Challenge: Variational Auto-Encoder (VAE) has been widely adopted in text generation due to its ability to learn flexible representations.
Approach: They propose a Transformer-based recurrent VAE structure that imposes recurrence on segment-wise latent variables with arbitrarily separated text segments and constructs the posterior distribution with residual parameterization.
Outcome: The proposed structure can deduce a non-zero lower bound of the KL term and enhance the entanglement of each segment and preceding latent variables, providing a theoretical guarantee of generation diversity.

Similar Papers

Empirical Prior for Text Autoencoders (2024.findings-emnlp)

Copied to clipboard

Challenge: Variational Autoencoders (VAE) are used to train generative models with latent variables.
Approach: They propose a transition from Variational Autoencoders (VAE) to text autoencodeurs (AE) which model a compact latent space and preserves the capability of the language model itself.
Outcome: The proposed method generates higher quality and more diverse text than the VAE-based Transformer baselines, and is more efficient than previous approaches.
Fuse It More Deeply! A Variational Transformer with Layer-Wise Latent Variable Inference for Text Generation (2022.naacl-main)

Copied to clipboard

Challenge: Variational Auto-Encoders are often used for text generation tasks due to the sequential nature of the text.
Approach: They propose a variational Transformer framework that learns a series of layer-wise latent variables with each inferred from those of lower layers and tightly coupled with the hidden states by low-rank tensor product.
Outcome: The proposed framework can learn latent variables from lower layers and incorporate more information.
Variational Autoregressive Decoder for Neural Response Generation (D18-1)

Copied to clipboard

Challenge: Existing variational Bayesian models generate responses from a single latent variable, which is not sufficient to model high variability in responses.
Approach: They propose a conditional variable auto-encoder that sequentially introduces latent variables to condition the generation of each word in the response sequence.
Outcome: Empirical results show that the proposed model improves on state-of-the-art models on Opensubtitle and Reddit datasets.
VOLTA: Improving Generative Diversity by Variational Mutual Information Maximizing Autoencoder (2024.findings-naacl)

Copied to clipboard

Challenge: generative diversity is a critical yet underexplored issue in natural language generation . previous approaches to enhance diversity of Transformer models have been limited by their latent variables .
Approach: They propose a framework that bridges Transformer with VAE to enhance generative diversity.
Outcome: The proposed framework improves generative diversity while maintaining generative quality.
Generative Text Modeling through Short Run Inference (2021.eacl-main)

Copied to clipboard

Challenge: Latent variable models for text capture global semantic and syntactic features when trained correctly.
Approach: They propose a short run dynamics for inference that initializes from the prior distribution of the latent variable and runs a small number of Langevin dynamics steps guided by its posterior distribution.
Outcome: The proposed model is able to generate coherent sentences with smooth transition and shows no sign of posterior collapse.
Improving Variational Autoencoder for Text Modelling with Timestep-Wise Regularisation (2020.coling-main)

Copied to clipboard

Challenge: Variational Autoencoders (VAEs) have been widely used in text modelling but posterior collapse is a problem when RNN-based models are employed.
Approach: They propose a timestep-wise regularisation VAE architecture which can effectively avoid posterior collapse when used in text modelling.
Outcome: The proposed model avoids posterior collapse and can be applied to any RNN-based VAE model.
TILGAN: Transformer-based Implicit Latent GAN for Diverse and Coherent Text Generation (2021.findings-acl)

Copied to clipboard

Challenge: Existing autoregressive models suffer from the exposure bias problem due to mismatches between training and generation stages.
Approach: They propose a Transformerbased Implicit Latent GAN which combines a transformer autoencoder and GAN in the latent space with a novel design and distribution matching based on the Kullback-Leibler divergence.
Outcome: The proposed model improves local and global coherence and quality-diversity trade-off on three benchmark datasets.
Implicit Deep Latent Variable Models for Text Generation (D19-1)

Copied to clipboard

Challenge: Variational auto-encoders have been used for text generation but their representation power is limited due to two reasons.
Approach: They advocate sample-based representations of variational distributions for natural language . they further develop an LVM to directly match the aggregated posterior to the prior .
Outcome: The proposed model can be viewed as a natural extension of VAEs with a regularization of maximizing mutual information, mitigating the "posterior collapse" issue.
Diffusion Glancing Transformer for Parallel Sequence-to-Sequence Learning (2024.naacl-long)

Copied to clipboard

Challenge: Experimental results show that non-autoregressive generation models are superior in generation efficiency but inferior in generation quality.
Approach: They propose a diffusion glancing transformer which employs a modality diffusion process and residual glancy sampling to improve multi-modality modeling.
Outcome: The proposed model outperforms autoregressive and non-autoregressive models on machine translation and text generation benchmarks.
Better Exploiting Latent Variables in Text Modeling (P19-1)

Copied to clipboard

Challenge: Consistent gains in performance on two datasets, Penn Treebank and Yahoo, indicate the generalizability of our method.
Approach: They propose a method to exploit latent variables through hidden state averaging by sampling latent variable multiple times at a gradient step.
Outcome: The proposed method shows consistent gains on two datasets showing that it is generalizable.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations