Skip to Main Content
Table 6: 
SARI scores on the DiscoFuse dataset when experimenting with different embedding setups. Each row also includes the setups of all previous rows.
 bertShare gpt 
DiscoFuse 89.3 88.0 
- embeddings from checkpoint 87.5 77.0 
+ task specific SentencePieces 87.5 84.2 
+ pre-training SentencePieces 88.0 69.7 
 bertShare gpt 
DiscoFuse 89.3 88.0 
- embeddings from checkpoint 87.5 77.0 
+ task specific SentencePieces 87.5 84.2 
+ pre-training SentencePieces 88.0 69.7 
Close Modal

or Create an Account

Close Modal
Close Modal