Metraipolis › archive › dc_64de26eb6770
GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling — arXiv:2608.29335
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.29335
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T09:58:58.088Z
- SHA-256 of the archived text
0067e503310e96b327bc167043a409c91b849b53d9de3be61a2836aebb8d7977- Stable URL of this page
https://metraipolis.com/documento/dc_64de26eb6770
How to cite
GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling — arXiv:2608.29335 — arXiv (en), https://arxiv.org/abs/2608.29335 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_64de26eb6770 (Metraipolis archive dc_64de26eb6770, sha256 0067e503310e96b327bc167043a409c91b849b53d9de3be61a2836aebb8d7977)
BibTeX
@misc{dc64de26eb6770,
title = {GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling — arXiv:2608.29335},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.29335},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_64de26eb6770; sha256 0067e503310e96b327bc167043a409c91b849b53d9de3be61a2836aebb8d7977},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.29335. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Latent generative models typically follow a two-stage pipeline, training a variational autoencoder for reconstruction and then a generative model on the frozen latent space. Since reconstruction-optimized latents are not necessarily generation-friendly, jointly training both models is an appealing alternative. However, direct end-to-end training remains challenging, as it is prone to latent collapse and faces a generation-reconstruction conflict. We revisit this problem by analyzing how different objectives shape the latent space and identify two key insights. First, the entropy term in the Kullback-Leibler divergence objective is essential for preventing collapse: reconstruction and prior fitting tend to shrink the posterior, while entropy preserves non-degenerate latent uncertainty. Second, reconstruction and generation exhibit asymmetric learning dynamics: reconstruction is fast and strongly supervised, whereas generation is slower and harder to optimize. Based on these insights, we achieve the first direct end-to-end training without latent collapse and propose GenFirst, a simple generation-before-reconstruction strategy. The generative objective first shapes the latent space under weak reconstruction pressure, after which reconstruction is progressively strengthened to recover visual details. We validate GenFirst with continuous autoregressive priors with exact likelihoods and SiT priors with implicit likelihoods. With our end-to-end objective and GenFirst, SiT achieves a gFID of 0.97 with CFG and 1.45 without CFG on ImageNet-256, while MMDiT reaches a GenEval score of 0.90 on text-to-image generation. Beyond image generation, we extend the framework to shared visual latents for generation and representation learning, and to continuous unified text-image generation. These results demonstrate the generality of stable end-to-end latent learning across generative priors and modalities. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2608.29335 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-17T09:28:48.028Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).