Update README.md

README.md

@@ -1,3 +1,36 @@
 ---
+language: 
+- multilingual
+- en
+- zh
 license: apache-2.0
 ---
+
+# KBioXLM
+
+The aligned corpus constructed using the knowledge-anchored method is combined with a multi task training strategy to continue training XLM-R, thus obtaining KBioXLM. It is the first multilingual biomedical pre-trained language model we know that has cross-lingual understanding capabilities in medical domain. It was introduced in the paper [KBioXLM: A Knowledge-anchored Biomedical
+Multilingual Pretrained Language Model]() by geng et al. and first released in [this repository](https://github.com/ngwlh-gl/KBioXLM/tree/main)
+
+## Model description
+KBioXLM model can be fintuned on downstream tasks. The downstream tasks here refer to biomedical cross-lingual understanding tasks, such as biomedical entity recognition, biomedical relationship extraction and biomedical text classification.
+
+## Usage
+
+You can follow the prompts below to load our model parameters:
+
+```python
+from transformers import AutoTokenizer, AutoModelForMaskedLM
+model=RobertaModel.from_pretrained(pretrained_model_name_or_path='ngwlh/KBioXLM', config=self.config)
+checkpoint = torch.load('ngwlh/KBioXLM'+'/pytorch_model.bin', map_location='cpu')
+new_state_dict=OrderedDict()
+for k,v in checkpoint.items():
+    if k.startswith('model.roberta.'):
+        k=k.replace('model.roberta.','')
+    new_state_dict[k]=v
+msg = model.load_state_dict(new_state_dict, strict=False)
+logger.info(msg)
+```
+
+### BibTeX entry and citation info
+
+Coming soon.