weight

[leobince]

sorry. is the trained weight provided, or we have to train the model by ourselves.

[Z-MU-Z]

same question!

[zhengzx-nlp]

Hello @leobince @Z-MU-Z, thank you for your interest in our work! As for your question, I will be working on providing model weights in an easier way shortly. Plus, reproducing model training should also be fairly easy and and we warmly encourage you to also have a try. 😊

Zaixiang

[BSharmi]

Hey Zaixiang!

I had the same question! I will try to train the model. But it would be super helpful if we have the weights as well :)

Thank you!

[BSharmi]

Actually I was able to train the model on CATH4.2!

[zhengzx-nlp]

Hey everyone seeing this thread, I have updated readme to include links to pretrained model weights. If there are any problems, please feel free to let me know!

Best Zaixiang

[benyaminjami]

Hi Zaixiang, Thank you for the update. Are you going to release the weights for the multichain dataset? I was going to train the model myself, but considering the dataset size, I may be unable to do it.

[BSharmi]

Hey Zaixiang!

Thanks so much for providing the weights. I noticed that the zip files had the weights but not the .yaml files which are needed to load the trained model. Is that the one in https://github.com/BytedProtein/ByProt/blob/main/configs/config.yaml

Thank you!

[zhengzx-nlp]

hi @BSharmi, actually it should have the yaml file but in .hydra folder which is invisible to GUI.

[zhengzx-nlp]

Hi @benyaminjami. I will upload the model weights for multichain shortly. I also needed to retrain ones to ensure the models got fully tested before uploading. Thank you for your interest!

Hi Zaixiang, Thank you for the update. Are you going to release the weights for the multichain dataset? I was going to train the model myself, but considering the dataset size, I may be unable to do it.

[BSharmi]

gotcha thank you!

[zhengzx-nlp]

Hi @benyaminjami I've uploaded the pretrained weight for lm_design_esm2_650m for multichain dataset. Please give it a try!