search

kuleshov-group / caduceus

Bi-Directional Equivariant Long-Range DNA Sequence Modeling
Apache License 2.0
137 stars 14 forks source link

Something about specifying the GPU #7

Closed yongrenr closed 3 months ago

yongrenr commented 3 months ago

Hello, I have a question for you. Due to the limited number of devices on our side, I want to specify the GPU, but I still can't change the environment after setting, so where do I specify the GPU?

yair-schiff commented 3 months ago

Hi @yongrenr, are you asking about specifying the GPU type or the number of GPUs/devices for the training runs. To change the number of devices, you add the following CLI flag: trainer.devices=<NUM_GPUS>. See here, for example.

Please let me know if this resolves the confusion.

yongrenr commented 3 months ago

Thanks for the reminder, it's working normally for now

yongrenr commented 3 months ago

I also want to ask if I divide the data set into 2 parts, a test set and a training set. By training your model and then testing it, can I feed 2 data sets at the same time, or can I train with the training set first and then use ckpt to test on the test data?

yair-schiff commented 3 months ago

You can definitely pass a train and test set at the same time. Take a look at this file to see how this is done for other datasets we use. You can hopefully pattern match from these to your dataset. Let me know if you’re having any issues.

yongrenr commented 3 months ago

Thank you for your reminder. I will try it. If I don’t understand something, please give me some advice.