Research map / Image generation / Autoregressive and latent generation
PLANNER: Generating Diversified Paragraph via Latent Language Diffusion Model
Yizhe Zhang, Jiatao Gu, Zhuofeng Wu, Shuangfei Zhai, Joshua M. Susskind, Navdeep Jaitly
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| select last k hidden vectors instead of first k (ablation) | PLANNER embedder · Hotel Review | BLEU_clean: 79.59 -> 78.96 | No clear effect |
| select interleaving hidden vectors instead of first k (ablation) | PLANNER embedder · Hotel Review | BLEU_clean: 79.59 -> 79.81 | No clear effect |
| reduce latent codes from k=16 to k=8 (ablation) | PLANNER embedder · Hotel Review | BLEU_clean: 79.59 -> 57.90 | Hurt |
| increase latent codes from k=16 to k=32 (ablation) | PLANNER embedder · Hotel Review | BLEU_clean: 79.59 -> 82.31 | Improved |
| increase latent codes from k=16 to k=32 (ablation) | PLANNER embedder · Hotel Review | BLEU_robust: 43.17 -> 45.14 | Improved |
| use PLANNER instead of Diff-LM | text diffusion · Hotel Review sentiment | accuracy: 83.6 -> 94.9 | Improved |
| use PLANNER instead of Diff-LM | text diffusion · Hotel Review completion | Rep-4: 4.54 -> 1.55 | Improved |
| use PLANNER instead of FT-sample | autoregressive finetuning · Hotel Review sentiment | Rep-4: 5.86 -> 2.37 | Improved |
| use PLANNER instead of Genie | text diffusion · CNN Dailymail | ROUGE-L: 30.47 -> 30.43 | No clear effect |
| generate 10 samples and take best Rouge | PLANNER · CNN Dailymail | ROUGE-L: 30.43 -> 36.81 | Improved |
| generate 10 samples and take best Rouge | PLANNER · XSum | ROUGE-L: 33.97 -> 41.23 | Improved |
| use deterministic search (T5-search) instead of PLANNER | summarization · CNN Dailymail | ROUGE-L: 34.48 -> 30.43 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- Genie — GENIE: Higher-Order Denoising Diffusion Solvers (NeurIPS 2022) · better on 2 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Scaling Diffusion Language Models via Adaptation from Autoregressive Models (ICLR 2025 · similarity 0.89)
- Energy-Based Diffusion Language Models for Text Generation (ICLR 2025 · similarity 0.89)
- Continuous Latent Diffusion Language Model (NeurIPS 2026 · similarity 0.89)
- Denoising Time Matters: Diverse Generation in Diffusion Language Models (NeurIPS 2026 · similarity 0.89)
- Compressed and Smooth Latent Space for Text Diffusion Modeling (NeurIPS 2025 · similarity 0.88)
- How to Train Your Latent Diffusion Language Model Jointly With the Latent Space (NeurIPS 2026 · similarity 0.88)
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.
