paper-with-me

홈 › Papers

Online Cascade Learning for Efficient Inference over Streams

2024-02-07 · Lunyiu Nie, Zhimin Ding, Erdong Hu, Christopher Jermaine, Swarat Chaudhuri

Large Language Models (LLMs) have a natural role in answering complex queries about data streams, but the high computational cost of LLM inference makes them infeasible in many such tasks. We propose online cascade learning, the first approach to address this challenge. The objective here is to learn a "cascade" of models, starting with lower-capacity models (such as logistic regression) and ending with a powerful LLM, along with a deferral policy that determines the model to be used on a given input. We formulate the task of learning cascades online as an imitation-learning problem, where smaller models are updated over time imitating the collected LLM demonstrations, and give a no-regret algorithm for the problem. Experimental results across four benchmarks show that our method parallels LLMs in accuracy while cutting down inference costs by as much as 90% with strong robustness against input distribution shifts, underscoring its efficacy and adaptability in stream processing.

📄 PDF Abstract BibTeX arXiv:2402.04513

Code (1)

flitternie/online_cascade_learning 공식 구현

Tasks

Imitation Learning

Similar Papers 제목 키워드 기반

CascadeServe: Unlocking Model Cascades for Inference Serving

2024-06-20 · Ferdi Kossmann, Ziniu Wu, Alex Turk, Nesime Tatbul 외

Machine learning (ML) models are increasingly deployed to production, calling for efficient inference serving systems. Efficient inference serving is complicated by two challenges: (i) ML models incur high computational …

modelScheduling

A Circular Window-based Cascade Transformer for Online Action Detection

2022-08-30 · Shuqiang Cao, Weixin Luo, Bairui Wang, Wei zhang 외

Online action detection aims at the accurate action prediction of the current frame based on long historical observations. Meanwhile, it demands real-time inference on online streaming videos. In this paper, we advocate …

Action DetectionAction SegmentationOnline Action Detection

Online Estimation and Community Detection of Network Point Processes for Event Streams

2020-09-03 · Guanhua Fang, Owen G. Ward, Tian Zheng

A common goal in network modeling is to uncover the latent community structure present among nodes. For many real-world networks, the true connections consist of events arriving as streams, which are then aggregated to f…

Community DetectionOnline Community DetectionPoint ProcessesVariational Inference

From Deferral to Learning: Online In-Context Knowledge Distillation for LLM Cascades

2025-09-26 · Yu Wu, Shuo Wu, Ye Tao, Yansong Li 외 arxiv

Standard LLM cascades improve efficiency by deferring difficult queries from weak to strong models. However, these systems are typically static: when faced with repeated or semantically similar queries, they redundantly …

Knowledge Distillation

Continual Learning with Optimal Transport based Mixture Model

2022-11-30 · Quyen Tran, Hoang Phan, Khoat Than, Dinh Phung 외

Online Class Incremental learning (CIL) is a challenging setting in Continual Learning (CL), wherein data of new tasks arrive in incoming streams and online learning models need to handle incoming data streams without re…

class-incremental learningClass Incremental LearningContinual LearningIncremental Learning+1