Hit the Sweet Spot! Span-Level Ensemble for Large Language Models

• URL: http://arxiv.org/abs/2409.18583v1
• Date: Fri, 27 Sep 2024 09:41:29 GMT
• Title: Hit the Sweet Spot! Span-Level Ensemble for Large Language Models
• Authors: Yangyifan Xu, Jianghao Chen, Junhong Wu, Jiajun Zhang,
• Abstract summary: We propose SweetSpan, a span-level ensemble method that effectively balances the need for real-time adjustments and the information required for accurate ensemble decisions. Our approach involves two key steps: First, we have each candidate model independently generate candidate spans based on the shared prefix. Second, we calculate perplexity scores to facilitate mutual evaluation among the candidate models and achieve robust span selection by filtering out unfaithful scores.
• Score: 8.34562564266839
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Ensembling various LLMs to unlock their complementary potential and leverage their individual strengths is highly valuable. Previous studies typically focus on two main paradigms: sample-level and token-level ensembles. Sample-level ensemble methods either select or blend fully generated outputs, which hinders dynamic correction and enhancement of outputs during the generation process. On the other hand, token-level ensemble methods enable real-time correction through fine-grained ensemble at each generation step. However, the information carried by an individual token is quite limited, leading to suboptimal decisions at each step. To address these issues, we propose SweetSpan, a span-level ensemble method that effectively balances the need for real-time adjustments and the information required for accurate ensemble decisions. Our approach involves two key steps: First, we have each candidate model independently generate candidate spans based on the shared prefix. Second, we calculate perplexity scores to facilitate mutual evaluation among the candidate models and achieve robust span selection by filtering out unfaithful scores. To comprehensively evaluate ensemble methods, we propose a new challenging setting (ensemble models with significant performance gaps) in addition to the standard setting (ensemble the best-performing models) to assess the performance of model ensembles in more realistic scenarios. Experimental results in both standard and challenging settings across various language generation tasks demonstrate the effectiveness, robustness, and versatility of our approach compared with previous ensemble methods.

Related papers

• An incremental preference elicitation-based approach to learning potentially non-monotonic preferences in multi-criteria sorting [53.36437745983783]
  We first construct a max-margin optimization-based model to model potentially non-monotonic preferences. We devise information amount measurement methods and question selection strategies to pinpoint the most informative alternative in each iteration. Two incremental preference elicitation-based algorithms are developed to learn potentially non-monotonic preferences.
  arXiv  Detail & Related papers  (2024-09-04T14:36:20Z)
• One-Shot Learning as Instruction Data Prospector for Large Language Models [108.81681547472138]
  textscNuggets uses one-shot learning to select high-quality instruction data from extensive datasets. We show that instruction tuning with the top 1% of examples curated by textscNuggets substantially outperforms conventional methods employing the entire dataset.
  arXiv  Detail & Related papers  (2023-12-16T03:33:12Z)
• Universal Semi-supervised Model Adaptation via Collaborative Consistency Training [92.52892510093037]
  We introduce a realistic and challenging domain adaptation problem called Universal Semi-supervised Model Adaptation (USMA) We propose a collaborative consistency training framework that regularizes the prediction consistency between two models. Experimental results demonstrate the effectiveness of our method on several benchmark datasets.
  arXiv  Detail & Related papers  (2023-07-07T08:19:40Z)
• Probabilistic Bilevel Coreset Selection [24.874967723659022]
  We propose a continuous probabilistic bilevel formulation of coreset selection by learning a probablistic weight for each training sample. We develop an efficient solver to the bilevel optimization problem via unbiased policy gradient without trouble of implicit differentiation.
  arXiv  Detail & Related papers  (2023-01-24T09:37:00Z)
• Differentiable Model Selection for Ensemble Learning [37.99501959301896]
  This paper proposes a novel framework for differentiable model selection integrating machine learning and optimization. The framework is tailored for ensemble learning, a strategy that combines the outputs of individually pre-trained models, and learns to select appropriate ensemble members for a particular input sample.
  arXiv  Detail & Related papers  (2022-11-01T03:37:49Z)
• Model ensemble instead of prompt fusion: a sample-specific knowledge transfer method for few-shot prompt tuning [85.55727213502402]
  We focus on improving the few-shot performance of prompt tuning by transferring knowledge from soft prompts of source tasks. We propose Sample-specific Ensemble of Source Models (SESoM) SESoM learns to adjust the contribution of each source model for each target sample separately when ensembling source model outputs.
  arXiv  Detail & Related papers  (2022-10-23T01:33:16Z)
• A Lagrangian Duality Approach to Active Learning [119.36233726867992]
  We consider the batch active learning problem, where only a subset of the training data is labeled. We formulate the learning problem using constrained optimization, where each constraint bounds the performance of the model on labeled samples. We show, via numerical experiments, that our proposed approach performs similarly to or better than state-of-the-art active learning methods.
  arXiv  Detail & Related papers  (2022-02-08T19:18:49Z)
• A New Approach to Overgenerating and Scoring Abstractive Summaries [9.060597430218378]
  We propose a two-staged strategy to generate a diverse set of candidate summaries from the source text in stage one, then score and select admissible ones in stage two. Our generator gives a precise control over the length of the summary, which is especially well-suited when space is limited. Our selectors are designed to predict the optimal summary length and put special emphasis on faithfulness to the original text.
  arXiv  Detail & Related papers  (2021-04-05T00:29:45Z)
• Towards Model-Agnostic Post-Hoc Adjustment for Balancing Ranking Fairness and Algorithm Utility [54.179859639868646]
  Bipartite ranking aims to learn a scoring function that ranks positive individuals higher than negative ones from labeled data. There have been rising concerns on whether the learned scoring function can cause systematic disparity across different protected groups. We propose a model post-processing framework for balancing them in the bipartite ranking scenario.
  arXiv  Detail & Related papers  (2020-06-15T10:08:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.