Update README.md

README.md

@@ -815,7 +815,7 @@ dataset_info:
   download_size: 12941004
   dataset_size: 11452216
 ---
-# MERA (Multimodal Evaluation for Russian-language Architectures)
+# MERA v.1.2.0 (Multimodal Evaluation for Russian-language Architectures)
 
 ## Dataset Description
 
@@ -834,6 +834,8 @@ The benchmark covers 23 evaluation tasks comprising knowledge about the world, l
 
 NB that 8 datasets are diagnostic and not used in the overall model evaluation.
 
+**License and access.** All datasets are released under the MIT license. Train and Dev splits and the Public Tests (with ground-truth answers) are fully open. The ground-truth answers of the private Test splits are intentionally withheld to prevent data leakage and benchmark contamination; scoring on these splits is performed through the official MERA platform at https://mera.a-ai.ru, where participants submit model outputs and receive scores and leaderboard placement at no cost.
+
 ## MERA tasks & datasets
 
 1. [BPS: Balanced Parentheses Sequence](https://huggingface.co/datasets/MERA-evaluation/MERA#bps) (diagnostic)