Pykeen testing/validation loss

Question

I am unable to find out how to compute the test/validation loss, so that I could make a graph depicting the evolution of training and test/validation loss over epochs.

I have tried the pipeline approach, i.e. for TransE

result = pipeline(
  dataset = PharmKG(
    random_state = 0, 
    cache_root = f'{project_path}/data/raw/pykeen'
  ),
  model = 'TransE',
  training_loop = 'sLCWA', 
  negative_sampler = 'basic',
  epochs = 100,
  training_kwargs = dict(batch_size = 8192),
  stopper = 'EarlyStopper',  #stoppers['TransE'],
  evaluator = 'RankBasedEvaluator', 
  device = 'cuda',
  random_seed = 0,
)

as well as more direct approach

dataset = PharmKG(
  random_state = 0, 
  cache_root = f'{project_path}/data/raw/pykeen'
) 

models_dict = {
    'TransE': TransE,
    'ConvKB': ConvKB,
    'ComplEx': ComplEx
}

models = {}
optimizers = {}
training_loops = {}
loss = {}


for model_name, model_class in models_dict.items():
  models[model_name] = model_class(triples_factory=dataset.training, random_seed=0).to('cuda')  # Must be explicitely directed to GPU
  optimizers[model_name] = Adam(params = models[model_name].get_grad_params())
  training_loops[model_name] = SLCWATrainingLoop(
    triples_factory = triples_factory, 
    model = models[model_name], 
    optimizer = optimizers[model_name]
  )

  loss[model_name] = training_loops[model_name].train(
    triples_factory = triples_factory,
    num_epochs = 30,
    batch_size = 8192,
  )