Token-level Ensembling of Models with Different Vocabularies

• URL: http://arxiv.org/abs/2502.21265v1
• Date: Fri, 28 Feb 2025 17:41:27 GMT
• Title: Token-level Ensembling of Models with Different Vocabularies
• Authors: Rachel Wicks, Kartik Ravisankar, Xinchen Yang, Philipp Koehn, Matt Post,
• Abstract summary: Model ensembling is a technique to combine the predicted distributions of two or more models.<n>This paper proposes an inference-time only algorithm that allows for ensembling models with different vocabularies.
• Score: 16.094010998574753
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Model ensembling is a technique to combine the predicted distributions of two or more models, often leading to improved robustness and performance. For ensembling in text generation, the next token's probability distribution is derived from a weighted sum of the distributions of each individual model. This requires the underlying models to share the same subword vocabulary, limiting the applicability of ensembling, since many open-sourced models have distinct vocabularies. In research settings, experimentation or upgrades to vocabularies may introduce multiple vocabulary sizes. This paper proposes an inference-time only algorithm that allows for ensembling models with different vocabularies, without the need to learn additional parameters or alter the underlying models. Instead, the algorithm ensures that tokens generated by the ensembled models \textit{agree} in their surface form. We apply this technique to combinations of traditional encoder-decoder models and decoder-only LLMs and evaluate on machine translation. In addition to expanding to model pairs that were previously incapable of token-level ensembling, our algorithm frequently improves translation performance over either model individually.

Related papers

• Can bidirectional encoder become the ultimate winner for downstream applications of foundation models? [1.8120356834558644]
  Foundational models have the characteristics of pre-training, transfer learning, and self-supervised learning.<n>BERT broke through the limitation of only using one-way methods for language modeling in pre-training by using a masked language model.<n>This article analyzes one-way and bidirectional models based on GPT and BERT and compares their differences based on the purpose of the model.
  arXiv  Detail & Related papers  (2024-11-27T03:31:14Z)
• FUSE-ing Language Models: Zero-Shot Adapter Discovery for Prompt Optimization Across Tokenizers [55.2480439325792]
  We propose FUSE, an approach to approximating an adapter layer that maps from one model's textual embedding space to another, even across different tokenizers. We show the efficacy of our approach via multi-objective optimization over vision-language and causal language models for image captioning and sentiment-based image captioning.
  arXiv  Detail & Related papers  (2024-08-09T02:16:37Z)
• Promises and Pitfalls of Generative Masked Language Modeling: Theoretical Framework and Practical Guidelines [74.42485647685272]
  We focus on Generative Masked Language Models (GMLMs) We train a model to fit conditional probabilities of the data distribution via masking, which are subsequently used as inputs to a Markov Chain to draw samples from the model. We adapt the T5 model for iteratively-refined parallel decoding, achieving 2-3x speedup in machine translation with minimal sacrifice in quality.
  arXiv  Detail & Related papers  (2024-07-22T18:00:00Z)
• CharED: Character-wise Ensemble Decoding for Large Language Models [24.993790740335243]
  We present an inference-time ensembling algorithm aimed at "averaging" outputs from multiple large language models. Our proposed model is able to combine complimentary strengths of multiple LLMs, regardless of vocabulary, tokenization, or model size.
  arXiv  Detail & Related papers  (2024-06-25T22:35:07Z)
• Artificial Interrogation for Attributing Language Models [0.0]
  The challenge provides twelve open-sourced base versions of popular language models and twelve fine-tuned language models for text generation. The goal of the contest is to identify which fine-tuned models originated from which base model. We have employed four distinct approaches for measuring the resemblance between the responses generated from the models of both sets.
  arXiv  Detail & Related papers  (2022-11-20T05:46:29Z)
• Hierarchical Phrase-based Sequence-to-Sequence Learning [94.10257313923478]
  We describe a neural transducer that maintains the flexibility of standard sequence-to-sequence (seq2seq) models while incorporating hierarchical phrases as a source of inductive bias during training and as explicit constraints during inference. Our approach trains two models: a discriminative derivation based on a bracketing grammar whose tree hierarchically aligns source and target phrases, and a neural seq2seq model that learns to translate the aligned phrases one-by-one.
  arXiv  Detail & Related papers  (2022-11-15T05:22:40Z)
• Language Model Cascades [72.18809575261498]
  Repeated interactions at test-time with a single model, or the composition of multiple models together, further expands capabilities. Cases with control flow and dynamic structure require techniques from probabilistic programming. We formalize several existing techniques from this perspective, including scratchpads / chain of thought, verifiers, STaR, selection-inference, and tool use.
  arXiv  Detail & Related papers  (2022-07-21T07:35:18Z)
• Twist Decoding: Diverse Generators Guide Each Other [116.20780037268801]
  We introduce Twist decoding, a simple and general inference algorithm that generates text while benefiting from diverse models. Our method does not assume the vocabulary, tokenization or even generation order is shared.
  arXiv  Detail & Related papers  (2022-05-19T01:27:53Z)
• Introducing various Semantic Models for Amharic: Experimentation and Evaluation with multiple Tasks and Datasets [19.855120632909124]
  We introduce different semantic models for Amharic. Models are build using word2Vec embeddings, distributional thesaurus (DT), contextual embeddings, and DT embeddings. We find that newly trained models perform better than pre-trained multilingual models.
  arXiv  Detail & Related papers  (2020-11-02T17:48:25Z)
• Grounded Compositional Outputs for Adaptive Language Modeling [59.02706635250856]
  A language model's vocabulary$-$typically selected before training and permanently fixed later$-$affects its size. We propose a fully compositional output embedding layer for language models. To our knowledge, the result is the first word-level language model with a size that does not depend on the training vocabulary.
  arXiv  Detail & Related papers  (2020-09-24T07:21:14Z)

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.