Detailed sentence bert model for the MAG240M dataset

[tjb-tech]

Hi,the team of ogb. Thanks for your efforts on collecting and processing these valuable datasets! You said you used a RoBERTa sentence encoder to generate a 768-dimensional vector for each paper node in the MAG240M dataset. May I ask which RoBERTa sentence encoder you used? Could you give me a specific huggingface link? Best regards!

@weihua916 @rusty1s

[weihua916]

https://huggingface.co/sentence-transformers/roberta-base-nli-stsb-mean-tokens

from sentence_transformers import SentenceTransformer
SentenceTransformer('roberta-base-nli-stsb-mean-tokens')

[tjb-tech]

mean

Thank you very much for the quick reply! Because this model is deprecated, do you consider to provide the original feature processing script for the MAG240M paper nodes? So that we can process the feature with more powerful sentence bert model🤗 Best regards!