SimCIT replaces reconstruction-based item tokenization with a contrastive loss and soft residual quantization, improving generative recommendation accuracy on five datasets.
Sequence to sequence learning with neural networks,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Simple Contrastive Framework Of Item Tokenization For Generative Recommendation
SimCIT replaces reconstruction-based item tokenization with a contrastive loss and soft residual quantization, improving generative recommendation accuracy on five datasets.