SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.11784 · ICML · 2025

Data-Juicer Sandbox: A Feedback-Driven Suite for Multimodal Data-Model Co-development

Yaliang Li, Bolin Ding, Jingren Zhou, Daoyuan Chen, Haibin Wang, Ce Ge, Yilun Huang, Model Op, Op

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 10 functions out of this paper's own repositories and ran 4 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
modelscope/data-juicer alias 4 of 10
FunctionStatusWhere it lives
PartitionedDAGStrategy Ran modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("3ffdac1f9fe42290")
PipelineDAG Ran modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("b8e00b9ff0a49608")
ScatterGatherNode Ran modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("2216173936ba88c4")
is_global_operation Ran modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("20647d3165062ad5")
DAGExecutionMixin Not yet run modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("2ad62cd47a823b20")
DAGExecutionStrategy Not yet run modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("a0472b5af9cfd25c")
DAGNodeStatusTransition Not yet run modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("1f8ece291249d70a")
DAGNodeType Not yet run modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("5dd6308a11d247fc")
EventType Not yet run modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("c63fd2ec15a736f3")
NonPartitionedDAGStrategy Not yet run modelscope/data-juicer/data_juicer/core/executor/dag_execution_mixin.py
code served (permissive licence) · get_code("27256aeeb1e862df")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The emergence of multimodal large models has advanced artificial intelligence, introducing unprecedented levels of performance and functionality. However, optimizing these models remains challenging due to historically isolated paths of model-centric and data-centric developments, leading to suboptimal outcomes and inefficient resource utilization. In response, we present a new sandbox suite tailored for integrated data-model co-development. This sandbox provides a feedback-driven experimental platform, enabling cost-effective iteration and guided refinement of both data and models. Our proposed "Probe-Analyze-Refine" workflow, validated through practical use cases on multimodal tasks such as image-text pre-training with CLIP, image-to-text generation with LLaVA-like models, and text-to-video generation with DiT-based models, yields transferable and notable performance boosts, such as topping the VBench leaderboard. A comprehensive set of over 100 experiments demonstrated the suite's usability and extensibility, while also uncovering insights into the interplay between data quality, diversity, model behavior, and computational costs. All codes, datasets, and models are open-sourced to foster future research and applications that would otherwise be infeasible due to the lack of a dedicated co-development infrastructure.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2407.11784")
get_code_for_paper("2407.11784")
have("2407.11784")

Connect an agent — have() is free.