Update README.md

README.md

@@ -24,10 +24,10 @@ size_categories:
 This is the dataset for two papers: **Quality and Quantity of Machine Translation References for Automated Metrics [[paper](https://arxiv.org/abs/2401.01283)]** - effect of reference quality and quantity on automatic metric performance, and **Evaluating Optimal Reference Translations [[paper]](https://arxiv.org/abs/2311.16787)** - creation of the data and human aspects of annotation and translation.
 Please see the [original repository](https://github.com/ufal/optimal-reference-translations) for more information or [contact the authors](mailto:vilem.zouhar@gmail.com) with any questions.
 
-You need to stream the dataset for it to work because `ort_human` and `ort_wmt` splits have different columns:
+Please make sure that you have the latest `datasets` installed:
 ```
-data_human = list(load_dataset("zouharvi/optimal-reference-translations", 'ort_human', streaming=True)["train"])
-data_wmt = list(load_dataset("zouharvi/optimal-reference-translations", 'ort_wmt', streaming=True)["train"])
+data_human = load_dataset("zouharvi/optimal-reference-translations", 'ort_human')["train"]
+data_wmt = load_dataset("zouharvi/optimal-reference-translations", 'ort_wmt')["train"]
 ```
 
 # Quality and Quantity of Machine Translation References for Automated Metrics [[paper](https://arxiv.org/abs/2401.01283)]
@@ -75,7 +75,7 @@ Note: If you you also want to use the WMT2020 system submissions, please contact
 
 ```python3
 from datasets import load_dataset
-data = list(load_dataset("zouharvi/optimal-reference-translations", 'ort_human', streaming=True)["train"])
+data = load_dataset("zouharvi/optimal-reference-translations", 'ort_human')["train"]
 
 # 220 annotated documents
 len(data)