Hanchen
Hanchen
is there a way to automatically convert the `np.float64` with `np.float32` in the `test` code...
谢谢，那我接下来就是直接提交PR来Merge，是吗？
Thanks, I am running the version of `1.4.5.1`
Thanks @steven60411. I've also changed is to the `module: torchmdnet.calculators` Hi @MaciejMajew ! I can't find the .ckpt file in the provided address, it is either included here: https://github.com/torchmd/torchmd-net/tree/main/examples/pretrained
@MaciejMajew Thanks providing the pre-training weights. I do encounter a few issues in the following steps: 1. Simulation from pre-trained weights: ``` (torchmd) [hanchen@qpu-4v100-32g0003 tutorial]$ python ./torchmd/torchmd/run.py --conf simulate.yaml Force...
@MaciejMajew Hi, Maciej. Thank you so much for the clarification! I can now use the current version of `tutorial` (with newly provided pre-trained weights} + `torchmd-net-legacy` to run the simulation,...
Hi @priceee , thank you for your question. I didn't have the data splits, as the data stored on the Cambridge's server should be erased now.. Yes we pretrain the...
Hi @YangZiz. There might be two reasons: - LMDB files are corrupted, you may need to re-download it. - The version of msgpack version is not correct, see [OcCo_Torch/Requirements_Torch.txt](https://github.com/hansen7/OcCo/blob/master/OcCo_Torch/Requirements_Torch.txt)
We are using the same data source as provided in the PointNet. The transformation of .ply format is for the blender steps.