It has become standard to solve NLP tasks by fine-tuning pre-trained language models (LMs), especially in low-data settings. There is minimal theoretical understanding of empirical success, e.g., why fine-tuning a model with $10^8$ or more parameters on a couple dozen training points does not result in overfitting. We investigate whether the Neural Tangent Kernel (NTK)—which originated as a model to study the gradient descent dynamics of infinitely wide networks with suitable random initialization—describes fine-tuning of pre-trained LMs. This study was inspired by the decent performance of NTK for computer vision tasks (Wei et al., 2022). We extend the NTK formalism to Adam and use Tensor Programs (Yang, 2020) to characterize conditions under which the NTK lens may describe fine-tuning updates to pre-trained language models. Extensive experiments on 14 NLP tasks validate our theory and show that formulating the downstream task as a masked word prediction problem through prompting often induces kernel-based dynamics during fine-tuning. Finally, we use this kernel view to propose an explanation for the success of parameter-efficient subspace-based fine-tuning methods.
more »
« less
SMART: Robust and Efficient Fine-Tuning for Pre-trained Natural Language Models through Principled Regularized Optimization
Transfer learning has fundamentally changed
the landscape of natural language processing
(NLP). Many state-of-the-art models are first
pre-trained on a large text corpus and then
fine-tuned on downstream tasks. However,
due to limited data resources from downstream
tasks and the extremely high complexity of
pre-trained models, aggressive fine-tuning of-
ten causes the fine-tuned model to overfit the
training data of downstream tasks and fail to
generalize to unseen data. To address such an
issue in a principled manner, we propose a new
learning framework for robust and efficient
fine-tuning for pre-trained models to attain
better generalization performance. The pro-
posed framework contains two important in-
gredients: 1. Smoothness-inducing regulariza-
tion, which effectively manages the complex-
ity of the model; 2. Bregman proximal point
optimization, which is an instance of trust-
region methods and can prevent aggressive up-
dating. Our experiments show that the pro-
posed framework achieves new state-of-the-art
performance on a number of NLP tasks includ-
ing GLUE, SNLI, SciTail and ANLI. More-
over, it also outperforms the state-of-the-art T5
model, which is the largest pre-trained model
containing 11 billion parameters, on GLUE.
more »
« less
- Award ID(s):
- 1717916
- NSF-PAR ID:
- 10162625
- Date Published:
- Journal Name:
- Annual Meeting of the Association for Computational Linguistics
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Recent Transformer-based contextual word representations, including BERT and XLNet, have shown state-of-the-art performance in multiple disciplines within NLP. Fine-tuning the trained contextual models on task-specific datasets has been the key to achieving superior performance downstream. While fine-tuning these pre-trained models is straightforward for lexical applications (applications with only language modality), it is not trivial for multimodal language (a growing area in NLP focused on modeling face-to-face communication). More specifically, this is due to the fact that pre-trained models don’t have the necessary components to accept two extra modalities of vision and acoustic. In this paper, we proposed an attachment to BERT and XLNet called Multimodal Adaptation Gate (MAG). MAG allows BERT and XLNet to accept multimodal nonverbal data during fine-tuning. It does so by generating a shift to internal representation of BERT and XLNet; a shift that is conditioned on the visual and acoustic modalities. In our experiments, we study the commonly used CMU-MOSI and CMU-MOSEI datasets for multimodal sentiment analysis. Fine-tuning MAG-BERT and MAG-XLNet significantly boosts the sentiment analysis performance over previous baselines as well as language-only fine-tuning of BERT and XLNet. On the CMU-MOSI dataset, MAG-XLNet achieves human-level multimodal sentiment analysis performance for the first time in the NLP community.more » « less
-
Fine-tuning pre-trained language models is a common practice in building NLP models for various tasks, including the case with less supervision. We argue that under the few-shot setting, formulating fine-tuning closer to the pre-training objective shall be able to unleash more benefits from the pre-trained language models. In this work, we take few-shot named entity recognition (NER) for a pilot study, where existing fine-tuning strategies are much different from pre-training. We propose a novel few-shot fine-tuning framework for NER, FFF-NER. Specifically, we introduce three new types of tokens, “is-entity”, “which-type” and “bracket”, so we can formulate the NER fine-tuning as (masked) token prediction or generation, depending on the choice of the pre-training objective. In our experiments, we apply to fine-tune both BERT and BART for few-shot NER on several benchmark datasets and observe significant improvements over existing fine-tuning strategies, including sequence labeling, prototype meta-learning, and prompt-based approaches. We further perform a series of ablation studies, showing few-shot NER performance is strongly correlated with the similarity between fine-tuning and pre-training.more » « less
-
null (Ed.)Recent years have witnessed the enormous success of text representation learning in a wide range of text mining tasks. Earlier word embedding learning approaches represent words as fixed low-dimensional vectors to capture their semantics. The word embeddings so learned are used as the input features of task-specific models. Recently, pre-trained language models (PLMs), which learn universal language representations via pre-training Transformer-based neural models on large-scale text corpora, have revolutionized the natural language processing (NLP) field. Such pre-trained representations encode generic linguistic features that can be transferred to almost any text-related applications. PLMs outperform previous task-specific models in many applications as they only need to be fine-tuned on the target corpus instead of being trained from scratch. In this tutorial, we introduce recent advances in pre-trained text embeddings and language models, as well as their applications to a wide range of text mining tasks. Specifically, we first overview a set of recently developed self-supervised and weakly-supervised text embedding methods and pre-trained language models that serve as the fundamentals for downstream tasks. We then present several new methods based on pre-trained text embeddings and language models for various text mining applications such as topic discovery and text classification. We focus on methods that are weakly-supervised, domain-independent, language-agnostic, effective and scalable for mining and discovering structured knowledge from large-scale text corpora. Finally, we demonstrate with real world datasets how pre-trained text representations help mitigate the human annotation burden and facilitate automatic, accurate and efficient text analyses.more » « less
-
Prompting has shown impressive success in enabling large pre-trained language models (LMs) to perform diverse NLP tasks, especially with only few downstream data. Automatically finding the optimal prompt for each task, however, is challenging. Most existing work resorts to tuning *soft* prompts (e.g., embeddings) which fall short of interpretability, reusability across LMs, and applicability when gradients are not accessible. *Discrete* prompts, on the other hand, are difficult to optimize, and are often created by “enumeration (e.g., paraphrasing)-then-selection” heuristics that do not explore the prompt space systematically. This paper proposes RLPrompt, an efficient discrete prompt optimization approach with reinforcement learning (RL). RLPrompt formulates a parameter-efficient policy network that generates the optimized discrete prompt after training with reward. To harness the complex and stochastic reward signals from the large LM environment, we incorporate effective reward stabilization that substantially enhances training efficiency. RLPrompt is flexibly applicable to different types of LMs, such as masked (e.g., BERT) and left-to-right models (e.g., GPTs), for both classification and generation tasks. Experiments on few-shot classification and unsupervised text style transfer show superior performance over a wide range of existing fine-tuning or prompting methods. Interestingly, the resulting optimized prompts are often ungrammatical gibberish text; and surprisingly, those gibberish prompts are transferrable between different LMs to retain significant performance, indicating that LM prompting may not follow human language patterns.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.18653/v1/2020.acl-main.197
