Research map / Probabilistic and scientific modeling / Variational and generative sampling
Improved Variational Autoencoders for Text Modeling using Dilated Convolutions
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Replace the LSTM language-model decoder with an LCNN VAE decoder and initialize the encoder from an LSTM language model | Text language modeling · Yahoo | PPL: 66.2 -> 63.9 | Helped |
| Replace the LSTM language-model decoder with an LCNN VAE decoder and initialize the encoder from an LSTM language model | Text language modeling · Yelp | PPL: 42.6 -> 41.1 | Helped |
| Use an LSTM VAE decoder instead of an LSTM language model | Text language modeling · Yahoo | PPL: 66.2 -> 72.5 | Hurt |
| Initialize the LCNN-VAE encoder from an LSTM language model (ablation) | LCNN-VAE · Yahoo | PPL: 65.4 -> 63.9 | Helped |
| Initialize the LCNN-VAE encoder from an LSTM language model (ablation) | LCNN-VAE · Yelp | PPL: 42.3 -> 41.1 | Helped |
| Increase VAE decoder contextual capacity from SCNN through LCNN (ablation) | CNN-VAE · Yahoo | PPL: 68.7 -> 67.3 -> 65.4 -> 67.6 | Mixed |
| Increase VAE decoder contextual capacity from SCNN through LCNN (ablation) | CNN-VAE · Yelp | PPL: 43.9 -> 42.8 -> 42.3 -> 43.4 | Mixed |
| Use a smaller SCNN instead of a larger decoder (ablation) | Semi-supervised VAE · Yahoo, 500 labels | accuracy: 57.2 -> 65.5 | Helped |
| Use an LCNN instead of an SCNN decoder (ablation) | Semi-supervised VAE · Yahoo, 500 labels | NLL: 335.7 -> 331.3 | Helped |
| Add semi-supervised labels to the SCNN-VAE objective | SCNN-VAE · Yahoo | NLL: 337.8 to 335.7 | Helped |
| Initialize SCNN-VAE-Semi encoder from a language model (ablation) | Semi-supervised classification · Yahoo test, 100 labels | accuracy: 55.4 -> 63.8 | Helped |
| Initialize SCNN-VAE-Semi encoder from a language model (ablation) | Semi-supervised classification · Yelp test, 500 labels | accuracy: 53.5 -> 57.3 | Helped |
Closest papers by meaning
- A Comprehensive Study of Decoder-Only LLMs for Text-to-Image Generation (CVPR 2025 · similarity 0.88)
- Fast and Fluent Diffusion Language Models via Convolutional Decoding and Rejective Fine-tuning (NeurIPS 2025 · similarity 0.88)
- Improving Sparse Decomposition of Language Model Activations with Gated Sparse Autoencoders (NeurIPS 2024 · similarity 0.87)
- Sparse Autoencoders, Again? (ICML 2025 · similarity 0.87)
- Low-Rank Adapting Models for Sparse Autoencoders (ICML 2025 · similarity 0.87)
- Scaling and evaluating sparse autoencoders (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.