Update README.md

README.md

@@ -9,13 +9,11 @@ tags:
 - e-commerce
 - encoder
 ---
-Encoder-model for e-commerce search query similarity task. The search queries are short.
+A sentencepiece tokenizer was applied to a corpus of 269 million Russian search queries. 
 
-Sentencepiece tokenizer fitted on 269 million Russian search queries log.
+The encoder-model was trained for the e-commerce search query similarity task, and the search queries were short.
 
-Short sequence length is used to reduce the memory footprint.
-
-The dataset for validation with manual markup consisted of 362 thousand examples.
+The dataset for validation, which was manually annotated, comprised 362,000 instances.
 
 ![Validation results](https://huggingface.co/fkrasnov2/SBE/resolve/main/bvf_recall1k_query_len_eng.svg)