About the description-based molecule generation task #9
Description
Hello,
I noticed that while your model is capable of handling reaction-related tasks—indicating that it can generate SELFIES—it doesn't seem to address the description-based molecule generation (the opposite of molecule description generation) task.
Even though you cannot input molecule in this task (i.e., cannot use projector), this task is still necessary. In addition, baseline models like MolT5 and MoMu have experimented with this task. It seems unusual that this area hasn't been explored or addressed in your work.
Could you please share your thoughts on this? Is there a particular reason for not pursuing this task, or are there plans to include it in future work?
Thank you for your time and consideration.
Sincerely, Syzseisus