DatasetPlatform

L+M-24 Dataset

A benchmark dataset pairing molecules with natural-language descriptions, for models that connect chemistry and language.

Language-molecule models have emerged as an exciting direction for molecular discovery and understanding. However, training these models is challenging due to the scarcity of molecule-language pair datasets. At this point, datasets have been released which are 1) small and scraped from existing databases, 2) large but noisy and constructed by performing entity linking on the scientific literature, and 3) template-based built on prediction datasets. In this document, we detail the L+M-24 dataset, which has been created for the Language + Molecules Workshop shared task at ACL 2024. In particular, L+M-24 is designed to focus on three key benefits of natural language in molecule design: compositionality, functionality, and abstraction.

Please see the manuscript for this dataset here.

View Repository

Related Resources

AlphaSynthesis

Platform to design molecules, plan synthesis routes, and optimize chemical processes.

ChemistryBiologyMachine Learning

DataComp

First rigorous benchmark for advancing multimodal dataset creation.

Machine LearningLLM

787

FoodAtlas

A quality-controlled food knowledge graph constructed from scientific literature with large language models.

AgricultureBiologyChemistry

ICICLE Services Board

Cyberinfrastructure services for automated data labeling, image segmentation, model serving, and edge deployment, usable across any field.

Machine LearningComputer VisionBiology

Magpie

Data synthesis pipeline generating high-quality alignment data for LLMs

LLM

874

MutateEverything

MutateEverything predicts the effect of all single and double protein mutations in one forward pass. It is even versatile enough to predict higher-order mutations with minimal computational overhead.

BiologyMachine Learning

OpenCLIP

An open source implementation of OpenAI's CLIP

Machine LearningLLM

14.1k