SPLADE models produce wacky expansion terms whose prevalence rises with larger vocabularies and falls with stricter sparsity; these terms primarily aid in-domain retrieval rather than out-of-domain generalization.
Wacky weights in learned sparse representations and the revenge of score-at-a-time query evaluation.arXiv preprint arXiv:2110.11540
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.IR 3representative citing papers
Replacing SPLADE's MLM vocabulary with SAE-learned semantic concepts achieves comparable retrieval performance with improved efficiency.
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.
citing papers explorer
-
Understanding Wacky Weights: A Dissection of SPLADE's Learned Term Importance
SPLADE models produce wacky expansion terms whose prevalence rises with larger vocabularies and falls with stricter sparsity; these terms primarily aid in-domain retrieval rather than out-of-domain generalization.
-
From Tokens to Concepts: Leveraging SAE for SPLADE
Replacing SPLADE's MLM vocabulary with SAE-learned semantic concepts achieves comparable retrieval performance with improved efficiency.
-
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.