracy":"0.8462 ± 0.0003","Validation Accuracy":"0.9319 ± 0.0003"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"SAGN+MCR","rank_in_archive_order":22,"of":64,"metrics":{"Ext. data":"No","Number of params":"2179678","Test Accuracy":"0.8441 ± 0.0005","Validation Accuracy":"0.9325 ± 0.0004"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2112.04319","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2112.04319"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/thudm/crgnn","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/thudm/scr","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/THUDM/SCR","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/THUDM/CRGNN","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran_honours":1,"unverified":2},"by_repo_kind":{"official":{"samples":2,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":1,"samples":[{"code_sha256_prefix":"608920c78f1ae5e9","entry":"get_n_params","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"608920c78f1ae5e9"}},{"code_sha256_prefix":"7fe187a46e4207d3","entry":"load_data","repo":"THUDM/CRGNN","repo_kind":"official","path":"ogbn-mag/heteo_data.py","file_url":"https://github.com/THUDM/CRGNN/blob/HEAD/ogbn-mag/heteo_data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"7fe187a46e4207d3"}},{"code_sha256_prefix":"198da4ebc08d7f6a","entry":"read_relation_subsets","repo":"THUDM/CRGNN","repo_kind":"official","path":"ogbn-mag/heteo_data.py","file_url":"https://github.com/THUDM/CRGNN/blob/HEAD/ogbn-mag/heteo_data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"198da4ebc08d7f6a"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}