Metraipolis › archive › dc_256ce4394c2c
Sol-H3: Recursive Self-Improvement for MiniMax-H3 Inference Acceleration on Sol-Engine across Cloud and Edge — arXiv:2609.35110
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.35110
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:58:32.743Z
- SHA-256 of the archived text
9f2d6a8af8032ad5497d35987b0b18ecbb20c0cf275750114c37a2fcd45b19b5- Stable URL of this page
https://metraipolis.com/documento/dc_256ce4394c2c
How to cite
Sol-H3: Recursive Self-Improvement for MiniMax-H3 Inference Acceleration on Sol-Engine across Cloud and Edge — arXiv:2609.35110 — arXiv (en), https://arxiv.org/abs/2609.35110 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_256ce4394c2c (Metraipolis archive dc_256ce4394c2c, sha256 9f2d6a8af8032ad5497d35987b0b18ecbb20c0cf275750114c37a2fcd45b19b5)
BibTeX
@misc{dc256ce4394c2c,
title = {Sol-H3: Recursive Self-Improvement for MiniMax-H3 Inference Acceleration on Sol-Engine across Cloud and Edge — arXiv:2609.35110},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.35110},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_256ce4394c2c; sha256 9f2d6a8af8032ad5497d35987b0b18ecbb20c0cf275750114c37a2fcd45b19b5},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.35110. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Video diffusion models are rapidly scaling and exhibiting enhanced generation capabilities. Among these recent advancements, MiniMax-H3 stands out as a highly capable, production-level open-source model. However, its 33-billion parameters and multi-step iterative denoising process introduce substantial computational overhead. Consequently, their practical production is hindered by generation latency in the cloud deployment like NVIDIA-GB200, alongside strict memory limits that pose further challenges at the edge device like DGX-Spark. To address these diverse hardware bottlenecks from cloud to edge device, we present a full-stack inference pipeline that integrates efficient algorithmic design with optimized operator implementations. Algorithmically, we introduce a cross-resolution two-stage generation scheduler that exploits the step-wise nature of diffusion: early low-resolution steps rapidly establish the global layout, while later high-resolution steps focus refinements of local and perceptual details. These stages are connected by a learned latent-to-latent mapping module, completely eliminating the computationally expensive VAE decode-reencode cycle for resolution transferring cross different resolutions. For operator implementation, we deploy a Recursive Self-Improvement (RSI) loop that searches kernel fusions and memory layouts, evaluating latency together with numerical agreement. Together, these optimizations deliver up to 30x end-to-end speedup and 20% lower memory: a 5-second 1344x768 video with audio is generated 3.5x faster than real time on an 8xGB200 node, and in under a minute fully memory-resident on a single DGX Spark. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI) ; Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.35110 — 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 2026-10-07T20:29:01.783Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).