Open guy4261 opened 1 year ago
Right now I found a way to overcome this, but I'm still not sure what will become of the final model :sweat_smile:
def try_repr(iteration, jiteration, obj):
try:
repr(obj)
return True
except:
return False
with nlp.disable_pipes(*unaffected_pipes):
for iteration in tqdm(range(10)):
...
batches = list(minibatch(train_data, size=compounding(4.0, 32.0, 1.001)))
for batch in tqdm(batches):
lst = [_ for j, _ in enumerate(batch) if try_repr(iteration, j, _)] # <= filter out bad records
nlp.update(lst, losses=losses, drop=0.1, sgd=sgd)
Right now I found a way to overcome this, but I'm still not sure what will become of the final model :sweat_smile:
def try_repr(iteration, jiteration, obj):
try:
repr(obj)
return True
except:
return False
with nlp.disable_pipes(*unaffected_pipes):
for iteration in tqdm(range(10)):
...
batches = list(minibatch(train_data, size=compounding(4.0, 32.0, 1.001)))
for batch in tqdm(batches):
lst = [_ for j, _ in enumerate(batch) if try_repr(iteration, j, _)] # <= filter out bad records
nlp.update(lst, losses=losses, drop=0.1, sgd=sgd)
Hi! First and foremost - thank you for these notebooks!!!
In your spancat notebook, I have followed your lead and used your
data.csv
but sadly, I am getting an error. This is withspacy==3.5.3
, which is the latest version as I am writing this (June 15, 2023).In [8]
, which in your notebook is the following cell:I am getting this error right when I start running: