chemfoundationmodels / chemllmbench Goto Github PK

One of your project is to convert SMILES to chemical name. I wonder whether you have tested on the reverse.
One of my projects involves converting chemical name to SMILES. However, I find that the performance of GPT-4 is not good.
I am wondering whether you can give me some suggestions to improve the performance and accuracy.

Hi, thanks for this amazing work. Can you provide the PubChem data used in the paper? Thank you so much.

There is only Reaction Prediction, do you have any method to test models?

Hello, I am trying to test the performance of my model on chemllmbench, but I meet some issues as following:

1. For property prediction tasks, there is a lack of code for calculating the AUC-ROC metric.
2. For Name Prediction task, it seems that the previous dataset download location has become invalid.
   Could you help me resolve these issues?

Hi! Thanks for your work!

I have a question about the prompt for molecular property prediction.

In the Section 4.1 (Tasks with selectively competitive (SC) performance), it said "the prompt includes inhibit HIV replication or drugs failed clinical trials for toxicity reason, and we observed a significant decline in the performance of GPT models upon removing property labels from the prompt", but the template for the property prediction in this repository seems not including such detailed information.

So could you please explain it more detailedly? Thanks a lot!

Hi,

Could you please provide all the prompts for property prediction? It seems the only prompt in pp_example.ipynb is for the BACE dataset.

Thx

请问数据为什么下载不了了？

https://deepchem.readthedocs.io/en/latest/api_reference/moleculenet.html#uspto-datasets