Chinese contentHarness-S：借助兄弟 Harness 训练更通用的智能体
一篇训练智能体的新论文，思路挺巧：基座 harness 跑挂了就让一堆兄弟 harness 救场，再把本事转移回策略本身，AppWorld 等三个基准都涨了。
#AppWorld
6 stories · 1 in 7 days · 1 in 30 days
Stories, products and papers tagged "AppWorld", newest first.
Nov 4
Oct 7
Chinese content新研究：用 Agent 历史预测上下文压缩何时造成损害
做长任务 Agent 的朋友看看这篇：TRACE 用 590 个真实压缩边界回放数据测了压缩时机怎么选，结论是历史行为预测力很弱，数据和方法都开源了。
Oct 6
PAIR method diagnoses context compression failures in long-horizon agents
Compression damages run-to-run reliability before it makes tasks unsolvable, so single-run evals will miss the problem — repeated-run consistency is the metric to watch. Builders of long-horizon agent systems can use PAIR's replay-and-compare approach to pinpoint which compression events break their workflows. The paper's benchmarks, AppWorld and tau-Bench Retail, offer concrete testbeds for comparing compression prompts.
Oct 4
Microsoft paper ScholarEvolve evolves agent harnesses from research papers
The counterintuitive idea is that the literature, not runtime traces, is the highest-value source of harness improvements. Builders tuning agent scaffolds around a fixed model can replicate the module-and-topic-modeling loop with their own paper corpus instead of paying for a model upgrade. Results on AppWorld Challenge and Tau2-Bench Telecom give a concrete baseline for comparing it against failure-log-driven harness tuning.
Aug 4
Jul 20
Papers09:18
Chinese content模型合并 vs 联合多任务强化学习：任务向量几何分析这篇论文用Qwen3-8B做了AppWorld上的合并vs联合训练对比，发现合并效果不输联合训练，而且解释了为什么不同合并方法结果差不多。