--- language: - en license: cc-by-4.0 pretty_name: CrysMTM v2 dataset_info: - config_name: augmented features: - name: structure_id dtype: string - name: sample_id dtype: string - name: phase dtype: string - name: temperature dtype: int32 - name: split dtype: string - name: species list: string - name: atomic_numbers list: int32 - name: positions list: list: float64 length: 3 - name: xyz dtype: string - name: text dtype: string - name: text_kind dtype: string - name: image dtype: image - name: targets struct: - name: HOMO dtype: float64 - name: LUMO dtype: float64 - name: Eg dtype: float64 - name: Ef dtype: float64 - name: Et dtype: float64 - name: Eta dtype: float64 - name: disp dtype: float64 - name: vol dtype: float64 - name: bond dtype: float64 - name: units struct: - name: HOMO dtype: string - name: LUMO dtype: string - name: Eg dtype: string - name: Ef dtype: string - name: Et dtype: string - name: Eta dtype: string - name: disp dtype: string - name: vol dtype: string - name: bond dtype: string - name: positions dtype: string - name: rotation_matrix list: list: float64 length: 3 length: 3 - name: rotation_center list: float64 length: 3 - name: paper_rotations list: list: list: float64 length: 3 length: 3 - name: source_archive_sha256 dtype: string - name: source_xyz_path dtype: string - name: source_xyz_sha256 dtype: string - name: source_image_sha256 dtype: string - name: source_text_sha256 dtype: string - name: recipe dtype: string splits: - name: train num_bytes: 41716226 num_examples: 1416 - name: validation num_bytes: 10427359 num_examples: 354 - name: test_id num_bytes: 40651320 num_examples: 1380 - name: test_ood num_bytes: 194093960 num_examples: 6588 download_size: 223385489 dataset_size: 286888865 - config_name: core features: - name: structure_id dtype: string - name: sample_id dtype: string - name: phase dtype: string - name: temperature dtype: int32 - name: split dtype: string - name: species list: string - name: atomic_numbers list: int32 - name: positions list: list: float64 length: 3 - name: xyz dtype: string - name: text dtype: string - name: text_kind dtype: string - name: image dtype: image - name: targets struct: - name: HOMO dtype: float64 - name: LUMO dtype: float64 - name: Eg dtype: float64 - name: Ef dtype: float64 - name: Et dtype: float64 - name: Eta dtype: float64 - name: disp dtype: float64 - name: vol dtype: float64 - name: bond dtype: float64 - name: units struct: - name: HOMO dtype: string - name: LUMO dtype: string - name: Eg dtype: string - name: Ef dtype: string - name: Et dtype: string - name: Eta dtype: string - name: disp dtype: string - name: vol dtype: string - name: bond dtype: string - name: positions dtype: string - name: rotation_matrix list: list: float64 length: 3 length: 3 - name: rotation_center list: float64 length: 3 - name: paper_rotations list: list: list: float64 length: 3 length: 3 - name: source_archive_sha256 dtype: string - name: source_xyz_path dtype: string - name: source_xyz_sha256 dtype: string - name: source_image_sha256 dtype: string - name: source_text_sha256 dtype: string - name: recipe dtype: string splits: - name: train num_bytes: 807310 num_examples: 24 - name: validation num_bytes: 201861 num_examples: 6 - name: test_id num_bytes: 540273 num_examples: 15 - name: test_ood num_bytes: 1003261 num_examples: 18 download_size: 2299058 dataset_size: 2552705 configs: - config_name: augmented data_files: - split: train path: augmented/train-* - split: validation path: augmented/validation-* - split: test_id path: augmented/test_id-* - split: test_ood path: augmented/test_ood-* - config_name: core data_files: - split: train path: core/train-* - split: validation path: core/validation-* - split: test_id path: core/test_id-* - split: test_ood path: core/test_ood-* default: true tags: - chemistry - materials-science - crystalline-materials - multimodal --- # CrysMTM v2 A TiO₂ dataset with 63 original structures and 9,738 archived views, exposed as standard Hugging Face datasets with corrected, disjoint parent-level splits. ```python from datasets import load_dataset ds = load_dataset("kurbanintelligencelab/CrysMTM", "core", revision="v2.0.2") print(ds["train"][0]["targets"]) ``` Install `datasets[vision]` for decoded images. Select `augmented` for all archived views. Both configurations use the same `temperature-v2` protocol. Supported splits are `train`, `validation`, `test_id`, and `test_ood`; streaming is supported. | Configuration | Train | Validation | ID test | OOD test | Total | | --- | ---: | ---: | ---: | ---: | ---: | | core | 24 | 6 | 15 | 18 | 63 | | augmented | 1416 | 354 | 1380 | 6588 | 9738 | ## Contents Each record contains IDs, phase, temperature (K), species, atomic numbers, positions (angstrom), XYZ text, an embedded image, text, nine named targets with source units, source hashes, a rotation matrix and its center. `core` also contains `paper_rotations`, the recovered matrices for replaying every archived orientation. Each parent's rotated views share its physical labels. Electronic targets `HOMO`, `LUMO`, `Eg`, `Ef`, `Et` and `Eta` are in eV as labeled by the source. `disp` and `bond` are in Å, and `vol` in ų. Structures use Cartesian XYZ coordinates. `text_kind=legacy-descriptors-unrevalidated` identifies original archive descriptions. Source descriptor definitions are documented in the paper; v2 validation covers data integrity and rotations. Select task inputs separately from labels and fit normalization on training parents. Targets are returned in their source units. ## Splits - Train: 150, 200, 300, 400, 500, 600, 700, 850 K. - Validation: 350, 550 K. - ID test: 250, 450, 650, 750, 800 K. - OOD test: 0, 50, 100, 900, 950, 1000 K. Every temperature assignment applies to all three phases. Every view remains with its parent. The original repository's training and OOD lists overlapped at 0, 50 and 100 K; the README's 13,032 total counted overlapping memberships. This release contains 9,738 distinct view IDs. Report benchmark results with their protocol version. This release provides the `temperature-v2` data and split definitions. ## Custom generation The CrysMTM toolkit supports arbitrary rotation counts per parent, seeds, explicit Euler angles or matrices, optional rendering, export and a `paper` orientation preset. The preset replays the recovered archive matrices around the origin, including identity, to coordinate precision. Generated images and text use the v2 renderer and geometric template. Use `augmented` for the original archived XYZ files, images and descriptions. Install the [versioned wheel](toolkit/crysmtm-2.0.2-py3-none-any.whl) with pip, or download the [source distribution](toolkit/crysmtm-2.0.2.tar.gz). The toolkit source is available in the [GitHub repository](https://github.com/KurbanIntelligenceLab/CrysMTM). ```bash pip install "https://huggingface.co/datasets/kurbanintelligencelab/CrysMTM/resolve/v2.0.2/toolkit/crysmtm-2.0.2-py3-none-any.whl" crysmtm fetch --output ./data crysmtm generate ./data --rotations-per-structure 1000 --seed 42 --output ./generated ``` Generation defaults to XYZ/coordinates and geometric text. Install the optional rendering extra from the wheel to generate images. See the source README for rendering, explicit angles/matrices, phase/temperature filters and the `paper` preset. ## Provenance and license Converted from [Figshare v1](https://doi.org/10.6084/m9.figshare.29497604.v1), archive MD5 `74315f19453b7e2f0f76855046707e81`. SHA-256 and the full validation summary are in `audit.json`. The data retains CC BY 4.0 and attribution to Can Polat, Erchin Serpedin, Mustafa Kurban and Hasan Kurban. The original repository snapshot is `3aa724da3c7ac64bbbe79dcf6dfbfc275fad5507`. Cite the paper and pin the dataset revision used in an experiment. ## Citation ```bibtex @article{polat2025crysmtm, title={CrysMTM: a multiphase, temperature-resolved, multimodal dataset for crystalline materials}, author={Polat, Can and Serpedin, Erchin and Kurban, Mustafa and Kurban, Hasan}, journal={Machine Learning: Science and Technology}, volume={6}, number={3}, pages={030603}, year={2025}, publisher={IOP Publishing} } ```