RecPFN: In-Context Recommendation via Synthetic Data Pretraining
_reachsumit · x · 2026-08-21
SAP presents RecPFN, a prior-fitted network bringing in-context learning to sequential recommendation. Pretrained entirely on synthetic clickstream data, it amortizes Bayesian inference from a small support set. It adapts to new domains in a single forward pass without weight updates, achieving SOTA zero-shot performance on eight benchmarks.
More from Research
- ShikharMurty: Pass@k gains may just sharpen distribution — ShikharMurty · 2026-08-21
- FasterFASTA tool achieves 1.69 GB/s with multithreaded BGZ decoding — viglovikov · 2026-08-21
- Princeton Paper: Legal Search Benchmarks Fail in Practice, New Dataset Released — burkov · 2026-08-21
- MIT Paper Finds Deleting Artist Data Doesn't Stop AI Recreating Images — technollama · 2026-08-21
- New journal to adopt GEB board, questioning value of legacy publishers — Afinetheorem · 2026-08-21
- HarnessEval-W: An Agentic Benchmark for World Models Evaluation — 青稞AI · 2026-08-21