ReactionSmiles: 3.4 Million Unique Chemical Reaction Strings

Derify compiled 3.4 million unique chemical reaction SMILES strings from multiple public sources. The dataset represents complete chemical reactions, encoding reactants, reagents/catalysts, and products in a single text string using the > delimiter. It was last updated on HuggingFace in January 2026.

Use Cases

• Pre-training chemical language models based on the 3.4M unique SMILES strings.
• Fine-tuning models for reaction outcome prediction based on the encoded reactant and reagent sequences.
• Developing sequence-to-sequence models for chemical synthesis planning based on the structured reaction format.
• Training models to generate novel reaction SMILES based on the large-scale collection of examples.

Strengths

• Contains 3.4 million unique chemical reaction SMILES strings.
• Compiled from multiple public sources, suggesting a broad scope.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.
• Freshness should be verified as the last update timestamp is in the future (2026-01-22).

Provenance

Source

Multiple public sources (unspecified).

Collection Method

Collated and compiled by author Derify.

Freshness

Last updated 2026-01-22 06:20:44.