cd /news/large-language-models/sdo-structure-aware-data-organizatio… · home topics large-language-models article
[ARTICLE · art-81319] src=arxiv.org ↗ pub= topic=large-language-models verified=true sentiment=· neutral

SDO: Structure-Aware Data Organization for Efficient LLM Post-Training

Researchers propose SDO (Structure-Aware Data Organization), a plug-and-play framework that organizes mini-batch composition and sample exposure based on representation-space structure to improve efficiency in large language model post-training. SDO uses locality-aware batching via KNN neighborhood traversal within epochs and exposure-balanced scheduling across epochs to reduce redundant updates and improve convergence across SFT, DPO, and GRPO, with the largest gains in the early-to-mid training phase.

read1 min views1 publishedJul 31, 2026

arXiv:2607.27273v1 Announce Type: new Abstract: Post-training of large language models is expensive, and existing efficiency improvements mainly focus on selecting informative samples or designing training schedules. However, data organization itself is usually treated as a static preprocessing step: embedding-based grouping methods construct fixed partitions before training and cannot adapt to the evolving sample exposure during optimization. As a result, all samples receive similar exposure despite their different optimization needs, leading to redundant updates for some samples while leaving others under-optimized. To address this problem, we propose SDO (Structure-Aware Data Organization), a plug-and-play data organization framework with an exposure-driven feedback mechanism that organizes mini-batch composition and sample exposure according to representation-space structure. SDO operates epoch by epoch on frozen external embeddings, avoiding model warm-up training overhead: within each epoch, locality-aware batching forms coherent mini-batches via KNN neighborhood traversal; across epochs, exposure-balanced scheduling records per-sample participation and reduces the sampling probability of over-exposed samples to preserve long-term coverage. Across SFT, DPO, and GRPO, SDO accelerates convergence, with the largest gains observed in the early-to-mid phase, producing more coherent gradients and more balanced accuracy across question types without permanently excluding training samples.

── more in #large-language-models 4 stories · sorted by recency
── more on @sdo 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/sdo-structure-aware-…] indexed:0 read:1min 2026-07-31 ·