s434749/ium_434749
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

except
Some checks failed
s434749-training/pipeline/head There was a failure building this commit
Details

Browse Source
This commit is contained in:
Alagris 2021-05-10 11:57:53 +02:00
parent 96ce6341cb
commit 16db0c341e
1 changed files with 8 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
train_model.py
View File

@ -18,13 +18,19 @@ import sys
from tqdm import tqdm
from Levenshtein import distance as levenshtein_distance
from sacred import Experiment
import traceback
ex = Experiment("CNN")
ex.observers.append(FileStorageObserver('sacred_file_observer'))
ex.observers.append(MongoObserver(url='mongodb://mongo_user:mongo_password_IUM_2021@localhost:27017',
try:
ex.observers.append(MongoObserver(url='mongodb://mongo_user:mongo_password_IUM_2021@localhost:27017',
db_name='sacred'))
except Exception as e:
traceback.print_exc()
device = torch.device('cuda:0') if torch.cuda.is_available() else torch.device('cpu')
class CNN(nn.Module):
def __init__(self, kernel_size, hidden_layers, channels, embedding_size, in_alphabet, max_len):
super(CNN, self).__init__()
@ -206,4 +212,3 @@ def run(kernel_size, hidden_layers, data_file, epochs, teacher_forcing_probabili
if mode == 'eval':
cnn.eval()
evaluate_monte_carlo(cnn, 1, data, batch_size, in_alphabet, max_len)