{
  "id": 579,
  "url": "https://arxiv.org/abs/2606.26671v1",
  "title": "NebulaExp-8B: An Empirical Post-Training Pipeline via Full-Scale Ablation Research",
  "summary": "Post-training alignment determines the reasoning and human preference following capabilities of large language models, yet most existing works withhold detailed data construction, filtering rules and training recipes, which hinders community reproducibility and lightweight model optimization. This work presents NebulaExp, a fully transparent, ablation-driven post-training pipeline built on Qwen3-8B-base, covering two orthogonal model branches: general instruct model and complex reasoning-special",
  "authors": "Qiaobo Hao, Yangqian Wu, Shunyi Wang, Zhongjian Zhang, Ziqun Li, Yayin He et al.",
  "category": "research",
  "topics": "safety-alignment,transparency",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-25T07:03:25.000Z",
  "fetched_at": "2026-07-14T14:14:37.249Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/579",
  "original_url": "https://arxiv.org/abs/2606.26671v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}