SamsungSAILMontreal/Conjugated-xTB_2M_molecules

这是一个包含200万个有机发光二极管(OLED)分子数据的Conjugated-xTB数据集，来源于arxiv.org/abs/2502.14842论文。数据集中的字段包括振荡强度（与亮度相关）f_osc，吸收波长wavelength，以及其他分子相关信息。振荡强度越大，OLED的亮度越高。吸收波长在1000nm及以上对应短波红外吸收范围，这对生物医学成像非常重要，因为在这一波长范围内，组织的吸收和散射相对较低，允许光线更深地穿透。该数据集非常适合训练生成模型或强化学习代理，以最大化振荡强度。同时，提供了一些代码来评估新分子的振荡强度和波长。

This is the Conjugated-xTB dataset of 2M OLED molecules from the paper arxiv.org/abs/2502.14842. The dataset includes fields such as oscillator strength (f_osc, correlated with brightness) which should be maximized for bright OLEDs, and absorption wavelength (wavelength) with values >=1000nm indicating short-wave infrared absorption range, which is important for biomedical imaging due to low absorption and scattering in tissues, allowing for deeper light penetration. The dataset is suitable for training generative models or RL agents to maximize the oscillator strength, and code is provided to evaluate the oscillator strength and wavelength of new molecules.