Research map / Probabilistic and scientific modeling / Variational and generative sampling
ShaLa: Multimodal Shared Latent Generative Modelling
Jiali Cui, Yan-Ying Chen, Yanxia Zhang, Matthew Klenk
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 |
|---|---|---|---|
| condition diffusion prior on embeddings h1:m instead of raw inputs x1:m (ablation) | ShaLa · PolyMNIST unconditional | coherence: 0.584 -> 0.815 | Improved |
| condition diffusion prior on embeddings h1:m instead of raw inputs x1:m (ablation) | ShaLa · PolyMNIST conditional | coherence: 0.612 -> 0.897 | Improved |
| condition diffusion prior on embeddings h1:m instead of raw inputs x1:m (ablation) | ShaLa · MNIST-SVHN-Text unconditional | coherence: 0.23 -> 0.44 | Improved |
| condition diffusion prior on embeddings h1:m instead of raw inputs x1:m (ablation) | ShaLa · MNIST-SVHN-Text conditional | coherence: 0.35 -> 0.75 | Improved |
| halve diffusion prior size (1/2x parameters) (ablation) | ShaLa · PolyMNIST unconditional | FID: 47.30 -> 50.24 | Hurt |
| shrink diffusion prior to 1/4x parameters (ablation) | ShaLa · PolyMNIST unconditional | FID: 47.30 -> 56.44 | Hurt |
| shrink diffusion prior to 1/4x parameters (ablation) | ShaLa · PolyMNIST conditional | FID: 40.18 -> 49.75 | Hurt |
| double diffusion prior size (2x parameters) (ablation) | ShaLa · PolyMNIST unconditional | FID: 47.30 -> 45.24 | Improved |
| double diffusion prior size (2x parameters) (ablation) | ShaLa · PolyMNIST conditional | FID: 40.18 -> 37.15 | Improved |
| replace CMVAE with ShaLa | CMVAE · PolyMNIST unconditional | coherence: 0.781 -> 0.815 | Improved |
| replace CMVAE with ShaLa | CMVAE · PolyMNIST conditional | coherence: 0.897 -> 0.897 | No clear effect |
| replace CMVAE with ShaLa | CMVAE · CUB conditional | FID: 28.00 -> 25.58 | Improved |
Closest papers by meaning
- Unity by Diversity: Improved Representation Learning for Multimodal VAEs (NeurIPS 2024 · similarity 0.88)
- MSP: Modality Self-Play for Training Multimodal LLMs Without Paired Cross-Modal Data (NeurIPS 2026 · similarity 0.87)
- Towards General Modality Translation with Contrastive and Predictive Latent Diffusion Bridge (NeurIPS 2025 · similarity 0.87)
- Scaling Language-centric Omnimodal Representation Learning (NeurIPS 2025 · similarity 0.87)
- LoMo: Local Modality Substitution for Deeper Vision-Language Fusion (NeurIPS 2026 · similarity 0.87)
- UME-R1: Exploring Reasoning-Driven Generative Multimodal Embeddings (ICLR 2026 · 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.
