ai-forever Samoed commited on
Commit
89fb165
1 Parent(s): ef258fd

Add prompts to config (#3)

Browse files

- add prompts (14917896b1fe196ef57efab5b1f0215673d5041f)
- Update config_sentence_transformers.json (79a7c744cfd0160a3988b824f0e409f23b053560)
- replace "retrieval" with "search" (dd0056fb02b3e4d3a506d12adc1ba811e01c3ae9)


Co-authored-by: Solomatin Roman <[email protected]>

Files changed (2) hide show
  1. README.md +21 -1
  2. config_sentence_transformers.json +10 -0
README.md CHANGED
@@ -1651,6 +1651,26 @@ print(sim_scores.diag().tolist())
1651
  # [0.47968706488609314, 0.940900444984436, 0.7761018872261047]
1652
  ```
1653
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1654
  ## Citation
1655
 
1656
  ```
@@ -1667,4 +1687,4 @@ print(sim_scores.diag().tolist())
1667
 
1668
  ## Limitations
1669
 
1670
- The model is designed to process texts in Russian, the quality in English is unknown. Maximum input text length is limited to 512 tokens.
 
1651
  # [0.47968706488609314, 0.940900444984436, 0.7761018872261047]
1652
  ```
1653
 
1654
+ or using prompts (sentence-transformers>=2.4.0):
1655
+
1656
+ ```python
1657
+ from sentence_transformers import SentenceTransformer
1658
+
1659
+
1660
+ # loads model with CLS pooling
1661
+ model = SentenceTransformer("ai-forever/ru-en-RoSBERTa")
1662
+
1663
+ classification = model.encode(["Он нам и <unk> не нужон ваш Интернет!", "What a time to be alive!"], prompt_name="classification")
1664
+ print(classification[0] @ classification[1].T) # 0.47968706488609314
1665
+
1666
+ clustering = model.encode(["В Ярославской области разрешили работу бань, но без посетителей", "Ярославским баням разрешили работать без посетителей"], prompt_name="clustering")
1667
+ print(clustering[0] @ clustering[1].T) # 0.940900444984436
1668
+
1669
+ query_embedding = model.encode("Сколько программистов нужно, чтобы вкрутить лампочку?", prompt_name="search_query")
1670
+ document_embedding = model.encode("Чтобы вкрутить лампочку, требуется три программиста: один напишет программу извлечения лампочки, другой — вкручивания лампочки, а третий проведет тестирование.", prompt_name="search_document")
1671
+ print(query_embedding @ document_embedding.T) # 0.7761018872261047
1672
+ ```
1673
+
1674
  ## Citation
1675
 
1676
  ```
 
1687
 
1688
  ## Limitations
1689
 
1690
+ The model is designed to process texts in Russian, the quality in English is unknown. Maximum input text length is limited to 512 tokens.
config_sentence_transformers.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "prompts": {
3
+ "classification": "classification: ",
4
+ "search_query": "search_query: ",
5
+ "search_document": "search_document: ",
6
+ "clustering": "clustering: "
7
+ },
8
+ "default_prompt_name": null,
9
+ "similarity_fn_name": null
10
+ }