English
Related papers

Related papers: Seen to Unseen: Exploring Compositional Generaliza…

200 papers

People easily recognize new visual categories that are new combinations of known components. This compositional generalization capacity is critical for learning in real-world domains like vision and language because the long tail of new…

Computer Vision and Pattern Recognition · Computer Science 2020-11-03 Yuval Atzmon , Felix Kreuk , Uri Shalit , Gal Chechik

There has been considerable progress made towards conversational models that generate coherent and fluent responses; however, this often involves training large language models on large dialogue datasets, such as Reddit. These large…

Computation and Language · Computer Science 2020-10-12 Andrea Madotto , Etsuko Ishii , Zhaojiang Lin , Sumanth Dathathri , Pascale Fung

In this paper, a novel Generation-Evaluation framework is developed for multi-turn conversations with the objective of letting both participants know more about each other. For the sake of rational knowledge utilization and coherent…

Computation and Language · Computer Science 2019-06-04 Siqi Bao , Huang He , Fan Wang , Rongzhong Lian , Hua Wu

Current end-to-end neural conversation models inherently lack the flexibility to impose semantic control in the response generation process, often resulting in uninteresting responses. Attempts to boost informativeness alone come at the…

While most research on controllable text generation has focused on steering base Language Models, the emerging instruction-tuning and prompting paradigm offers an alternate approach to controllability. We compile and release ConGenBench, a…

Computation and Language · Computer Science 2024-05-03 Dhananjay Ashok , Barnabas Poczos

Multiturn dialogue models aim to generate human-like responses by leveraging conversational context, consisting of utterances from previous exchanges. Existing methods often neglect the interactions between these utterances or treat all of…

Computation and Language · Computer Science 2025-04-15 Akanksha Mehndiratta , Krishna Asawa

Prompt-tuning has become an increasingly popular parameter-efficient method for adapting large pretrained language models to downstream tasks. However, both discrete prompting and continuous prompting assume fixed prompts for all data…

Computation and Language · Computer Science 2023-07-12 Runcheng Liu , Ahmad Rashid , Ivan Kobyzev , Mehdi Rezagholizadeh , Pascal Poupart

Humans excel at applying learned behavior to unlearned situations. A crucial component of this generalization behavior is our ability to compose/decompose a whole into reusable parts, an attribute known as compositionality. One of the…

Artificial Intelligence · Computer Science 2024-07-24 Prasanna Vijayaraghavan , Jeffrey Frederic Queisser , Sergio Verduzco Flores , Jun Tani

Open-domain conversation models have become good at generating natural-sounding dialogue, using very large architectures with billions of trainable parameters. The vast training data required to train these architectures aggregates many…

Computation and Language · Computer Science 2020-09-24 Eric Michael Smith , Diana Gonzalez-Rico , Emily Dinan , Y-Lan Boureau

Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe…

Artificial Intelligence · Computer Science 2026-01-16 Boaz Carmeli , Ron Meir , Yonatan Belinkov

Compositional generalization is a fundamental trait in humans, allowing us to effortlessly combine known phrases to form novel sentences. Recent works have claimed that standard seq-to-seq models severely lack the ability to compositionally…

Computation and Language · Computer Science 2022-03-16 Arkil Patel , Satwik Bhattamishra , Phil Blunsom , Navin Goyal

Although neural sequence-to-sequence models have been successfully applied to semantic parsing, they fail at compositional generalization, i.e., they are unable to systematically generalize to unseen compositions of seen components.…

Computation and Language · Computer Science 2021-09-10 Hao Zheng , Mirella Lapata

Controlled generation refers to the problem of creating text that contains stylistic or semantic attributes of interest. Many approaches reduce this problem to training a predictor of the desired attribute. For example, researchers hoping…

Computation and Language · Computer Science 2023-06-02 Carolina Zheng , Claudia Shi , Keyon Vafa , Amir Feder , David M. Blei

We study domain-adaptive image synthesis, the problem of teaching pretrained image generative models a new style or concept from as few as one image to synthesize novel images, to better understand the compositional image synthesis. We…

Computer Vision and Pattern Recognition · Computer Science 2023-06-02 Kihyuk Sohn , Albert Shaw , Yuan Hao , Han Zhang , Luisa Polania , Huiwen Chang , Lu Jiang , Irfan Essa

Although many pretrained models exist for text or images, there have been relatively fewer attempts to train representations specifically for dialog understanding. Prior works usually relied on finetuned representations based on generic…

Computation and Language · Computer Science 2022-05-04 Bishal Santra , Sumegh Roychowdhury , Aishik Mandal , Vasu Gurram , Atharva Naik , Manish Gupta , Pawan Goyal

Despite the recent impressive breakthroughs in text-to-image generation, generative models have difficulty in capturing the data distribution of underrepresented attribute compositions while over-memorizing overrepresented attribute…

Computer Vision and Pattern Recognition · Computer Science 2023-01-05 Yuren Cong , Martin Renqiang Min , Li Erran Li , Bodo Rosenhahn , Michael Ying Yang

Multi-aspect controllable text generation aims to control the generated texts in attributes from multiple aspects (e.g., "positive" from sentiment and "sport" from topic). For ease of obtaining training samples, existing works neglect…

Computation and Language · Computer Science 2024-05-31 Yi Liu , Xiangyu Liu , Xiangrong Zhu , Wei Hu

Deep latent variable models have been shown to facilitate the response generation for open-domain dialog systems. However, these latent variables are highly randomized, leading to uncontrollable generated responses. In this paper, we…

Computation and Language · Computer Science 2017-07-07 Xiaoyu Shen , Hui Su , Yanran Li , Wenjie Li , Shuzi Niu , Yang Zhao , Akiko Aizawa , Guoping Long

Improving the emotional awareness of pre-trained language models is an emerging important problem for dialogue generation tasks. Although prior studies have introduced methods to improve empathetic dialogue generation, few have discussed…

Computation and Language · Computer Science 2023-02-06 Yiren Liu , Halil Kilicoglu

We study knowledge-grounded dialogue generation with pre-trained language models. To leverage the redundant external knowledge under capacity constraint, we propose equipping response generation defined by a pre-trained language model with…

Computation and Language · Computer Science 2020-10-20 Xueliang Zhao , Wei Wu , Can Xu , Chongyang Tao , Dongyan Zhao , Rui Yan