analiza aiqbench
FlashEvaluator introduces a novel joint evaluation framework that scores multiple candidate sequences simultaneously using QKV-Cache mechanisms inspired by autoregressive caching. This approach significantly reduces inference latency and increases throughput in recommendation systems and text summarization without compromising model quality. The method effectively mitigates the linear growth of computational costs associated with scaling the number of candidates evaluated per request.
FlashEvaluator: Expanding Search Space with Parallel Sequence-Level Evaluation
arXiv:2603.02565v2 Announce Type: replace-cross Abstract: The Generator-Evaluator (G-E) framework generates K candidate sequences and uses an evaluator to select the highest-scoring one, which is widely used in recommender systems (RecSys) and natural language processing (NLP). Existing evaluators commonly score candidates independently. Although such evaluations can be batched, independent scoring neither models interactions among candidates nor eliminates repeated computation of request-level context and recurring candidate elements, causing the total evaluation work to grow approximately linearly with K. To handle with, we propose FlashEvaluator, a joint evaluator that scores all candidate sequences in a single forward pass. FlashEvaluator factorizes evaluation into shared request-level encoding, reusable candidate-side computation, sequence assembly by indexing, and cross-sequence interaction for setwise comparison. We call this request-local reuse scheme QKV-Cache: inspired by autoregressive KV caching, it reuses context-side key/value representations across candidate sequences and, when candidate elements recur, reuses their request-conditioned representations on the query side. In repeated-item settings, the dominant item-encoding cost therefore depends on the number of distinct items rather than their total occurrences across sequences, reducing the marginal cost of evaluating additional candidates. We provide a computational analysis and evaluate FlashEvaluator on recommendation and text summarization. The results show lower latency and higher throughput with competitive recommendation and summarization quality. In an online deployment at Kuaishou with K=50, FlashEvaluator reduces inference latency by 44% and increases QPS by 114% relative to the production baseline, while yielding statistically significant gains in retention, engagement, and ecosystem metrics.
Abstract:The Generator-Evaluator (G-E) framework generates K candidate sequences and uses an evaluator to select the highest-scoring one, which is widely used in recommender systems (RecSys) and natural language processing (NLP). Existing evaluators commonly score candidates independently. Although such evaluations can be batched, independent scoring neither models interactions among candidates nor eliminates repeated computation of request-level context and recurring candidate elements, causing the total evaluation work to grow approximately linearly with K. To handle with, we propose FlashEvaluator, a joint evaluator that scores all candidate sequences in a single forward pass. FlashEvaluator factorizes evaluation into shared request-level encoding, reusable candidate-side computation, sequence assembly by indexing, and cross-sequence interaction for setwise comparison. We call this request-local reuse scheme QKV-Cache: inspired by autoregressive KV caching, it reuses context-side key/value representations across candidate sequences and, when candidate elements recur, reuses their request-conditioned representations on the query side. In repeated-item settings, the dominant item-encoding cost therefore depends on the number of distinct items rather than their total occurrences across sequences, reducing the marginal cost of evaluating additional candidates. We provide a computational analysis and evaluate FlashEvaluator on recommendation and text summarization. The results show lower latency and higher throughput with competitive recommendation and summarization quality. In an online deployment at Kuaishou with K=50, FlashEvaluator reduces inference latency by 44% and increases QPS by 114% relative to the production baseline, while yielding statistically significant gains in retention, engagement, and ecosystem metrics.
Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG)
arXiv:2603.02565 [cs.IR]
(or arXiv:2603.02565v2 [cs.IR] for this version)
https://doi.org/10.48550/arXiv.2603.02565 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 334 cuvinte
Subiecte conexe
From Vector Search to Agentic RAG: Building an Enterprise Research Analyst with hipVS
AMD · 15 iul. 2026
NVIDIA CUDA 13.3 Enhances GPU Development with Tile Programming in C++, Compiler Autotuning, and Python Updates
NVIDIA · 26 mai 2026
Synthesis Superintelligence: from Semiconductors to Superconductors — Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk
Latent Space · 8 oct. 2026
From CUDA to MLX: How K-Search Brings Decades of Kernel Expertise to Apple Silicon
BAIR Blog · 29 iul. 2026