Hello,
I am integrating the released ConforFormer encoder into FluorCast for frozen molecular embedding extraction.
The provided training and inference code references:
dict_omol_full.txt
However, I could not find this file in either the GitHub repository or the Hugging Face checkpoint repository.
Could you please provide:
The exact dict_omol_full.txt file, preserving its original token order.
The dictionary associated with ConforFormer.pt.
The mapping between the released checkpoint files and the models reported in the paper.
The recommended checkpoint and architecture arguments for frozen molecular embedding extraction.
Confirmation of the expected hydrogen policy for that checkpoint.
The local preprocessing and direct encoder adapter are complete, and the pinned ConforFormer code now imports successfully under WSL2. I am avoiding real inference until the checkpoint and dictionary pairing can be verified.
Thank you.
Hello,
I am integrating the released ConforFormer encoder into FluorCast for frozen molecular embedding extraction.
The provided training and inference code references:
dict_omol_full.txt
However, I could not find this file in either the GitHub repository or the Hugging Face checkpoint repository.
Could you please provide:
The exact dict_omol_full.txt file, preserving its original token order.
The dictionary associated with ConforFormer.pt.
The mapping between the released checkpoint files and the models reported in the paper.
The recommended checkpoint and architecture arguments for frozen molecular embedding extraction.
Confirmation of the expected hydrogen policy for that checkpoint.
The local preprocessing and direct encoder adapter are complete, and the pinned ConforFormer code now imports successfully under WSL2. I am avoiding real inference until the checkpoint and dictionary pairing can be verified.
Thank you.