paper-with-me

홈 › Papers

On the Generalization Ability of Retrieval-Enhanced Transformers

2023-02-23 · Tobias Norlund, Ehsan Doostmohammadi, Richard Johansson, Marco Kuhlmann

Recent work on the Retrieval-Enhanced Transformer (RETRO) model has shown that off-loading memory from trainable weights to a retrieval database can significantly improve language modeling and match the performance of non-retrieval models that are an order of magnitude larger in size. It has been suggested that at least some of this performance gain is due to non-trivial generalization based on both model weights and retrieval. In this paper, we try to better understand the relative contributions of these two components. We find that the performance gains from retrieval largely originate from overlapping tokens between the database and the test data, suggesting less non-trivial generalization than previously assumed. More generally, our results point to the challenges of evaluating the generalization of retrieval-augmented language models such as RETRO, as even limited token overlap may significantly decrease test-time loss. We release our code and model at https://github.com/TobiasNorlund/retro

📄 PDF Abstract BibTeX arXiv:2302.12128

Code (0)

등록된 구현이 없습니다.

Tasks

Language ModelingLanguage ModellingRetrieval

Methods 이 논문이 사용한 방법론

Attention 설명 없음
Test 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Adam 설명 없음
Multi-Head Attention 설명 없음
Residual Connection 설명 없음
Layer Normalization Unlike batch normalization, Layer Normalization directly estimates the normalization statistics from the summed inputs…
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…

Similar Papers 제목 키워드 기반

RNNs are not Transformers (Yet): The Key Bottleneck on In-context Retrieval

2024-02-28 · Kaiyue Wen, Xingyu Dang, Kaifeng Lyu

This paper investigates the gap in representation powers of Recurrent Neural Networks (RNNs) and Transformers in the context of solving algorithmic problems. We focus on understanding whether RNNs, known for their memory…

RAGRetrievalRetrieval-augmented Generation

Characterizing Verbatim Short-Term Memory in Neural Language Models

2022-10-24 · Kristijan Armeni, Christopher Honey, Tal Linzen

When a language model is trained to predict natural language sequences, its prediction at each moment depends on a representation of prior context. What kind of information about the prior context can language models ret…

Language ModellingRetrieval

Decoupling Knowledge and Reasoning in Transformers: A Modular Architecture with Generalized Cross-Attention

2025-01-01 · Zhenyu Guo, WenGuang Chen

Transformers have achieved remarkable success across diverse domains, but their monolithic architecture presents challenges in interpretability, adaptability, and scalability. This paper introduces a novel modular Transf…

Retrieval

Retrieval-Enhanced Machine Learning

2022-05-02 · Hamed Zamani, Fernando Diaz, Mostafa Dehghani, Donald Metzler 외

Although information access systems have long supported people in accomplishing a wide range of tasks, we propose broadening the scope of users of information access systems to include task-driven machines, such as machi…

BIG-bench Machine LearningInformation RetrievalRetrieval

Retrieval-Enhanced Visual Prompt Learning for Few-shot Classification

2023-06-04 · Jintao Rong, Hao Chen, Linlin Ou, Tianxiao Chen 외

The Contrastive Language-Image Pretraining (CLIP) model has been widely used in various downstream vision tasks. The few-shot learning paradigm has been widely adopted to augment its capacity for these tasks. However, cu…

ClassificationDomain GeneralizationFew-Shot LearningImage Segmentation+3