[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-188059-en":3,"doc-seo-188059-105":30,"detail-sidebar-cat-1-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":11,"category_id":12,"category_name":13,"doc_title":14,"doc_description":15,"doc_content":16,"file_id":17,"file_url":18,"file_type":19,"file_size":20,"view_count":4,"is_deleted":4,"is_public":11,"is_downloadable":11,"audit_status":11,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":15,"update_tm":28,"read_time":29},188059,962088121634,"supergirl","https://ap-avatar.wpscdn.com/davatar_9964176cb1d06d4a9deccf72a44ae3dc",1,158,"General","Template-based Probing - Template-free Probing","Template-based probing and template-free probing are compared using multiple biomedical and general-purpose language models. The content contrasts an explicit pattern such as “[X](born [MASK])” against a template-free setup, and evaluates downstream knowledge extraction across several benchmark settings. Model configurations are described in terms of parameter scale and pretraining sources (e.g., PubMed, PMC, Wikipedia, and clinical corpora), followed by performance tables reporting accuracy-like metrics (A@1, A@5, A@10) and recall-style scores across datasets such as Google-RE, SQuAD, T-REx, and biomedical knowledge tasks.","| Template-based Probing | Template-free Probing |\n| --- | --- |\n| Template: “[X](born [MASK])” | N/A |\n| Peter F. Martin (born [MASK]) | Peter F Martin (born [MASK]) is an American politician [...] |\n| Dennis B. Sullivan (born [MASK]) | Sullivan was born in Chippewa Falls Wisconsin in [MASK] |\n| Tan Jiexi (born [MASK]) | Tan Jiexi (born December 2,[MASK] in Shenzhen,China), is a Chinese singer-songwriter |\n| Tasos Neroutsos (born [MASK]) | Neroutsos was born in Athens in [MASK] to a wealthy family |\n\n\n| Model | Parameters Data |  |\n| --- | --- | --- |\n|  |  |  |\n| ⋆ PubMedBERT | 109M | PubMed abstracts+PMC full-text articles (3.2B words/21GB) |\n| ⋆ Bioformer | 42M | 33M PubMed abstracts+1M PMC full-text articles |\n| ⋆ BioM-ELECTRA-Generator | 49M | PubMed Abstracts |\n|  †BioMed-RoBERTa  | \u003Cbr>124M | \u003Cbr>RoBERTa (160GB)+Semantic Scholar corpus (2.68M papers/47GB) |\n| †COVID Bert | 108M | N/A |\n| †BlueBert | 109M | Bert+PubMed abstracts+MIMIC-III clinical notes (4500M words/27GB) |\n| †Bio Discharge Summary BERT | 108M | Biobert (18B words)+MIMIC III discharge summaries (880M words) |\n| †PMC RoBERTa | 355M | RoBERTa (160GB)+ PMC and PubMd abstracts |\n| †Bio ClinicalBERT | 108M | Biobert (18B words)+MIMIC notes (880M words) |\n|  ⋄ RoBERTa-base  | \u003Cbr>124M |  BookCorpus, English Wikipedia, CC-News, OpenWebText, Stories (160GB)  |\n| ⋄ RoBERTa-large | 355M | BookCorpus, English Wikipedia, CC-News, OpenWebText, Stories (160GB) |\n| ⋄ BERT-base | 109M | BookCorpus, English Wikipedia (16GB) |\n| ⋄ BERT-large | 334M | BookCorpus, English Wikipedia (16GB) |\n| ⋄ ALBERT-base | 12M | BookCorpus, English Wikipedia (16GB) |\n| ⋄ ALBERT-base | 18M | BookCorpus, English Wikipedia (16GB) |\n| ⋄ DistilBERT | 12M | BookCorpus, English Wikipedia (16GB) |\n\n\n| Model | A@1 | Google-RE |  | R | A@1 | T-REx |  | R | Biomed-Wikidata |  |  | R | A@1 | CTD\u003Cbr>A@5 | A@10 | R |\n| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |\n|  |  | A@5 | A@10 |  |  | A@5 | A@10 |  | A@1 | A@5 | A@10 |  |  |  |  |  |\n|  ⋆ PubMedBERT  |  5.4e−3  |  1.7e−2  |  3.2e−2  |  8  |  1.1e − 1  |  2.1e − 1  |  2.7e − 1  |  6  |  4.4e−2  |  1.3e − 1  |  1.9e − 1  |  1  |  7.8e−3  |  2.9e−2  |  4.5e−2  |  1  |\n| ⋆ Bioformer | 9.8e−4 | 6.2e−3 | 1.3e−2 | 15 | 9.3e−2 | 1.7e − 1 | 2.1e − 1 | 7 | 3.7e−2 | 1.0e − 1 | 1.6e − 1 | 2 | 6.0e−3 | 2.3e−2 | 3.6e−2 | 2 |\n| ⋆ BioM-ELECTRA | 1.3e−3 | 7.8e−2 | 1.5e−2 | 13 | 5.4e−2 | 1.3e − 1 | 1.8e − 1 | 9 | 3.0e−2 | 1.0e − 1 | 1.5e − 1 | 3 | 3.4e−3 | 1.4e−2 | 2.6e−2 | 3 |\n|  †BioMed-RoBERTa  | \u003Cbr>1.2e−2 |  3.0e−2  |  4.3e−2  | \u003Cbr>6 |  5.1e−2  | \u003Cbr>1.2e − 1 | \u003Cbr>1.7e − 1 |  11  |  2.3e−3  | \u003Cbr>1.4e−2 |  3.4e−2  |  12  |  3.6e−4  |  3.7e−3  |  8.4e−3  | \u003Cbr>6 |\n| †COVID Bert | 9.8e−4 | 4.5e−3 | 1.2e−2 | 16 | 2.8e−2 | 6.8e−2 | 1.1e − 1 | 13 | 8.1e−3 | 4.3e−2 | 6.4e−2 | 6 | 4.0e−3 | 1.0e−2 | 2.0e−2 | 4 |\n| †BlueBert | 1.6e−3 | 9.9e−3 | 1.6e−2 | 12 | 3.5e−2 | 1.0e − 1 | 1.5e − 1 | 14 | 1.3e−2 | 5.1e−2 | 8.7e−2 | 4 | 5.4e−4 | 5.4e−3 | 9.8e−3 | 12 |\n| †Discharge BERT | 9.8e−4 | 7.2e−3 | 1.3e−2 | 14 | 2.1e−2 | 6.2e−2 | 9.7e−2 | 15 | 6.8e−3 | 3.7e−2 | 6.0e−2 | 10 | 3.9e−3 | 1.2e−2 | 2.0e−2 | 5 |\n| †PMC RoBERTa | 3.2e−3 | 1.7e−2 | 3.5e−2 | 10 | 4.0e−2 | 9.4e−2 | 1.3e − 1 | 12 | 2.0e−3 | 1.7e−2 | 3.0e−2 | 14 | 8.1e−4 | 3.7e−3 | 8.5e−3 | 11 |\n| †Bio ClinicalBERT | 1.9e−3 | 5.0e−3 | 1.0e−2 | 11 | 1.2e−2 | 4.1e−2 | 6.6e−2 | 16 | 7.8e−3 | 3.6e−2 | 6.2e−2 | 7 | 1.9e−3 | 1.0e−2 | 1.5e−2 | 7 |\n|  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |\n| ⋄ RoBERTa-base | 6.3e−3 | 2.5e−2 | 4.9e−2 | 7 | 5.3e−2 | 9.4e−2 | 1.2e − 1 | 9 | 1.3e−3 | 2.2e−2 | 3.6e−2 | 15 | 3.6e−4 | 4.4e−3 | 7.9e−3 | 16 |\n| ⋄ RoBERTa-large | 4.2e−3 | 2.3e−2 | 4.2e−2 | 9 | 5.6e−2 | 1.1e − 1 | 1.5e − 1 | 8 | 2.0e−3 | 2.0e−2 | 3.7e−2 | 13 | 5.4e−4 | 3.7e−3 | 7.4e−3 | 13 |\n| ⋄ BERT-base | 1.1e − 1 | 2.3e − 1 | 3.2e − 1 | 2 | 3.0e − 1 | 5.3e − 1 | 6.4e − 1 | 2 | 7.8e−3 | 3.4e−2 | 6.0e−2 | 8 | 8.1e−4 | 3.9e−3 | 9.6e−3 | 10 |\n| ⋄ BERT-large | 1.1e − 1 | ","cbCaiuBVaFxqSZ0u","https://ap.wps.com/l/cbCaiuBVaFxqSZ0u","pdf",548205,11,"English","en",105,"# Template-based Probing\n## Template-free Probing\n# Model Parameters Data\n# Evaluation Results","[{\"question\":\"What is the difference between template-based and template-free probing in this document?\",\"answer\":\"Template-based probing uses an explicit text pattern like “[X](born [MASK])” to create masked queries, while template-free probing skips that templated formulation.\"},{\"question\":\"Which kinds of language models are evaluated?\",\"answer\":\"The document compares multiple transformer-based models, including biomedical variants (e.g., PubMedBERT, Bioformer, BioM-ELECTRA, Bio ClinicalBERT) and general models (e.g., BERT, RoBERTa, ALBERT, DistilBERT).\"},{\"question\":\"How are model performance results presented?\",\"answer\":\"Results are shown in tables using metrics such as A@1, A@5, A@10 and an additional recall-like measure R across multiple benchmarks (e.g., Google-RE, SQuAD, T-REx, and biomedical relation tasks).\"}]","Template-based Probing - Template-free Probing | PDF",1788387043,4,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":14,"keywords":34,"description":15,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"template-based-probing-template-free-probing","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":11},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/template/","Template",2,{"item":49,"name":13,"@type":43,"position":50},"https://docshare.wps.com/template/general/",3,{"item":52,"name":14,"@type":43,"position":29},"https://docshare.wps.com/template/template-based-probing-template-free-probing/188059/",{"url":52,"name":14,"@type":54,"author":55,"headline":14,"publisher":57,"fileFormat":60,"inLanguage":23,"description":15,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-09-04","2026-09-02",true,{"@type":65,"interactionType":66,"userInteractionCount":11},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What is the difference between template-based and template-free probing in this document?","Question",{"text":75,"@type":76},"Template-based probing uses an explicit text pattern like “[X](born [MASK])” to create masked queries, while template-free probing skips that templated formulation.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which kinds of language models are evaluated?",{"text":80,"@type":76},"The document compares multiple transformer-based models, including biomedical variants (e.g., PubMedBERT, Bioformer, BioM-ELECTRA, Bio ClinicalBERT) and general models (e.g., BERT, RoBERTa, ALBERT, DistilBERT).",{"name":82,"@type":73,"acceptedAnswer":83},"How are model performance results presented?",{"text":84,"@type":76},"Results are shown in tables using metrics such as A@1, A@5, A@10 and an additional recall-like measure R across multiple benchmarks (e.g., Google-RE, SQuAD, T-REx, and biomedical relation tasks).","https://schema.org",{"og:url":52,"og:type":87,"og:title":14,"og:site_name":58,"og:description":15},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,102,107,112,117,122,127,132],{"id":21,"doc_module":11,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Presentations",90,"presentations",{"id":98,"doc_module":11,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},12,"Resumes",80,"resumes",{"id":103,"doc_module":11,"doc_module_name":46,"category_name":104,"show_sort_weight":105,"slug":106},14,"Invoices",70,"invoices",{"id":108,"doc_module":11,"doc_module_name":46,"category_name":109,"show_sort_weight":110,"slug":111},15,"Posters",60,"posters",{"id":113,"doc_module":11,"doc_module_name":46,"category_name":114,"show_sort_weight":115,"slug":116},16,"Social Media",50,"social-media",{"id":118,"doc_module":11,"doc_module_name":46,"category_name":119,"show_sort_weight":120,"slug":121},17,"Forms",40,"forms",{"id":123,"doc_module":11,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},18,"Letters",30,"letters",{"id":128,"doc_module":11,"doc_module_name":46,"category_name":129,"show_sort_weight":130,"slug":131},21,"Paper Templates",5,"papers-templates",{"id":12,"doc_module":11,"doc_module_name":46,"category_name":13,"show_sort_weight":4,"slug":133},"general-158"]