FUSE-ing Language Models: Zero-Shot Adapter Discovery for Prompt Optimization Across Tokenizers

• URL: http://arxiv.org/abs/2408.04816v1
• Date: Fri, 9 Aug 2024 02:16:37 GMT
• Title: FUSE-ing Language Models: Zero-Shot Adapter Discovery for Prompt Optimization Across Tokenizers
• Authors: Joshua Nathaniel Williams, J. Zico Kolter,
• Abstract summary: 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.
• Score: 55.2480439325792
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The widespread use of large language models has resulted in a multitude of tokenizers and embedding spaces, making knowledge transfer in prompt discovery tasks difficult. In this work, we propose FUSE (Flexible Unification of Semantic Embeddings), an inexpensive approach to approximating an adapter layer that maps from one model's textual embedding space to another, even across different tokenizers. We introduce a third-order tensor-based representation of a model's embedding space that aligns semantic embeddings that have been split apart by different tokenizers, and use this representation to derive an approximation of the gradient of one model's outputs with respect to another model's embedding space. 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.

Related papers

• Towards Retrieval-Augmented Architectures for Image Captioning [81.11529834508424]
  This work presents a novel approach towards developing image captioning models that utilize an external kNN memory to improve the generation process. Specifically, we propose two model variants that incorporate a knowledge retriever component that is based on visual similarities. We experimentally validate our approach on COCO and nocaps datasets and demonstrate that incorporating an explicit external memory can significantly enhance the quality of captions.
  arXiv  Detail & Related papers  (2024-05-21T18:02:07Z)
• Fine-tuning CLIP Text Encoders with Two-step Paraphrasing [83.3736789315201]
  We introduce a straightforward fine-tuning approach to enhance the representations of CLIP models for paraphrases. Our model, which we call ParaCLIP, exhibits significant improvements over baseline CLIP models across various tasks.
  arXiv  Detail & Related papers  (2024-02-23T06:11:50Z)
• Subobject-level Image Tokenization [60.80949852899857]
  Transformer-based vision models typically tokenize images into fixed-size square patches as input units. Inspired by the subword tokenization widely adopted in language models, we propose an image tokenizer at a subobject level.
  arXiv  Detail & Related papers  (2024-02-22T06:47:44Z)
• Generalizable Entity Grounding via Assistance of Large Language Model [77.07759442298666]
  We propose a novel approach to densely ground visual entities from a long caption. We leverage a large multimodal model to extract semantic nouns, a class-a segmentation model to generate entity-level segmentation, and a multi-modal feature fusion module to associate each semantic noun with its corresponding segmentation mask.
  arXiv  Detail & Related papers  (2024-02-04T16:06:05Z)
• Prompting Large Vision-Language Models for Compositional Reasoning [12.908633583017359]
  We develop a novel generative method that prompts large vision-language models to depict images and perform compositional reasoning. Our method outperforms other embedding-based methods on the Winoground dataset, and obtains further improvement of up to 10% accuracy when enhanced with the optimal description.
  arXiv  Detail & Related papers  (2024-01-20T22:04:28Z)
• User-Aware Prefix-Tuning is a Good Learner for Personalized Image Captioning [35.211749514733846]
  Traditional image captioning methods often overlook the preferences and characteristics of users. Most existing methods emphasize the user context fusion process by memory networks or transformers. We propose a novel personalized image captioning framework that leverages user context to consider personality factors.
  arXiv  Detail & Related papers  (2023-12-08T02:08:00Z)
• Self-Supervised Open-Ended Classification with Small Visual Language Models [60.23212389067007]
  We present Self-Context Adaptation (SeCAt), a self-supervised approach that unlocks few-shot abilities for open-ended classification with small visual language models. By using models with approximately 1B parameters we outperform the few-shot abilities of much larger models, such as Frozen and FROMAGe.
  arXiv  Detail & Related papers  (2023-09-30T21:41:21Z)
• Guiding Attention using Partial-Order Relationships for Image Captioning [2.620091916172863]
  A guided attention network mechanism exploits the relationship between the visual scene and text-descriptions. A pairwise ranking objective is used for training this embedding space which allows similar images, topics and captions in the shared semantic space. The experimental results based on MSCOCO dataset shows the competitiveness of our approach.
  arXiv  Detail & Related papers  (2022-04-15T14:22:09Z)
• Fusion Models for Improved Visual Captioning [18.016295296424413]
  This paper proposes a generic multimodal model fusion framework for caption generation and emendation. We employ the same fusion strategies to integrate a pretrained Masked Language Model (MLM) with a visual captioning model, viz. Show, Attend, and Tell. Our caption emendation experiments on three benchmark image captioning datasets, viz. Flickr8k, Flickr30k, and MSCOCO, show improvements over the baseline.
  arXiv  Detail & Related papers  (2020-10-28T21:55:25Z)

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.