跳到正文
arXiv cs.AI· Sushant Mehta, Edwin Chen·· 4 小时前AI 评分49

在专家构建任务上做 RL 能给 Nemotron 3.5 Super 带来什么

What RL on Expert-Built Tasks Teaches Nemotron 3.5 Super

AI 导读

NVIDIA 用五个专家构建的任务数据集对 Nemotron 3.5 Super(120B 总参数、12B 激活参数)逐个做强化学习后训练,每次用 CISPO 在 Tinker 上训练全新的 rank-32 LoRA 适配器,无监督阶段。

正文 · 原文

View PDF HTML (experimental)

Abstract:We post-train NVIDIA's Nemotron 3.5 Super (120B total, 12B active parameters) with reinforcement learning (RL) on five expert-built task datasets, one at a time: real-world software engineering, complex constraints, long-horizon tool use, professional chart understanding and enterprise document reasoning. Each run trains a fresh rank-32 LoRA adapter with CISPO on Tinker, with no supervised stage, and each adapter is compared with the base model on the same tasks, harness and settings. On twelve headline benchmarks, eight built by other groups and four by Surge AI, the gains range from +4.5 points on IFBench to +13.8 on GDP_pdf, which rises from 10.4% to 24.2%; SWE-bench Pro V2 rises from 86.9% to 92.1%, Terminal-Bench 2.1 from 63.9% to 70.3%, MultiChallenge from 58.3% to 66.6%, AutomationBench from 14.8% to 27.5% and Chartography from 39.4% to 50.6%. The document adapter, trained without API tools, also gains +8.0 points on AutomationBench. In attempts read side by side with the base model's on tasks each adapter newly solves, the base model's failures recur in a few modes, and each adapter does more often what the base model missed: the tool-use adapter finds the data and the rules before acting, the coding adapter writes its own test or scratch files on SWE-bench Pro V2, the constraints adapter carries earlier turns into its answers, the chart adapter measures the marks in code, and the document adapter reads AutomationBench's spreadsheets.
Comments: 18 pages, 2 figures, 9 tables
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE)
Cite as: arXiv:2611.22161 [cs.LG]
  (or arXiv:2611.22161v1 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2611.22161

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Sushant Mehta [view email]
[v1] Tue, 17 Nov 2026 03:48:06 UTC (77 KB)

来源:arXiv cs.AI · arxiv.org