GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay

类型：arxiv
标识：2609.25001
链接：https://arxiv.org/abs/2609.25001
主分类：evaluation
形态：benchmark
TLDR：Modern video games provide a measurable testbed for AI models, combining abilities of visual understanding, instruction decomposition, goal planning, and precise action control over multiple temporal horizons. Existing datasets and benchmarks, however, either cover a narrow range of games, lack language instructions, or rely on high-variance online rollouts. To address these challenges, we introduce GameHorizon, a unified data and evaluation suite that measures gameplay capabilities at different horizons for diverse model families. GameHorizon Suite consists of three components. First, GameHor
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-09-22-agent-rag-longcontext-candidates.json

2026-09-19

笔记 ByteByteGo · RSS 摘要

2026-07-26

笔记 Cameron R. Wolfe (Deep Learning Focus) · RSS 摘要

2026-08-14

笔记精读 · Perception-R1（Visual Perception Reward for MLLM RLVR）

2026-07-06

笔记 ABot-World-0 精读与批判 — 单桌面 GPU 实时长视野交互世界模型(LongForcing 因果蒸馏 + 多源数据基础设施)

2026-07-23

笔记 Cameron Wolfe Substack 短评（双稿）· Agentic World Models + Agentic RL — flyP

2026-07-27

arXiv

2609.25001

TLDR

Modern video games provide a measurable testbed for AI models, combining abilities of visual understanding, instruction decomposition, goal planning, and precise action control over multiple temporal horizons. Existing datasets and benchmarks, however, either cover a narrow range of games, lack language instructions, or rely on high-variance online rollouts. To address these challenges, we introduce GameHorizon, a unified data and evaluation suite that measures gameplay capabilities at different horizons for diverse model families. GameHorizon Suite consists of three components. First, GameHor

arXiv ↗ 中文全文

GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay

相关 · 按语义匹配