{"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/chemberta-2-towards-chemical-foundation","title":"ChemBERTa-2: Towards Chemical Foundation Models","arxiv_id":"2209.01712","date":"2022-09-05","proceeding":null,"authors":["Walid Ahmad","Elana Simon","Seyone Chithrananda","Gabriel Grand","Bharath Ramsundar"],"abstract":"Large pretrained models such as GPT-3 have had tremendous impact on modern natural language processing by leveraging self-supervised learning to learn salient representations that can be used to readily finetune on a wide variety of downstream tasks. We investigate the possibility of transferring such advances to molecular machine learning by building a chemical foundation model, ChemBERTa-2, using the language of SMILES. While labeled data for molecular prediction tasks is typically scarce, libraries of SMILES strings are readily available. In this work, we build upon ChemBERTa by optimizing the pretraining process. We compare multi-task and self-supervised pretraining by varying hyperparameters and pretraining dataset size, up to 77M compounds from PubChem. To our knowledge, the 77M set constitutes one of the largest datasets used for molecular pretraining to date. We find that with these pretraining improvements, we are competitive with existing state-of-the-art architectures on the MoleculeNet benchmark suite. We analyze the degree to which improvements in pretraining translate to improvement on downstream tasks.","url_abs":"https://arxiv.org/abs/2209.01712v1","url_pdf":"https://arxiv.org/pdf/2209.01712v1.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":"chemberta-2-towards-chemical-foundation","repo_url":"https://github.com/SylwiaNowakowska/LLM_Fine_Tuning_Molecular_Properties","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"chemberta-2-towards-chemical-foundation","repo_url":"https://github.com/seyonechithrananda/bert-loves-chemistry","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"molecular-property-prediction","task_name":"Molecular Property Prediction"},{"task_slug":"self-supervised-learning","task_name":"Self-Supervised Learning"}],"methods":[{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"attention-dropout","method_name":"Attention Dropout"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"cosine-annealing","method_name":"Cosine Annealing"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"gpt-3","method_name":"GPT-3"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"linear-warmup-with-cosine-annealing","method_name":"Linear Warmup With Cosine Annealing"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"weight-decay","method_name":"Weight Decay"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/molecular-property-prediction-on-bace-1","task":"Molecular Property Prediction","dataset":"BACE","model":"ChemBERTa-2 (MTR-77M)","rank_in_archive_order":12,"of":20,"metrics":{"RMSE":"1.363","ROC-AUC":"79.9"},"uses_additional_data":false},{"leaderboard":"/sota/molecular-property-prediction-on-bbbp-1","task":"Molecular Property Prediction","dataset":"BBBP","model":"ChemBERTa-2 (MTR-77M)","rank_in_archive_order":17,"of":29,"metrics":{"ROC-AUC":"72.8"},"uses_additional_data":false},{"leaderboard":"/sota/molecular-property-prediction-on-clearance","task":"Molecular Property Prediction","dataset":"Clearance","model":"ChemBERTa-2 (MTR-77M)","rank_in_archive_order":2,"of":2,"metrics":{"RMSE":"48.515"},"uses_additional_data":false},{"leaderboard":"/sota/molecular-property-prediction-on-esol","task":"Molecular Property Prediction","dataset":"ESOL","model":"ChemBERTa-2 (MTR-77M)","rank_in_archive_order":18,"of":20,"metrics":{"RMSE":"0.889"},"uses_additional_data":false},{"leaderboard":"/sota/molecular-property-prediction-on","task":"Molecular Property Prediction","dataset":"Lipophilicity","model":"ChemBERTa-2 (MTR-77M)","rank_in_archive_order":9,"of":13,"metrics":{"RMSE":"0.798"},"uses_additional_data":true},{"leaderboard":"/sota/molecular-property-prediction-on-clintox-1","task":"Molecular Property Prediction","dataset":"clintox","model":"ChemBERTa-2 (MTR-77M)","rank_in_archive_order":19,"of":20,"metrics":{"Molecules (M)":"77","ROC-AUC":"56.3"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2209.01712","atlas_url":"https://app.syntology.ai/?focus=2209.01712","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}