# RNG-Bench：多模态大语言模型在可控非马尔可夫游戏中的评估基准

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-06-17 08:00
- AIHOT 分数：57
- AIHOT 链接：https://aihot.virxact.com/items/cmqiu67yt04a1sl5we3fy3kw4
- 原文链接：https://arxiv.org/abs/2606.19338

## AI 摘要

RNG-Bench是一个评估多模态大语言模型在闭环交互中重建过去观测并据此行动能力的基准套件，包含Matching Pairs和3D Maze两个互补游戏。基准通过网格大小、视觉模式和观测模态三个难度轴控制复杂度，并引入对决赛制消除实例方差及Memory Gap指标分离遗忘与决策错误。最难配置需约128K tokens上下文和每轮350张图像，前沿模型尚未饱和。Memory Gap分析显示主要错误源于遗忘。微调Qwen3.5-9B可提升该基准性能并迁移至现有基准，不损失通用多模态能力。

## 正文

Deploying multimodal foundation models as closed-loop policies increasingly requires conditioning actions on observations that are no longer visible. However, existing benchmarks either expose the full state, conflate hidden-state reconstruction with other agent skills, or test recall only after an episode has ended. We introduce RNG-Bench (Reconstructive Non-Markov Games), a benchmark suite designed to isolate a base model's ability to reconstruct past observations and act on them during multi-step interaction. RNG-Bench includes two complementary games: Matching Pairs, where card identities briefly revealed at specific locations must later be recalled, and 3D Maze, where egocentric views must be integrated into a spatial map. Both games are evaluated under a unified harness with three controlled difficulty axes: grid size, visual pattern, and observation modality. The benchmark further introduces a head-to-head duel protocol to control for instance-level variance and a Memory Gap metric that disentangles forgetting from poor action selection. The hardest configurations require contexts of roughly 128K tokens and 350 image inputs per episode, and remain far from saturated by frontier MLLMs. Memory Gap analysis shows that most residual errors stem from forgetting earlier observations rather than from suboptimal decision making. Finally, fine-tuning Qwen3.5-9B on optimal-policy rollouts and filtered model demonstrations improves performance on RNG-Bench and transfers to existing benchmarks without degrading general multimodal capability.