Larger 100K vocabularies in SPLADE models, especially those initialized with ESPLADE pretraining, improve retrieval effectiveness after pruning compared to 32K baselines while keeping similar efficiency.
Sparsifying sparse representations for passage retrieval by top-kmasking.arXiv preprint arXiv:2112.09628,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
The Role of Vocabularies in Learning Sparse Representations for Ranking
Larger 100K vocabularies in SPLADE models, especially those initialized with ESPLADE pretraining, improve retrieval effectiveness after pruning compared to 32K baselines while keeping similar efficiency.