2DNMRGym: 22,000+ 2D HSQC NMR Spectra for Graph Learning

More than 22,000 2D HSQC NMR spectra compiled for atom-level graph learning tasks. The dataset was created by siriusxiao and uploaded to Hugging Face on May 28, 2025. The train split is annotated by an established algorithm, providing silver-standard labels for surrogate tasks.

Use Cases

• Train graph neural networks for molecular discrimination based on 2D HSQC spectra.
• Develop models for molecular structure elucidation using silver-standard annotated NMR data.
• Benchmark machine learning algorithms on surrogate tasks defined by the dataset's annotation algorithm.

Strengths

• Contains more than 22,000 2D HSQC NMR spectra.
• Train split provides silver-standard labels generated by an established algorithm.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file format details are unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Collection Method

Likely compiled from experimental or simulated NMR data, with algorithmically generated annotations.

Freshness

Last updated 2025-05-28 20:04:05.