Multy GPU and early stopping in TSAI

[TDL77]

/ ------------------------------------------------------------------

• Question :1

---

Please tell me, I have 2GPU
os : Windows-10-10.0.19044-SP0
python: 3.10.12
tsai: 0.3.7
fastai: 2.7.12
fastcore: 1.5.29
torch: 2.0.1
device : 2 gpus (['NVIDIA GeForce RTX 3090 Ti', 'NVIDIA GeForce RTX 3090 Ti'])
cpu cores: 16
threads per cpu : 2
RAM: 127.94 GB
GPU memory : [23.99, 23.99] GB

, before training models, I prescribe and run
import os
os.environ['CUDA_VISIBLE_DEVICES']='0:1'
%env CUDA_VISIBLE_DEVICES=0.1

but all the same, models are trained on only one GPU: 1

/ ----------------------------------------------------------------------

• Question :2

---

how to set an early stop, by valid_loss, when training models? for example in this code:
dls = TSDataLoaders.from_dsets(dsets.train,
dsets.valid,
bs=[64, 128],
batch_tfms=[TSStandardize()],
num_workers=0)

model = InceptionTime(dls.vars, dls.c)
learn = Learner(dls, model, metrics=accuracy ,cbs=PredictionDynamics())
learn.fit_one_cycle(25, lr_max=1e-3)
learn.recorder.plot_metrics()

PS dataset is large and they train for quite a long time