模型介绍
**[news]** A cross-lingual extension of SapBERT will appear in the main onference of **ACL 2021**! **[news]** SapBERT will appear in the conference proceedings of **NAACL 2021**! SapBERT by Liu et al. (2020). Trained with UMLS 2020AA (English only), using microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext as the base model. The input should be a string of biomedical entity names, e.g., "covid infection" or "Hydroxychloroquine". The [CLS] embedding of the last layer is regarded as the output.
下方摘要由共用资料处理模型依据来源整理,非本站实测;选型结论请结合原始资料和实际业务验证。
主要特点
- Specialized for biomedical entity name understanding using UMLS 2020AA (English-only) ontology
- Built on microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext, pretrained on PubMed abstracts and full-text articles
- Outputs the [CLS] token embedding from the last transformer layer for semantic similarity tasks
局限与注意事项
- English-only; no multilingual capability despite ACL 2021 cross-lingual extension announcement (model itself is not cross-lingual)
- Trained exclusively on biomedical entity names—untested or unsupported for general domain text or sentence-level NLU
- Requires input as raw entity strings (e.g., "covid infection"); no built-in preprocessing, normalization, or disambiguation
来源报价与费用分析
同一模型在不同渠道、地区和配置下可能采用不同价格。下表独立保留各渠道报价,不将渠道价格视为厂商官方统一价格。
| 来源 / 渠道 | 输入 | 输出 | 计费单位 | 采集时间 |
|---|---|---|---|---|
| 暂未取得可核验报价,请查阅来源渠道。 | ||||
按月估算基础 Token 费用
预估不包含缓存优惠、阶梯定价、多模态附加费用、网络检索、税费及推理额外输出。
公开评测与运行配置
暂无已采集的可核验公开评分。模型未上榜不代表能力较低。
查看完整已收录榜单 →常见问题
Is this model multilingual?
No—this specific checkpoint (SapBERT-from-PubMedBERT-fulltext) is English-only. A cross-lingual extension was announced for ACL 2021, but it is a separate development not reflected in this model.
What biomedical knowledge source was used for training?
It was trained using the Unified Medical Language System (UMLS) 2020AA, an English-language biomedical ontology and terminology resource.
What is the output format of this model?
The model outputs the [CLS] token embedding from the last transformer layer—a fixed-dimensional vector suitable for similarity computation or downstream classification.
开放权重是否代表可以免费商用?
不一定。应核对具体版本许可证、商用门槛和使用政策。自部署还会产生算力和维护成本。
这些价格是实时结算价格吗?
不是实时结算价格,而是带采集时间的来源报价快照。请到对应渠道核对当前版本、区域、折扣和计费方式。
资料来源与可追溯记录
许可证:apache-2.0。模型介绍与第三方评测均以原来源的具体版本和方法为准。