Continual Learning for Encoder-only Language Models via a Discrete Key-Value Bottleneck
- URL: http://arxiv.org/abs/2412.08528v1
- Date: Wed, 11 Dec 2024 16:38:34 GMT
- Title: Continual Learning for Encoder-only Language Models via a Discrete Key-Value Bottleneck
- Authors: Andor Diera, Lukas Galke, Fabian Karl, Ansgar Scherp,
- Abstract summary: We introduce a discrete key-value bottleneck for encoder-only language models.
Inspired by the success of a discrete key-value bottleneck in vision, we address new and NLP-specific challenges.
- Score: 6.137272725645159
- License:
- Abstract: Continual learning remains challenging across various natural language understanding tasks. When models are updated with new training data, they risk catastrophic forgetting of prior knowledge. In the present work, we introduce a discrete key-value bottleneck for encoder-only language models, allowing for efficient continual learning by requiring only localized updates. Inspired by the success of a discrete key-value bottleneck in vision, we address new and NLP-specific challenges. We experiment with different bottleneck architectures to find the most suitable variants regarding language, and present a generic discrete key initialization technique for NLP that is task independent. We evaluate the discrete key-value bottleneck in four continual learning NLP scenarios and demonstrate that it alleviates catastrophic forgetting. We showcase that it offers competitive performance to other popular continual learning methods, with lower computational costs.
Related papers
- Multi-Stage Knowledge Integration of Vision-Language Models for Continual Learning [79.46570165281084]
We propose a Multi-Stage Knowledge Integration network (MulKI) to emulate the human learning process in distillation methods.
MulKI achieves this through four stages, including Eliciting Ideas, Adding New Ideas, Distinguishing Ideas, and Making Connections.
Our method demonstrates significant improvements in maintaining zero-shot capabilities while supporting continual learning across diverse downstream tasks.
arXiv Detail & Related papers (2024-11-11T07:36:19Z) - Multi-View Class Incremental Learning [57.14644913531313]
Multi-view learning (MVL) has gained great success in integrating information from multiple perspectives of a dataset to improve downstream task performance.
This paper investigates a novel paradigm called multi-view class incremental learning (MVCIL), where a single model incrementally classifies new classes from a continual stream of views.
arXiv Detail & Related papers (2023-06-16T08:13:41Z) - Deep Active Learning Using Barlow Twins [0.0]
The generalisation performance of a convolutional neural networks (CNN) is majorly predisposed by the quantity, quality, and diversity of the training images.
The goal of the Active learning for the task is to draw most informative samples from the unlabeled pool.
We propose Deep Active Learning using BarlowTwins(DALBT), an active learning method for all the datasets.
arXiv Detail & Related papers (2022-12-30T12:39:55Z) - Language Model Pre-Training with Sparse Latent Typing [66.75786739499604]
We propose a new pre-training objective, Sparse Latent Typing, which enables the model to sparsely extract sentence-level keywords with diverse latent types.
Experimental results show that our model is able to learn interpretable latent type categories in a self-supervised manner without using any external knowledge.
arXiv Detail & Related papers (2022-10-23T00:37:08Z) - Discrete Key-Value Bottleneck [95.61236311369821]
Deep neural networks perform well on classification tasks where data streams are i.i.d. and labeled data is abundant.
One powerful approach that has addressed this challenge involves pre-training of large encoders on volumes of readily available data, followed by task-specific tuning.
Given a new task, however, updating the weights of these encoders is challenging as a large number of weights needs to be fine-tuned, and as a result, they forget information about the previous tasks.
We propose a model architecture to address this issue, building upon a discrete bottleneck containing pairs of separate and learnable key-value codes.
arXiv Detail & Related papers (2022-07-22T17:52:30Z) - Multimodal Knowledge Alignment with Reinforcement Learning [103.68816413817372]
ESPER extends language-only zero-shot models to unseen multimodal tasks, like image and audio captioning.
Our key novelty is to use reinforcement learning to align multimodal inputs to language model generations without direct supervision.
Experiments demonstrate that ESPER outperforms baselines and prior work on a variety of zero-shot tasks.
arXiv Detail & Related papers (2022-05-25T10:12:17Z) - Detecting Requirements Smells With Deep Learning: Experiences,
Challenges and Future Work [9.44316959798363]
This work aims to improve the previous work by creating a manually labeled dataset and using ensemble learning, Deep Learning (DL), and techniques such as word embeddings and transfer learning to overcome the generalization problem.
The current findings show that the dataset is unbalanced and which class examples should be added more.
arXiv Detail & Related papers (2021-08-06T12:45:15Z) - Generalized Zero-Shot Learning using Multimodal Variational Auto-Encoder
with Semantic Concepts [0.9054540533394924]
Recent techniques try to learn a cross-modal mapping between the semantic space and the image space.
We propose a Multimodal Variational Auto-Encoder (M-VAE) which can learn the shared latent space of image features and the semantic space.
Our results show that our proposed model outperforms the current state-of-the-art approaches for generalized zero-shot learning.
arXiv Detail & Related papers (2021-06-26T20:08:37Z) - Reinforced Iterative Knowledge Distillation for Cross-Lingual Named
Entity Recognition [54.92161571089808]
Cross-lingual NER transfers knowledge from rich-resource language to languages with low resources.
Existing cross-lingual NER methods do not make good use of rich unlabeled data in target languages.
We develop a novel approach based on the ideas of semi-supervised learning and reinforcement learning.
arXiv Detail & Related papers (2021-06-01T05:46:22Z) - Fine-tuning BERT for Low-Resource Natural Language Understanding via
Active Learning [30.5853328612593]
In this work, we explore fine-tuning methods of BERT -- a pre-trained Transformer based language model.
Our experimental results show an advantage in model performance by maximizing the approximate knowledge gain of the model.
We analyze the benefits of freezing layers of the language model during fine-tuning to reduce the number of trainable parameters.
arXiv Detail & Related papers (2020-12-04T08:34:39Z)
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.