AI Search
Re3Rank changes what looks at the candidates in attention-based reranking: one model first reads the question and candidates and writes a reasoning trace, then a second model looks with the trace's tokens and ranks candidates by the attention they receive.
AI Search
Letting a research agent read page screenshots, instead of first converting PDFs to text and searching that, raises question-answering accuracy and needs fewer searches.
AI Search
A search reranker can respect the requirements users attach to their queries without training any parameters: written for beginners, in a certain language, official sources only.
AI Search
When an LLM uses MCTS to find answers step by step over a knowledge base, intermediate steps can be given useful scores without training a reward model and without simulating to the end.
Personalization
There are a dozen or so ways for a CTR model to turn numerical features such as prices and counts into vectors, and in one fair comparison no single method wins everywhere.
Personalization
Ad CVR models are trained on conversion events such as purchases and installs, but conversions are rare, and about half of the training samples in this Meta model have no conversion label.
Personalization
A news publisher can ask an LLM to make a headline more eye-catching without changing a single fact, and the ranking still goes up.
Personalization
Shilling detection almost always reports a single AUROC, and that number cannot say when the detector can be trusted.
Infra
Prefill queueing under a prefix cache can keep both the median latency low and the tail short.
Infra
In learned sparse attention, as used by DeepSeek-V4 and GLM-5.3, each query attends to only k keys, but the small scoring network that picks those k keys still scores every key in the prefix.
Infra
A ten-minute traffic surge can leave an agent serving engine with prefill and decode disaggregated unable to recover.
Infra
For high-recall vector search on GPUs, graph search is held back by its chain of dependent rounds, not by distance computation itself.
Infra
When the KV cache runs out of room, vLLM and SGLang both evict from the end of cached sequences, to keep the opening prefix that many requests can share.
Infra
With a few changes, vLLM's existing batched LoRA kernel can run adapters trained with DoRA, (IA)³, VeRA and OFT exactly, in the same batch as LoRA adapters, without converting them to LoRA first.
Generative
On-policy distillation (OPD), where the student generates on its own and a teacher gives a full probability distribution at each position, makes generative recommenders train faster, but the student drifts further and further toward popular items.
Generative
This Kuaishou paper adds video utility to two places in generative recommendation that used to look only at content and probability.
Generative
In this CIKM paper from Alibaba's Taobao and Tmall Group, one large language model handles click retrieval, per-scenario retrieval, long-tail retrieval, interest exploration and retrieval with a search query at the same time, and overall does better than the multi-task retrieval models common in industry.
Generative
This Meta paper gives every creator its own token in the language model's vocabulary and uses it in place of the SID.
Generative
A generative recommender uses beam search to produce a SID one token at a time, and the usual way to miss fewer good results is to widen every step at once.
Generative
In SID generative recommendation, a user's newly emerging interest is often buried under a pile of old behavior, and every user shares one output transformation.
Embodied AI
When a humanoid robot reaches for a box, the box blocks its head depth camera as soon as the hands get close, yet the controller still has to finish the whole sequence of picking up, carrying and putting down.