DOFA

weights

open(ungated CC-BY-4.0 DOFA and DOFA v2 checkpoints on Hugging Face)

data

documented-not-released(paper section 6 gives the composition — ~11.5M images across Sentinel-1, Sentinel-2, NAIP, Gaofen and EnMAP, with per-modality channel counts — and a download link per subset, but no assembled DOFA corpus and no script that rebuilds it end to end)

code

open(pretraining entry point, masked-autoencoder engine, samplers, and launch script under pretraining/)

license

CC-BY-4.0(attribution only, not OSI-approved)

The pretraining pipeline is public, the checkpoints are ungated, and the paper documents the corpus well enough to audit the mixture: section 6 Data availability statements names all five subsets with a download link each. That pairs a described corpus with an open pipeline, which is the rung the ladder writes for exactly this case. It is recorded as documented-not-released rather than open because the constituents are third-party datasets and nothing assembles them — no DOFA corpus is published and no script rebuilds one. The score is 4 on either reading. The repository LICENSE is MIT and covers the code; the weights the Hub distributes are CC-BY-4.0, and that is the license this axis scores.

• https://huggingface.co/api/models/earthflow/DOFA recorded 2026-09-10

  gated: false; private: false; license cc-by-4.0; DOFA_ViT_base_e100.pth, DOFA_ViT_large_e100.pth, dofav2_vit_base_e150.pth, and dofav2_vit_large_e150.pth present.

• https://huggingface.co/earthflow/DOFA/raw/main/README.md recorded 2026-09-10

  Card front matter and the model details both state License: CC-BY-4.0 for the released DOFA weights.

• https://api.github.com/repos/zhu-xlab/DOFA/contents/pretraining recorded 2026-09-10

  Lists main_pretrain_ofa.py, engine_pretrain.py, models_base_ofa_mae.py, train_mae_all.sh, samplers, and a datasets directory holding only a loader and wavelength tables.

• https://arxiv.org/abs/2403.15356 recorded 2026-09-10

  Abstract states that the code and pretrained weights are publicly available. It says nothing either way about the corpus, so it establishes the artifacts and not the data dimension.

• https://arxiv.org/pdf/2403.15356v3 recorded 2026-09-10

  Section 4.5 — "DOFA is pretrained on a large-scale multimodal EO dataset comprising approximately 11.5 million images from five different modalities: Sentinel-1, Sentinel-2, NAIP, Gaofen, and EnMAP". Section 6 Data availability statements gives a download link per subset (SatlasPretrain for Sentinel-1 and Sentinel-2, a named Drive folder for Gaofen, hyspecnet.rsim.berlin for EnMAP).

The Hugging Face mirror reports zero downloads because the checkpoints are pulled as raw files, so 211 GitHub stars are used as a last-resort reach signal.

• https://api.github.com/repos/zhu-xlab/DOFA recorded 2026-09-10

  stargazers_count: 211; archived: false; fork: false.

Two attributes are stated directly. The pretraining objective is masked autoencoding, which the category does not count as generative, and the sensor diversity is already counted as multiple instruments rather than a second time as multiple scales.

• https://raw.githubusercontent.com/zhu-xlab/DOFA/master/README.md recorded 2026-09-10

  Documents GeoBench classification linear probing and UPerNet segmentation fine-tuning under downstream_tasks, and shows one loaded model run over Sentinel-1 SAR, Sentinel-2, and NAIP RGB inputs with arbitrary channel counts.

• https://arxiv.org/abs/2403.15356 recorded 2026-09-10

  Describes a wavelength-conditioned dynamic hypernetwork processing inputs from five distinct satellite sensors and continual pretraining on five EO modalities.