[
    {
        "id": "osp-16045",
        "type": "article-journal",
        "title": "MMPostTrainBench: Benchmarking Autonomous Research for Multimodal Post-Training",
        "author": [
            {
                "family": "Liu",
                "given": "Yuxin"
            },
            {
                "family": "Wang",
                "given": "Yuxuan"
            },
            {
                "family": "Lei",
                "given": "Zhenxin"
            },
            {
                "family": "Meng",
                "given": "Lingchen"
            },
            {
                "family": "Sun",
                "given": "Yuchong"
            },
            {
                "family": "Lin",
                "given": "Junming"
            },
            {
                "family": "Liu",
                "given": "Hongcheng"
            },
            {
                "family": "Chu",
                "given": "Yunfei"
            },
            {
                "family": "Yang",
                "given": "Qize"
            },
            {
                "family": "Xu",
                "given": "Jin"
            },
            {
                "family": "Zhang",
                "given": "Lei"
            },
            {
                "family": "Mao",
                "given": "Zhendong"
            }
        ],
        "URL": "https://omanscience.com/en/articles/mmposttrainbench-benchmarking-autonomous-research-for-multimodal-post-training",
        "language": "en",
        "issued": {
            "date-parts": [
                [
                    2026
                ]
            ]
        },
        "abstract": "Autonomous research seeks sustained model improvements through iterative experimentation and feedback. LLM agents show promise in automating machine learning and language-model post-training, but their ability to sustain multimodal improvement remains unclear. We introduce MMPostTrainBench, a benchmark spanning eight tasks in image, audio, video, and joint audio-video understanding and image-grounded software repair. Agents operate from a common base model within fixed budgets, using development feedback before independent evaluation of their submitted models. Evaluation covers target and non-target model outcomes, iterative model improvement and selection, and research integrity. Across all eight tasks, 52.1% of model--task means fall below the base, and evaluated submissions also exhibit non-target regressions. Model performance does not consistently improve across research iterations, and agents do not reliably select the best evaluated candidate for submission; final submissions trail that candidate by up to 5.38 percentage points. Extending autonomous research from text-only to multimodal tasks introduces additional sources of error in perception, cross-modal alignment, and temporal grounding. The observed regressions and selection gaps highlight the need to balance targeted improvements with non-target capability preservation and to retain gains across research iterations. These requirements motivate MMResearch, a multimodal research framework that connects media-grounded evidence to hypotheses and interventions, carries findings across rounds through hierarchical memory, and retains candidates using development evaluation. Added to existing code-agent runtimes, it improves submitted-model accuracy by up to 7.75 percentage points for Claude Opus 4.8 with Claude Code and 2.33 points for GPT-5.6-sol with Codex."
    }
]