Related papers: Why Task-Based Training is Superior to Traditional…
Achieving human-level translations requires leveraging context to ensure coherence and handle complex phenomena like pronoun disambiguation. Sparsity of contextually rich examples in the standard training data has been hypothesized as the…
Deep learning models are often deployed in downstream tasks that the training procedure may not be aware of. For example, models solely trained to achieve accurate predictions may struggle to perform well on downstream tasks because…
Modern data and applications pose very different challenges from those of the 1950s or even the 1980s. Students contemplating a career in statistics or data science need to have the tools to tackle problems involving massive, heavy-tailed…
Recently, Language Models (LMs) instruction-tuned on multiple tasks, also known as multitask-prompted fine-tuning (MT), have shown the capability to generalize to unseen tasks. Previous work has shown that scaling the number of training…
The effective teaching of data modelling concepts is very important; it constitutes the fundament of database planning methods and the handling of databases with the help of database management lan-guages, typically SQL. We examined three…
Instruction tuning -- tuning large language models on instruction-output pairs -- is a promising technique for making models better adapted to the real world. Yet, the key factors driving the model's capability to understand and follow…
Since the calculation engine of Excel was the subject of a major upgrade to accommodate Dynamic Arrays in 2018 there has been a series of seismic changes to the art of building spreadsheet solutions. This paper will show the ad-hoc end user…
To train neural machine translation models simultaneously on multiple tasks (languages), it is common to sample each task uniformly or in proportion to dataset sizes. As these methods offer little control over performance trade-offs, we…
Bimodal, stochastic environments present a challenge to typical Reinforcement Learning problems. This problem is one that is surprisingly common in real world applications, being particularly applicable to pricing problems. In this paper we…
A business case study on how three simple guidelines: 1. Make it easy to check (and maintain) 2. Make it safe to use 3. Keep business logic out of code changed user attitudes and improved spreadsheet quality in a financial services…
Continual learning aims to train a model incrementally on a sequence of tasks without forgetting previous knowledge. Although continual learning has been widely studied in computer vision, its application to Vision+Language tasks is not…
The paper examines in the context of financial reporting, the controls that organisations have in place to manage spreadsheet risk and errors. There has been widespread research conducted in this area, both in Ireland and internationally.…
Task Arithmetic is a model merging technique that enables the combination of multiple models' capabilities into a single model through simple arithmetic in the weight space, without the need for additional fine-tuning or access to the…
Active learning (AL) is a widely-used training strategy for maximizing predictive performance subject to a fixed annotation budget. In AL one iteratively selects training examples for annotation, often those for which the current model is…
Deep learning methods such as multitask neural networks have recently been applied to ligand-based virtual screening and other drug discovery applications. Using a set of industrial ADMET datasets, we compare neural networks to standard…
Recent research shows that fine-tuning on benign instruction-following data can inadvertently undo the safety alignment process and increase a model's propensity to comply with harmful queries. While instruction-following fine-tuning is…
Many real-world applications require making multiple predictions from the same text. Fine-tuning a large pre-trained language model for each downstream task causes computational burdens in the inference time due to several times of forward…
A spreadsheet usually starts as a simple and single-user software artifact, but, as frequent as in other software systems, quickly evolves into a complex system developed by many actors. Often, different users work on different aspects of…
This paper deals with errors in using spreadsheets and analysis of automatic recording of user interaction with spreadsheets. After a review of literature devoted to spreadsheet errors, we advocate the importance of going from error…
Continual learning aims to enable models to adapt to new datasets without losing performance on previously learned data, often assuming that prior data is no longer available. However, in many practical scenarios, both old and new data are…