Reaction outcome prediction with AI models could be a powerful capability in workflows, including the exciting prospect of discovering new reactions!
But how well do they work on more realistic data?
Check the preprint to find out!
https://arxiv.org/abs/2501.06669
and see Connor’s long LinkedIn post:
But how well do they work on more realistic data?
Check the preprint to find out!
https://arxiv.org/abs/2501.06669
and see Connor’s long LinkedIn post:
Comments
https://www.linkedin.com/posts/connorcoley_we-and-others-have-spent-a-long-time-thinking-activity-7284923698220826625-CB4N?utm_medium=ios_app&utm_source=social_share_sheet&utm_campaign=copy_link
Joint work with John Bradshaw, Anji Zhang, Babak Mahjour, David Graff and Connor Coley
But really interesting benchmark, will give a couple of ideas I've had a go on this... if I ever have time again to code 😉