why the `model_type` setting in `training.yml` is `uni_o2`?

[daisy-zzy]

Thank you for your great work and open-source code!
According to the guide, to train the model from scratch, we have to run:
python scripts/train_diffusion.py configs/training.yml
However, the model_type setting in training.yml is uni_o2, which means the denoising process will use UniTransformerO2TwoUpdateGeneral instead of EGNN as the refine net. Is this a mistake or is it the common case in training? @guanjq

[PzySeere]

hi, i also want the answer to this question. could u pls share the results? :) @zooay-zzy

[ShuRenyi211]

这个应该是作者后面重新换了架构吧，不管是EGNN还是其他的一些模型架构都是希望努力挖掘口袋和配体的融合信息。我不知道这样理解是否正确。我是刚刚接触这个领域，如果有错误，请批评指正