今日速览
| 序号 | 标题 | 来源 | 日期 | 主题 | 推荐等级 |
|---|---|---|---|---|---|
| 1 | PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents | arXiv | 2026-08-04 | RAG | 高 |
| 2 | Should We Type or Talk to LLM Agents? A Comprehensive Study of Voice and Keyboard Input Perturbations | arXiv | 2026-08-04 | AI-Agent | 高 |
重点论文与技术动态
1. PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents
- 来源：arXiv
- 日期：2026-08-04
- 作者/机构：Shuhan Xue, Zixin Ding, Yichen Shen, Yinjie Wang, Zhenfei Yin, Yingcheng Wu
- 主题标签：
RAG,arXiv - 推荐等级：高
- 分类：cs.CL
一句话结论
PAST-Bench基准测试与Hermes+框架证实了个人AI代理通过保留经验实现递归自我改进的可行性，但改进效果在不同模型和任务能力上存在显著差异。
核心内容
- PAST-Bench系统隔离测试了个人AI代理保留经验对后续行为的改善效果，通过控制经验开启/关闭机制验证递归路径。
- 实验覆盖26个场景和204个任务剧集，涵盖记忆、程序重用、信息收集与更新四大核心能力。
- Hermes+通过五项针对性干预显著提升保留经验的平均收益，尤其在需替换过时状态的任务上效果突出。
方法与数据
- 采用控制变量法，对比代理在保留经验开启/关闭条件下的任务表现，测试递归自我改进路径。
- 涵盖七种基础模型和四种代理框架，基于26个场景及204个任务剧集的系统性实验数据。
价值判断
- 值得关注：首个系统性评估持久代理递归自我改进的基准框架，填补了该领域方法论空白。
- 可复用点：Hermes+的五项干预策略可迁移至其他代理系统，优化经验利用效率。
- 局限/待核查：改进效果高度依赖模型能力和任务类型，未在所有能力上实现一致提升。
摘要
Recursive self-improvement requires agents to turn accumulated experience into better future behavior. Personal AI agents offer a concrete setting for studying this capability because they retain preferences, task histories, tool routines, and learned skills across sessions. Yet whether retained experience actually improves them over time has not been systematically tested. We introduce PAST-Bench, a benchmark designed to isolate this question. Each agent runs through ordered sequences of fresh-session tasks under matched conditions that turn retained experience on and off. It spans 26 scenarios and 204 episodes across memory, procedural reuse, information gathering, and update. We report both later-task gains and whether those gains follow the intended save, retrieve, and update pathway. Across seven base models and four agent frameworks, improvement is real but uneven across capabilities. Agents with the same headline gain can differ markedly in whether that gain is supported by evidence of the intended pathway. Guided by these findings, we develop Hermes+, which extends Hermes with five targeted interventions across stages of the agent loop. Hermes+ raises the average gain from retained experience and provides clearer pathway evidence, with its strongest improvement on tasks requiring outdated state to be replaced, although the effect remains capability- and model-dependent. Together, PAST-Bench and Hermes+ provide an evaluation and diagnostic foundation for studying how persistent agents can progress from retaining experience to systematically improving through it. Code: https://github.com/Gen-Verse/PAST-Bench2. Should We Type or Talk to LLM Agents? A Comprehensive Study of Voice and Keyboard Input Perturbations
- 来源：arXiv
- 日期：2026-08-04
- 作者/机构：Zizhao Hu, Nathan Elijah Segura, Mohammad Rostami, Jesse Thomason
- 主题标签：
AI-Agent,arXiv - 推荐等级：高
- 分类：cs.AI
一句话结论
语音输入扰动对LLM性能的影响显著大于键盘输入扰动，尤其在需推理的任务中。
核心内容
- 语音转录扰动普遍降低所有测试指令微调模型的准确率，转录结构而非填充词是主要成本来源。
- 键盘扰动成本较低，模型能吸收大量扰动后准确率才下降。
- 问题token存活率决定性能：破坏token导致性能下降，添加新token影响微小。
方法与数据
- 方法：HIVE（Human Input-Variation Engine）工具，集成语音转录与QWERTY键盘扰动。
- 数据：摘要未明确具体数据集。
价值判断
- 值得关注：语音输入在需答案构建或推导的任务中显著降低LLM性能，影响实际应用可靠性。
- 可复用点：HIVE框架可作为评估LLM输入鲁棒性的通用工具。
- 局限/待核查：摘要未明确局限，但压缩语音输入效果更差需进一步验证。