mlsanityData to 14 Oct 2026

RASTP: Representation-Aware Semantic Token Pruning for Generative Recommendation with Semantic Identifiers

arXiv 2511.16943 ↗First posted 21 Nov 2025cs.IR cs.AIcited by 0

On the map: Efficient LLM inference and serving → Visual token pruning in multimodal models

Abstract (arXiv)

Generative recommendation systems typically leverage Semantic Identifiers (SIDs), which represent each item as a sequence of tokens that encode semantic information. However, representing item ID with multiple SIDs significantly increases input sequence length, which is a major determinant of computational complexity and memory consumption. While existing efforts primarily focus on optimizing attention computation and KV cache, we propose RASTP (Representation-Aware Semantic Token Pruning), which directly prunes less informative tokens in the input sequence. Specifically, RASTP evaluates token importance by combining semantic saliency, measured via representation magnitude, and attention centrality, derived from cumulative attention weights. Since RASTP dynamically prunes low-information or irrelevant semantic tokens, experiments on three real-world Amazon datasets show that RASTP reduces training time by 26.7%, while maintaining or slightly improving recommendation performance. The code has been open-sourced at https://github.com/Yuzt-zju/RASTP.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Visual token pruning in multimodal models, one of 12 directions in Efficient LLM inference and serving. Builds on 9 of its 13 references that are in the map. First posted on arXiv 21 Nov 2025.

Builds on

9 of 13 references.

References · 9

  1. 2022Autoregressive Image Generation using Residual Quantizationcited by 722
  2. 2024Attention Score is not All You Need for Token Importance Indicator in KV Cache Reduction: Value Also Matterscited by 73
  3. 2024IDGenRec: LLM-RecSys Alignment with Textual ID Learningcited by 53
  4. 2025Sparse Meets Dense: Unified Generative Recommendations with Cascaded Sparse-Dense Representationscited by 35
  5. 2025FORGE: Forming Semantic Identifiers for Generative Retrieval in Industrial Datasetscited by 17
  6. 2025Collaboration of Large Language Models and Small Recommendation Models for Device-Cloud Recommendationcited by 15
  7. 2025ThinkRec: Thinking-based recommendation via LLMcited by 11
  8. 2025EARN: Efficient Inference Acceleration for LLM-based Generative Recommendation by Register Tokenscited by 6
  9. 2025FLAME: A Serving System Optimized for Large-Scale Generative Recommendation with Efficiencycited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026PRISM: Purified Representation and Integrated Semantic Modeling for Generative Sequential Recommendation
  2. 2025Semantics Meet Signals: Dual Codebook Representationl Learning for Generative Recommendation
  3. 2026Learning Variable-Length Tokenization for Generative Recommendation
  4. 2025A Simple Contrastive Framework Of Item Tokenization For Generative Recommendation
  5. 2026Preserving Item Semantics for Free: Rethinking Token Initialization in LLM-Based Generative Recommendation
  6. 2026Conditional Memory Enhanced Item Representation for Generative Recommendation
  7. 2025LLaDA-Rec: Discrete Diffusion for Parallel Semantic ID Generation in Generative Recommendation
  8. 2026GenRec: A Preference-Oriented Generative Framework for Large-Scale Recommendation