Research map / Label-efficient and robust learning / Recognition and long-tail
Recommender Systems with Generative Retrieval
Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Q. Tran, Jonah Samost, Maciej Kula, Ed H. Chi and 1 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| generative retrieval of Semantic IDs (TIGER) vs best dual-encoder baseline | sequential recommendation · Beauty | NDCG@5: +29.04% | Improved |
| generative retrieval (TIGER) vs best baseline | sequential recommendation · Sports and Outdoors | NDCG@10: 0.0204 -> 0.0225 | Improved |
| generative retrieval (TIGER) vs best baseline | sequential recommendation · Beauty | Recall@10: +0.15% | No clear effect |
| generative retrieval (TIGER) vs best baseline | sequential recommendation · Toys and Games | NDCG@5: +21.24% | Improved |
| use RQ-VAE Semantic IDs instead of random IDs (ablation) | TIGER · Sports and Outdoors | NDCG@10: 0.0063 -> 0.0225 | Improved |
| use RQ-VAE Semantic IDs instead of random IDs (ablation) | TIGER · Beauty | NDCG@10: 0.0250 -> 0.0384 | Improved |
| use RQ-VAE instead of LSH for Semantic IDs (ablation) | TIGER · Beauty | NDCG@10: 0.0309 -> 0.0384 | Improved |
| use RQ-VAE instead of LSH for Semantic IDs (ablation) | TIGER · Sports and Outdoors | Recall@5: 0.0215 -> 0.0264 | Improved |
| increase transformer layers 3 -> 4 -> 5 (ablation) | TIGER · Beauty | NDCG@10: 0.03768 -> 0.0384 -> 0.03834 | Mixed |
| add user ID token to input (ablation) | TIGER · Beauty | NDCG@10: 0.0367 -> 0.0384 | Improved |
| generate Semantic IDs on combined datasets instead of Beauty only (ablation) | TIGER · Beauty | NDCG@10: 0.0384 -> 0.03676 | Hurt |
| raise decoding temperature 1.0 -> 2.0 (ablation) | TIGER · Beauty | Entropy@10: 0.76 -> 1.38 | Improved |
Closest papers by meaning
- Sparse Meets Dense: Unified Generative Recommendations with Cascaded Sparse-Dense Representations (NeurIPS 2025 · similarity 0.88)
- Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations (ICML 2024 · similarity 0.87)
- Language Models as Semantic Indexers (ICML 2024 · similarity 0.87)
- Ada-Retrieval: An Adaptive Multi-Round Retrieval Paradigm for Sequential Recommendations (AAAI 2024 · similarity 0.86)
- Echoes in Filter Bubble: Diagnosing and Curing Popularity Bias in Generative Recommender Systems (NeurIPS 2026 · similarity 0.86)
- SRA-CL: Semantic Retrieval Augmented Contrastive Learning for Sequential Recommendation (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
