Is MAP decoding all you need? the inadequacy of the mode in neural machine translation
Recent studies have revealed a number of pathologies of neural machine translation (NMT)
systems. Hypotheses explaining these mostly suggest there is something fundamentally …
systems. Hypotheses explaining these mostly suggest there is something fundamentally …
On reinforcement learning and distribution matching for fine-tuning language models with no catastrophic forgetting
The availability of large pre-trained models is changing the landscape of Machine Learning
research and practice, moving from a" training from scratch" to a" fine-tuning''paradigm …
research and practice, moving from a" training from scratch" to a" fine-tuning''paradigm …
Discriminative reranking for neural machine translation
Reranking models enable the integration of rich features to select a better output hypothesis
within an n-best list or lattice. These models have a long history in NLP, and we revisit …
within an n-best list or lattice. These models have a long history in NLP, and we revisit …
Residual energy-based models for text
Current large-scale auto-regressive language models (Radford et al., 2019; Liu et al., 2018;
Graves, 2013) display impressive fluency and can generate convincing text. In this work we …
Graves, 2013) display impressive fluency and can generate convincing text. In this work we …
Versatile energy-based probabilistic models for high energy physics
T Cheng, AC Courville - Advances in Neural Information …, 2024 - proceedings.neurips.cc
As a classical generative modeling approach, energy-based models have the natural
advantage of flexibility in the form of the energy function. Recently, energy-based models …
advantage of flexibility in the form of the energy function. Recently, energy-based models …
Metaphor generation based on noval evaluation method
Metaphor generation is a difficult research area to study. In this task, the generated content
must maintain certain elements of the original content, such as verbs, adjectives, and …
must maintain certain elements of the original content, such as verbs, adjectives, and …
Improving joint training of inference networks and structured prediction energy networks
Deep energy-based models are powerful, but pose challenges for learning and inference
(Belanger and McCallum, 2016). Tu and Gimpel (2018) developed an efficient framework for …
(Belanger and McCallum, 2016). Tu and Gimpel (2018) developed an efficient framework for …
Sampling from Discrete Energy-Based Models with Quality/Efficiency Trade-offs
B Eikema, G Kruszewski, H Elsahar… - arXiv preprint arXiv …, 2021 - arxiv.org
Energy-Based Models (EBMs) allow for extremely flexible specifications of probability
distributions. However, they do not provide a mechanism for obtaining exact samples from …
distributions. However, they do not provide a mechanism for obtaining exact samples from …
Versatile Energy-Based Models for High Energy Physics
T Cheng, A Courville - 2023 - openreview.net
Energy-Based Models (EBMs) have the natural advantage of flexibility in the form of the
energy function. Recently, EBMs have achieved great success in modeling high …
energy function. Recently, EBMs have achieved great success in modeling high …
E-Forcing: Improving Autoregressive Models by Treating it as an Energy-Based One
Autoregressive generative models are commonly used to solve tasks involving sequential
data. They have, however, been plagued by a slew of inherent flaws due to the intrinsic …
data. They have, however, been plagued by a slew of inherent flaws due to the intrinsic …