Molmo2-ER-SIMS-VSI

# Molmo2-ER · ellisbrown/SIMS-VSI 203K simulated video spatial QA pairs from AI2-THOR/ProcTHOR agent trajectories. This is a re-hosted, **loader-ready subset** of the upstream dataset, used to train [`allenai/Molmo2-ER-4B`](https://huggingface.co/allenai/Molmo2-ER-4B). Files mirror the upstream layout; nothing in the data has been modified. ## Upstream source - **Original dataset:** [ellisbrown/SIMS-VSI](https://huggingface.co/datasets/ellisbrown/SIMS-VSI) - **Paper:** *SIMS-V: Simulated Instruction-Tuning for Spatial Video Understanding* ([arXiv:2511.04668](https://arxiv.org/abs/2511.04668)) - **License:** `apache-2.0` (inherits from upstream) If you use this data, please cite the original authors: ```bibtex @misc{brown2025simsvsimulatedinstructiontuningspatial, title={SIMS-V: Simulated Instruction-Tuning for Spatial Video Understanding}, author={Ellis Brown and Arijit Ray and Ranjay Krishna and others}, year={2025}, eprint={2511.04668}, archivePrefix={arXiv} } ``` ## Extracting before training This release ships archives. Extract them in-place before pointing `SPATIAL_DATA_HOME` at this directory: ```bash zstd -d annos_shard_000.tar.zst && tar -xf annos_shard_000.tar for f in video_shard_*.tar.zst; do zstd -d $f -o ${f%.zst}; done for f in video_shard_*.tar; do tar -xf $f; done # Loader expects ./data/<shards>/... ``` ## Usage in Molmo2-ER See the [`allenai/molmo2`](https://github.com/allenai/molmo2) repository for the data loader and training recipe. The relevant loader class for this dataset lives in `olmo/data/spatial_datasets.py`.