ByteDance's X-Rec does generative retrieval via flow matching with 3.46x throughput
_reachsumit · x · 2026-09-25
ByteDance's X-Rec technical report models recommendation as flow matching in continuous item embedding space, generating multiple retrieval triggers in parallel instead of quantizing items into discrete semantic IDs.
- Anchor conditioning decomposes generation into coarse semantic-region selection and fine-grained refinement
- Riemannian flow matching aligns trajectories with the hyperspherical geometry of item embeddings
- A late-interaction diffusion Transformer restricts repeated velocity-field estimation to the final layer
On a streaming benchmark it substantially outperforms U2I baselines, matches SID-AR retrieval quality, and delivers 3.46x higher inference throughput.
More from Research
- PosteriorBench accepted to NeurIPS: point reconstruction accuracy misleads on inverse solvers — AnimaAnandkumar · 2026-09-25
- Dietterich: pretraining's next-token task already forces models to abstract and generalize — tdietterich · 2026-09-25
- Neural Spectral Capacity: a training-free architecture score that beats #Params and prunes LLaMA-7B 5900x faster — CityU-HongKong · 2026-09-25
- HRI 2027 Adds Archival Industry White Paper Track for Real-World Robot Deployments — petitegeek · 2026-09-25
- Study: Reasoning hurts 15.7% of multimodal embeddings; training-free SURE router fixes it — _reachsumit · 2026-09-25
- Google scales learned cross-task relationships in YouTube's production recommender — _reachsumit · 2026-09-25