MBA：面向真实世界商业构思的多模态基准与智能体

摘要

由大语言模型（LLM）驱动的智能体系统为商业构思开辟了新的机遇。然而，尽管现实世界情境本质上具有多模态特性，现有方法仍局限于纯文本范式。为此，我们提出MBA-Bench，这是首个用于训练和评估商业构思智能体的多模态基准测试，包含六个领域的30K个样本，每个领域具有独特的视觉线索，这些线索无法仅通过文本完整传达。具体而言，我们自动为图像生成字幕，并通过检索查询生成、市场证据检索和证据增强合成三个步骤，使用GPT-4o为三个商业问题各生成五个参考构思。遵循先前工作，我们使用MLLM-as-a-Judge（MLLM作为评审）方法，依据六项面向商业的准则对智能体进行评估。针对准则隐藏或公开的不同场景，我们分别提出MBA-b（盲评）和MBA-k（已知准则）。两者均使用两个新颖的奖励目标进行训练——创造性和可行性——而MBA-k进一步优化六项公开准则，共计八项目标。两者均通过基于LoRA的监督微调，随后结合特定场景奖励进行群体相对策略优化（GRPO）来训练。为在MBA-Bench上进行全面实验，我们设置了两个基线，分别适应仅字幕输入和多模态输入，其中后者在若干指标上接近闭源模型性能。MBA-b和MBA-k相较于字幕基线分别提升63.9%和77.1%，相较于多模态基线分别提升25.6%和35.8%。

English

Agentic systems powered by large language models (LLMs) have opened new opportunities for business ideation. Yet existing approaches remain confined to a text-only paradigm, despite the inherently multimodal nature of real-world contexts. We thus introduce MBA-Bench, the first multimodal benchmark for training and evaluating business ideation agents, comprising 30K samples across six domains, each domain characterized by distinct visual cues not fully conveyed by text alone. Concretely, we automatically caption images and employ GPT-4o to generate five reference ideas for each of three business questions through retrieval query generation, market evidence retrieval, and evidence-augmented synthesis. Following prior work, we evaluate agents across six business-oriented criteria using MLLM-as-a-Judge. To consider settings where criteria are hidden or disclosed, we present MBA-b and MBA-k for blind and known, respectively. We train both with two novel reward objectives---creativity and feasibility---while MBA-k further optimizes the six disclosed criteria for eight in total. Both are trained via LoRA-based supervised fine-tuning followed by group relative policy optimization with these setting-specific rewards. For extensive experiments on MBA-Bench, we set up two baselines accommodating either captions only or multimodal inputs, with the latter nearing closed-source performance on several metrics. MBA-b and MBA-k outperform caption baselines by 63.9% and 77.1%, and multimodal baselines by 25.6% and 35.8%, respectively.