DS-Frame: slow-fast adaptive inference routes recommender compute by user difficulty
_reachsumit · x · 2026-09-03
DS-Frame is an adaptive fast-slow inference framework for sequential recommendation: a lightweight Fast System handles routine predictions, a Slow System performs iterative latent refinement, and a learned selector routes each sample under a controllable compute budget. Motivated by static recommenders degrading on hard user groups (long histories, niche profiles), it consistently improves multiple backbones across five real-world datasets, with larger gains on challenging groups. Code is open-sourced.
More from Research
- Linear transforms beat deep learning (ScanVI) for single-cell batch correction, preprint shows — arjunrajlab · 2026-09-03
- SimLoss Enables Single-Pass Fine-Grained Image Captioning at Multi-Stage Quality — Suryaansh Jain · 2026-09-03
- Tenstorrent and AI & Inc launch JapanFold: free inference for open-source drug discovery models — DavidBennett__ · 2026-09-03
- Until Labs Scales Cryoprotectant Search to 250,000 Molecules With AI — NirantK · 2026-09-03
- X Debate: Is Chain-of-Thought Prompting a Form of Parameter Reuse? — aryaman2020 · 2026-09-03
- Constraining agents with LL(1) grammar + structured diagnostics: what it fixes and what slips through — Upstairs-Special-925 · 2026-09-03