OpenMobile data collection for mobile agent training. The dataset includes four JSON training files formatted for LlamaFactory and a raw_dumps folder containing original rollout trajectories with screenshots. The dataset was uploaded by cckevinn and last updated on April 16, 2026.
Use Cases
- Training mobile agents based on rollout trajectories.
- Developing multimodal models based on screenshots and trajectory data.
- Benchmarking agent performance using structured JSON training files.
Strengths
- Includes original rollout trajectories in raw_dumps.
- Training files are formatted for the LlamaFactory framework.
- Contains multimodal data with screenshots.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Source
- huggingface
- Collection Method
- Original rollout trajectories collected for mobile agent training.
- Freshness
- Last updated 2026-04-16 03:13:06; freshness should be verified.