Update README.md

README.md

@@ -17,4 +17,65 @@ tags:
   - transformers
 widget:
 - text: Метод опорных векторов
----
+---
+SciRus-tiny is the model to obtain embeddings of scientific texts in russian and english. Model was trained on [eLibrary](https://www.elibrary.ru/) data with contrastive technics described in [habr post]() and achive quite good metrics values on the [ruSciBench](https://github.com/mlsa-iai-msu-lab/ru_sci_bench/tree/main) benchmark.
+
+### How to get embeddings
+
+```python
+from transformers import AutoTokenizer, AutoModel
+import torch
+import torch.nn.functional as F
+
+
+tokenizer = AutoTokenizer.from_pretrained("mlsa-iai-msu-lab/sci-rus-tiny")
+model = AutoModel.from_pretrained("mlsa-iai-msu-lab/sci-rus-tiny")
+# model.cuda()  # if you want to use a GPU
+
+def mean_pooling(model_output, attention_mask):
+    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
+    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
+    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
+
+
+def get_sentence_embedding(title, abstract, model, tokenizer, max_length=None):
+    # Tokenize sentences
+    sentence = '</s>'.join([title, abstract])
+    encoded_input = tokenizer(
+        [sentence], padding=True, truncation=True, return_tensors='pt', max_length=max_length).to(model.device)
+
+    # Compute token embeddings
+    with torch.no_grad():
+        model_output = model(**encoded_input)
+
+    # Perform pooling
+    sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
+    
+    # Normalize embeddings
+    sentence_embeddings = F.normalize(sentence_embeddings, p=2, dim=1)
+    return sentence_embeddings.cpu().detach().numpy()[0]
+
+print(get_sentence_embedding('some title', 'some abstract', model, tokenizer).shape)
+# (312,)
+```
+
+Or you can use the `sentence_transformers`:
+```Python
+from sentence_transformers import SentenceTransformer
+
+
+model = SentenceTransformer('mlsa-iai-msu-lab/sci-rus-tiny')
+embeddings = model.encode(['привет мир'])
+print(embeddings[0].shape)
+# (312,)
+```
+
+
+### Authors
+Benchmark developed by MLSA Lab of Institute for AI, MSU.
+
+### Acknowledgement
+We would like to thank [eLibrary](https://elibrary.ru/) for provided datasets.
+
+### Contact
+Nikolai Gerasimenko (nikgerasimenko@gmail.com).