-
2024This paper introduces Q-tuning, a novel approach for continual prompt tuning that enables the lifelong learning of a pre-trained language model. When learning a new task, Q-tuning trains a task-specific prompt by adding it to a prompt queue consisting of the prompts from older tasks. To better transfer the knowledge of old tasks, we design an adaptive knowledge aggregation technique that reweighs previous
-
AISTATS 20242024We propose a notion of causal influence that describes the ‘intrinsic’ part of the contribution of a node on a target node in a DAG. By recursively writing each node as a function of the upstream noise terms, we separate the intrinsic information added by each node from the one obtained from its ancestors. To interpret the intrinsic information as a causal contribution, we consider ‘structure-preserving
-
AAAI 20242024The selection of the assumed effect size (AES) critically determines the duration of an experiment, and hence its accuracy and efficiency. Traditionally, experimenters determine AES based on domain knowledge. However, this method becomes impractical for online experimentation services managing numerous experiments, and a more automated approach is hence of great demand. We initiate the study of data-driven
-
Multi-Touch Attribution plays a crucial role in both marketing and advertising, offering insight into the complex series of interactions within customer journeys during transactions or impressions. This holistic approach empowers marketers to strategically allocate attribution credits for conversions across diverse channels, not only optimizing campaigns but also elevating overall marketplace strategies
-
CVPR 2024, CVPR 2024 Workshop on What is Next in Multimodal Foundation Models?, CVPR 2024 Workshop on Robustness in Large Language Models2024Generative Vision-Language Models (VLMs) are prone to generate plausible-sounding textual answers that, however, are not always grounded in the input image. We investigate this phenomenon, usually referred to as “hallucination” and show that it stems from an excessive reliance on the language prior. In particular, we show that as more tokens are generated, the reliance on the visual prompt decreases, and
Related content
-
August 12, 2020Classes previously only available to Amazon employees will now be available to the community.
-
July 24, 2020Amazon automated reasoning scientists showcase verification methods being applied across Amazon during CAV 2020.
-
July 24, 2020New position encoding scheme improves state-of-the-art performance on several natural-language-processing tasks.
-
July 20, 2020Method presented to ICML workshop works with any machine learning model and fairness criterion.
-
July 17, 2020Watch the keynote presentation by Alex Smola, AWS vice president and distinguished scientist, presented at the AutoML@ICML2020 workshop.
-
July 15, 2020New transferability metric is more accurate and more generally applicable than predecessors.