kdudzic/aitech-sd-lab
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

redid NLU task files upload

Browse Source
This commit is contained in:
Kacper Dudzic 2022-05-02 21:00:28 +02:00
parent c25548d9ea
commit 4de73a4a11
11 changed files with 2324 additions and 10871 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
NLU_lab_7-8/create_datasets.py
View File

@ -4,6 +4,10 @@ import pandas as pd
from nltk.tokenize import word_tokenize from nltk.tokenize import word_tokenize
import re import re
import random import random
import nltk
#nltk.download('punkt')
class LineContent: class LineContent:
@ -58,9 +62,11 @@ def process_file(file):
if email: if email:
email_address = email.group() email_address = email.group()
text = text.replace(email_address, '@') text = text.replace(email_address, '@')
text = text.replace("'", "")
tokens = word_tokenize(text) tokens = word_tokenize(text)
tokens = [token.replace('@', email_address) for token in tokens] tokens = [token.replace('@', email_address) for token in tokens]
else: else:
text = text.replace("'", "")
tokens = word_tokenize(text) tokens = word_tokenize(text)
return tokens return tokens
text_tokens = tokenize(text) text_tokens = tokenize(text)
@ -90,7 +96,7 @@ def process_file(file):
lines_contents = [] lines_contents = []
for _, row in df.iterrows(): for _, row in df.iterrows():
if row[0] == 'user' and row[1]: if row[0] == 'user' and row[1]:
# if row[1]: #if row[1]:
text = row[1] text = row[1]
intents = get_intents(row[2]) intents = get_intents(row[2])
slots = get_slots(row[2]) slots = get_slots(row[2])

170
NLU_lab_7-8/main.py
View File

@ -1,85 +1,85 @@
from conllu import parse_incr from conllu import parse_incr
from flair.data import Corpus, Sentence, Token from flair.data import Corpus, Sentence, Token
from flair.datasets import SentenceDataset from flair.datasets import SentenceDataset
from flair.embeddings import StackedEmbeddings from flair.embeddings import StackedEmbeddings
from flair.embeddings import WordEmbeddings from flair.embeddings import WordEmbeddings
from flair.embeddings import CharacterEmbeddings from flair.embeddings import CharacterEmbeddings
from flair.embeddings import FlairEmbeddings from flair.embeddings import FlairEmbeddings
from flair.models import SequenceTagger from flair.models import SequenceTagger
from flair.trainers import ModelTrainer from flair.trainers import ModelTrainer
import random import random
import torch import torch
from tabulate import tabulate from tabulate import tabulate
fields = ['id', 'form', 'frame', 'slot'] fields = ['id', 'form', 'frame', 'slot']
def nolabel2o(line, i): def nolabel2o(line, i):
return 'O' if line[i] == 'NoLabel' else line[i] return 'O' if line[i] == 'NoLabel' else line[i]
def conllu2flair(sentences, label=None): def conllu2flair(sentences, label=None):
fsentences = [] fsentences = []
for sentence in sentences: for sentence in sentences:
fsentence = Sentence() fsentence = Sentence()
for token in sentence: for token in sentence:
ftoken = Token(token['form']) ftoken = Token(token['form'])
if label: if label:
ftoken.add_tag(label, token[label]) ftoken.add_tag(label, token[label])
fsentence.add_token(ftoken) fsentence.add_token(ftoken)
fsentences.append(fsentence) fsentences.append(fsentence)
return SentenceDataset(fsentences) return SentenceDataset(fsentences)
def predict(model, sentence): def predict(model, sentence):
csentence = [{'form': word} for word in sentence] csentence = [{'form': word} for word in sentence]
fsentence = conllu2flair([csentence])[0] fsentence = conllu2flair([csentence])[0]
model.predict(fsentence) model.predict(fsentence)
return [(token, ftoken.get_tag('slot').value) for token, ftoken in zip(sentence, fsentence)] return [(token, ftoken.get_tag('slot').value) for token, ftoken in zip(sentence, fsentence)]
with open('train-pl-full.conllu', encoding='utf-8') as trainfile: with open('train-pl.conllu', encoding='utf-8') as trainfile:
trainset = list(parse_incr(trainfile, fields=fields, field_parsers={'slot': nolabel2o})) trainset = list(parse_incr(trainfile, fields=fields, field_parsers={'slot': nolabel2o}))
with open('test-pl-full.conllu', encoding='utf-8') as testfile: with open('test-pl.conllu', encoding='utf-8') as testfile:
testset = list(parse_incr(testfile, fields=fields, field_parsers={'slot': nolabel2o})) testset = list(parse_incr(testfile, fields=fields, field_parsers={'slot': nolabel2o}))
random.seed(42) random.seed(42)
torch.manual_seed(42) torch.manual_seed(42)
if torch.cuda.is_available(): if torch.cuda.is_available():
torch.cuda.manual_seed(0) torch.cuda.manual_seed(0)
torch.cuda.manual_seed_all(0) torch.cuda.manual_seed_all(0)
torch.backends.cudnn.enabled = False torch.backends.cudnn.enabled = False
torch.backends.cudnn.benchmark = False torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True torch.backends.cudnn.deterministic = True
corpus = Corpus(train=conllu2flair(trainset, 'slot'), test=conllu2flair(testset, 'slot')) corpus = Corpus(train=conllu2flair(trainset, 'slot'), test=conllu2flair(testset, 'slot'))
tag_dictionary = corpus.make_tag_dictionary(tag_type='slot') tag_dictionary = corpus.make_tag_dictionary(tag_type='slot')
embedding_types = [ embedding_types = [
WordEmbeddings('pl'), WordEmbeddings('pl'),
FlairEmbeddings('pl-forward'), FlairEmbeddings('pl-forward'),
FlairEmbeddings('pl-backward'), FlairEmbeddings('pl-backward'),
CharacterEmbeddings(), CharacterEmbeddings(),
] ]
embeddings = StackedEmbeddings(embeddings=embedding_types) embeddings = StackedEmbeddings(embeddings=embedding_types)
tagger = SequenceTagger(hidden_size=256, embeddings=embeddings, tagger = SequenceTagger(hidden_size=256, embeddings=embeddings,
tag_dictionary=tag_dictionary, tag_dictionary=tag_dictionary,
tag_type='slot', use_crf=True) tag_type='slot', use_crf=True)
"""
trainer = ModelTrainer(tagger, corpus) trainer = ModelTrainer(tagger, corpus)
trainer.train('slot-model-pl', trainer.train('slot-model-pl',
learning_rate=0.1, learning_rate=0.1,
mini_batch_size=32, mini_batch_size=32,
max_epochs=10, max_epochs=10,
train_with_dev=True) train_with_dev=True)
"""
try: try:
model = SequenceTagger.load('slot-model-pl/best-model.pt') model = SequenceTagger.load('slot-model-pl/best-model.pt')
except: except:
model = SequenceTagger.load('slot-model-pl/final-model.pt') model = SequenceTagger.load('slot-model-pl/final-model.pt')
print(tabulate(predict(model, 'Jeden bilet na imię Jan Kowalski na film Batman'.split()))) print(tabulate(predict(model, 'Jeden bilet na imię Jan Kowalski na film Batman'.split())))

166
NLU_lab_7-8/slot-model-pl/dev.tsv
View File

@ -1,166 +0,0 @@
poproszę O O
listę B-goal O
filmów I-goal O
granych O O
jutro B-date B-date
wieczorem B-interval I-date
wagon O O
z O O
przedziałami O O
miejsce O O
przy O O
oknie O O
23032022 O O
o B-goal B-goal
której I-goal I-goal
jest O O
na B-title B-title
noże I-title I-title
a O O
jakie O B-goal
są O I-goal
proszę O O
o O O
godzine O B-goal
20:19 B-time I-goal
a O O
jakie O B-goal
są O I-goal
dostępne O I-goal
ok O O
o O B-goal
jakich O I-goal
godzinach O I-goal
grają O O
te O O
filmy O O
wszystkie O O
dziękuję O O
witam O O
jakie B-goal B-goal
filmy I-goal I-goal
są I-goal I-goal
teraz O I-goal
w O O
kinach O O
wybieram O O
godzine O B-date
12:00 B-time I-date
29032022 O O
halo O O
halo O O
123123 O O
podaj O O
więcej O O
informacji O O
o O O
seansach O O
ok O O
z B-area B-area
tyłu I-area I-area
sali I-area O
nie I-area O
na I-area B-area
samym I-area I-area
końcu I-area I-area
na B-area B-area
środku I-area I-area
do O O
widzenia O O
a O O
z B-area B-area
przodu I-area I-area
gdzieś O O
dzień O O
dobry O O
jeden B-quantity B-quantity
normalny I-quantity O
i I-quantity O
ulgowy I-quantity O
witam O O
dzień O O
dobry O O
czy O O
sš O O
wcześniejsze B-goal B-goal
seanse I-goal I-goal
kim O O
jest O O
senior O O
3 B-quantity B-quantity
dzień O O
dobry O O
co B-goal O
można I-goal O
obejrzeć I-goal O
w O B-date
kwietniu B-interval I-date
tak O O
no O O
to O O
jakoś O O
niech O O
będzie O O
jakoś O O
to B-title O
będzie I-title O
wybieram O O
na B-title B-title
noże I-title I-title
o B-time B-time
09:30 I-time B-time
idę O O
na O O
drugą B-time B-time
na O O
batmana B-title B-title
trzy B-quantity B-quantity
bileciki O O
na B-time B-time
19:00 I-time B-time
na O O
batmana B-title B-title
na B-area B-area
środku I-area I-area
co O O
leci O O
w B-date B-date
poniedziałek I-date I-date
1 poproszę O O
2 listę B-goal O
3 filmów I-goal O
4 granych O O
5 jutro B-date B-date
6 wieczorem B-interval I-date
7 wagon O O
8 z O O
9 przedziałami O O
10 miejsce O O
11 przy O O
12 oknie O O
13 23032022 O O
14 o B-goal B-goal
15 której I-goal I-goal
16 jest O O
17 na B-title B-title
18 noże I-title I-title
19 a O O
20 jakie O B-goal
21 są O I-goal
22 proszę O O
23 o O O
24 godzine O B-goal
25 20:19 B-time I-goal
26 a O O
27 jakie O B-goal
28 są O I-goal
29 dostępne O I-goal
30 ok O O
31 o O B-goal
32 jakich O I-goal
33 godzinach O I-goal
34 grają O O
35 te O O
36 filmy O O
37 wszystkie O O
38 dziękuję O O
39 witam O O
40 jakie B-goal B-goal
41 filmy I-goal I-goal
42 są I-goal I-goal
43 teraz O I-goal
44 w O O
45 kinach O O
46 wybieram O O
47 godzine O B-date
48 12:00 B-time I-date
49 29032022 O O
50 halo O O
51 halo O O
52 123123 O O
53 podaj O O
54 więcej O O
55 informacji O O
56 o O O
57 seansach O O
58 ok O O
59 z B-area B-area
60 tyłu I-area I-area
61 sali I-area O
62 nie I-area O
63 na I-area B-area
64 samym I-area I-area
65 końcu I-area I-area
66 na B-area B-area
67 środku I-area I-area
68 do O O
69 widzenia O O
70 a O O
71 z B-area B-area
72 przodu I-area I-area
73 gdzieś O O
74 dzień O O
75 dobry O O
76 jeden B-quantity B-quantity
77 normalny I-quantity O
78 i I-quantity O
79 ulgowy I-quantity O
80 witam O O
81 dzień O O
82 dobry O O
83 czy O O
84 sš O O
85 wcześniejsze B-goal B-goal
86 seanse I-goal I-goal
87 kim O O
88 jest O O
89 senior O O
90 3 B-quantity B-quantity
91 dzień O O
92 dobry O O
93 co B-goal O
94 można I-goal O
95 obejrzeć I-goal O
96 w O B-date
97 kwietniu B-interval I-date
98 tak O O
99 no O O
100 to O O
101 jakoś O O
102 niech O O
103 będzie O O
104 jakoś O O
105 to B-title O
106 będzie I-title O
107 wybieram O O
108 na B-title B-title
109 noże I-title I-title
110 o B-time B-time
111 09:30 I-time B-time
112 idę O O
113 na O O
114 drugą B-time B-time
115 na O O
116 batmana B-title B-title
117 trzy B-quantity B-quantity
118 bileciki O O
119 na B-time B-time
120 19:00 I-time B-time
121 na O O
122 batmana B-title B-title
123 na B-area B-area
124 środku I-area I-area
125 co O O
126 leci O O
127 w B-date B-date
128 poniedziałek I-date I-date

11
NLU_lab_7-8/slot-model-pl/loss.tsv
View File

@ -1,11 +0,0 @@
EPOCH TIMESTAMP BAD_EPOCHS LEARNING_RATE TRAIN_LOSS
1 15:45:49 0 0.1000 1.107883760221383
2 15:46:06 0 0.1000 0.724391172370514
3 15:46:23 0 0.1000 0.6198507147675428
4 15:46:42 0 0.1000 0.5637349612763847
5 15:46:58 0 0.1000 0.48588470330256117
6 15:47:14 0 0.1000 0.4225153886549188
7 15:47:30 0 0.1000 0.38841035381494515
8 15:47:45 0 0.1000 0.3469537376117912
9 15:48:00 0 0.1000 0.30912264277005586
10 15:48:16 1 0.1000 0.31141101694209966
1 EPOCH TIMESTAMP BAD_EPOCHS LEARNING_RATE TRAIN_LOSS
2 1 15:45:49 0 0.1000 1.107883760221383
3 2 15:46:06 0 0.1000 0.724391172370514
4 3 15:46:23 0 0.1000 0.6198507147675428
5 4 15:46:42 0 0.1000 0.5637349612763847
6 5 15:46:58 0 0.1000 0.48588470330256117
7 6 15:47:14 0 0.1000 0.4225153886549188
8 7 15:47:30 0 0.1000 0.38841035381494515
9 8 15:47:45 0 0.1000 0.3469537376117912
10 9 15:48:00 0 0.1000 0.30912264277005586
11 10 15:48:16 1 0.1000 0.31141101694209966

1275
NLU_lab_7-8/slot-model-pl/test.tsv
View File

File diff suppressed because it is too large Load Diff

261
NLU_lab_7-8/slot-model-pl/training.log
View File

@ -1,261 +0,0 @@
2022-05-02 15:44:09,185 ----------------------------------------------------------------------------------------------------
2022-05-02 15:44:09,185 Model: "SequenceTagger(
(embeddings): StackedEmbeddings(
(list_embedding_0): WordEmbeddings('pl')
(list_embedding_1): FlairEmbeddings(
(lm): LanguageModel(
(drop): Dropout(p=0.25, inplace=False)
(encoder): Embedding(1602, 100)
(rnn): LSTM(100, 2048)
(decoder): Linear(in_features=2048, out_features=1602, bias=True)
)
)
(list_embedding_2): FlairEmbeddings(
(lm): LanguageModel(
(drop): Dropout(p=0.25, inplace=False)
(encoder): Embedding(1602, 100)
(rnn): LSTM(100, 2048)
(decoder): Linear(in_features=2048, out_features=1602, bias=True)
)
)
(list_embedding_3): CharacterEmbeddings(
(char_embedding): Embedding(275, 25)
(char_rnn): LSTM(25, 25, bidirectional=True)
)
)
(word_dropout): WordDropout(p=0.05)
(locked_dropout): LockedDropout(p=0.5)
(embedding2nn): Linear(in_features=4446, out_features=4446, bias=True)
(rnn): LSTM(4446, 256, batch_first=True, bidirectional=True)
(linear): Linear(in_features=512, out_features=50, bias=True)
(beta): 1.0
(weights): None
(weight_tensor) None
)"
2022-05-02 15:44:09,185 ----------------------------------------------------------------------------------------------------
2022-05-02 15:44:09,185 Corpus: "Corpus: 735 train + 82 dev + 152 test sentences"
2022-05-02 15:44:09,185 ----------------------------------------------------------------------------------------------------
2022-05-02 15:44:09,185 Parameters:
2022-05-02 15:44:09,185 - learning_rate: "0.1"
2022-05-02 15:44:09,185 - mini_batch_size: "32"
2022-05-02 15:44:09,185 - patience: "3"
2022-05-02 15:44:09,185 - anneal_factor: "0.5"
2022-05-02 15:44:09,185 - max_epochs: "10"
2022-05-02 15:44:09,185 - shuffle: "True"
2022-05-02 15:44:09,185 - train_with_dev: "True"
2022-05-02 15:44:09,185 - batch_growth_annealing: "False"
2022-05-02 15:44:09,185 ----------------------------------------------------------------------------------------------------
2022-05-02 15:44:09,185 Model training base path: "slot-model-pl"
2022-05-02 15:44:09,185 ----------------------------------------------------------------------------------------------------
2022-05-02 15:44:09,185 Device: cpu
2022-05-02 15:44:09,185 ----------------------------------------------------------------------------------------------------
2022-05-02 15:44:09,185 Embeddings storage mode: cpu
2022-05-02 15:44:09,212 ----------------------------------------------------------------------------------------------------
2022-05-02 15:44:12,896 epoch 1 - iter 2/26 - loss 5.40706334 - samples/sec: 17.37 - lr: 0.100000
2022-05-02 15:44:17,195 epoch 1 - iter 4/26 - loss 4.38706093 - samples/sec: 14.89 - lr: 0.100000
2022-05-02 15:44:20,984 epoch 1 - iter 6/26 - loss 3.63759864 - samples/sec: 16.90 - lr: 0.100000
2022-05-02 15:44:25,378 epoch 1 - iter 8/26 - loss 3.26681995 - samples/sec: 14.57 - lr: 0.100000
2022-05-02 15:44:29,757 epoch 1 - iter 10/26 - loss 3.05881263 - samples/sec: 14.62 - lr: 0.100000
2022-05-02 15:44:40,091 epoch 1 - iter 12/26 - loss 2.53006141 - samples/sec: 6.19 - lr: 0.100000
2022-05-02 15:44:52,707 epoch 1 - iter 14/26 - loss 1.93704781 - samples/sec: 5.07 - lr: 0.100000
2022-05-02 15:45:02,080 epoch 1 - iter 16/26 - loss 1.63138431 - samples/sec: 6.83 - lr: 0.100000
2022-05-02 15:45:14,009 epoch 1 - iter 18/26 - loss 1.40000228 - samples/sec: 5.37 - lr: 0.100000
2022-05-02 15:45:23,287 epoch 1 - iter 20/26 - loss 1.23378287 - samples/sec: 6.90 - lr: 0.100000
2022-05-02 15:45:34,691 epoch 1 - iter 22/26 - loss 1.12719827 - samples/sec: 5.61 - lr: 0.100000
2022-05-02 15:45:40,330 epoch 1 - iter 24/26 - loss 1.13188836 - samples/sec: 11.35 - lr: 0.100000
2022-05-02 15:45:49,236 epoch 1 - iter 26/26 - loss 1.10788376 - samples/sec: 7.19 - lr: 0.100000
2022-05-02 15:45:49,236 ----------------------------------------------------------------------------------------------------
2022-05-02 15:45:49,236 EPOCH 1 done: loss 1.1079 - lr 0.1000000
2022-05-02 15:45:49,236 BAD EPOCHS (no improvement): 0
2022-05-02 15:45:49,237 ----------------------------------------------------------------------------------------------------
2022-05-02 15:45:50,285 epoch 2 - iter 2/26 - loss 1.25997738 - samples/sec: 61.30 - lr: 0.100000
2022-05-02 15:45:51,915 epoch 2 - iter 4/26 - loss 0.91321364 - samples/sec: 39.27 - lr: 0.100000
2022-05-02 15:45:53,048 epoch 2 - iter 6/26 - loss 0.97971206 - samples/sec: 56.50 - lr: 0.100000
2022-05-02 15:45:54,340 epoch 2 - iter 8/26 - loss 0.87838664 - samples/sec: 49.56 - lr: 0.100000
2022-05-02 15:45:55,485 epoch 2 - iter 10/26 - loss 0.86177694 - samples/sec: 55.90 - lr: 0.100000
2022-05-02 15:45:56,488 epoch 2 - iter 12/26 - loss 0.81463133 - samples/sec: 63.85 - lr: 0.100000
2022-05-02 15:45:57,918 epoch 2 - iter 14/26 - loss 0.76334644 - samples/sec: 44.79 - lr: 0.100000
2022-05-02 15:45:59,393 epoch 2 - iter 16/26 - loss 0.78542696 - samples/sec: 43.41 - lr: 0.100000
2022-05-02 15:46:00,673 epoch 2 - iter 18/26 - loss 0.74084630 - samples/sec: 49.99 - lr: 0.100000
2022-05-02 15:46:02,245 epoch 2 - iter 20/26 - loss 0.71586100 - samples/sec: 40.74 - lr: 0.100000
2022-05-02 15:46:04,051 epoch 2 - iter 22/26 - loss 0.71469797 - samples/sec: 35.45 - lr: 0.100000
2022-05-02 15:46:05,132 epoch 2 - iter 24/26 - loss 0.71315625 - samples/sec: 59.21 - lr: 0.100000
2022-05-02 15:46:06,356 epoch 2 - iter 26/26 - loss 0.72439117 - samples/sec: 52.36 - lr: 0.100000
2022-05-02 15:46:06,356 ----------------------------------------------------------------------------------------------------
2022-05-02 15:46:06,356 EPOCH 2 done: loss 0.7244 - lr 0.1000000
2022-05-02 15:46:06,356 BAD EPOCHS (no improvement): 0
2022-05-02 15:46:06,356 ----------------------------------------------------------------------------------------------------
2022-05-02 15:46:07,363 epoch 3 - iter 2/26 - loss 0.93262965 - samples/sec: 63.62 - lr: 0.100000
2022-05-02 15:46:08,805 epoch 3 - iter 4/26 - loss 0.66342690 - samples/sec: 44.41 - lr: 0.100000
2022-05-02 15:46:10,199 epoch 3 - iter 6/26 - loss 0.69693404 - samples/sec: 45.93 - lr: 0.100000
2022-05-02 15:46:11,106 epoch 3 - iter 8/26 - loss 0.71254800 - samples/sec: 70.54 - lr: 0.100000
2022-05-02 15:46:12,661 epoch 3 - iter 10/26 - loss 0.68056002 - samples/sec: 41.17 - lr: 0.100000
2022-05-02 15:46:14,195 epoch 3 - iter 12/26 - loss 0.62003628 - samples/sec: 41.75 - lr: 0.100000
2022-05-02 15:46:15,549 epoch 3 - iter 14/26 - loss 0.62764929 - samples/sec: 47.29 - lr: 0.100000
2022-05-02 15:46:16,685 epoch 3 - iter 16/26 - loss 0.64616873 - samples/sec: 56.36 - lr: 0.100000
2022-05-02 15:46:18,469 epoch 3 - iter 18/26 - loss 0.65065601 - samples/sec: 35.88 - lr: 0.100000
2022-05-02 15:46:19,908 epoch 3 - iter 20/26 - loss 0.64878090 - samples/sec: 44.50 - lr: 0.100000
2022-05-02 15:46:21,278 epoch 3 - iter 22/26 - loss 0.63696184 - samples/sec: 46.72 - lr: 0.100000
2022-05-02 15:46:22,587 epoch 3 - iter 24/26 - loss 0.63006250 - samples/sec: 48.92 - lr: 0.100000
2022-05-02 15:46:23,866 epoch 3 - iter 26/26 - loss 0.61985071 - samples/sec: 50.08 - lr: 0.100000
2022-05-02 15:46:23,866 ----------------------------------------------------------------------------------------------------
2022-05-02 15:46:23,866 EPOCH 3 done: loss 0.6199 - lr 0.1000000
2022-05-02 15:46:23,866 BAD EPOCHS (no improvement): 0
2022-05-02 15:46:23,867 ----------------------------------------------------------------------------------------------------
2022-05-02 15:46:25,024 epoch 4 - iter 2/26 - loss 0.82507049 - samples/sec: 55.30 - lr: 0.100000
2022-05-02 15:46:26,129 epoch 4 - iter 4/26 - loss 0.78983147 - samples/sec: 57.98 - lr: 0.100000
2022-05-02 15:46:27,401 epoch 4 - iter 6/26 - loss 0.69410684 - samples/sec: 50.34 - lr: 0.100000
2022-05-02 15:46:28,974 epoch 4 - iter 8/26 - loss 0.62705834 - samples/sec: 40.69 - lr: 0.100000
2022-05-02 15:46:30,301 epoch 4 - iter 10/26 - loss 0.57534194 - samples/sec: 48.26 - lr: 0.100000
2022-05-02 15:46:32,177 epoch 4 - iter 12/26 - loss 0.55566517 - samples/sec: 34.13 - lr: 0.100000
2022-05-02 15:46:33,477 epoch 4 - iter 14/26 - loss 0.56243747 - samples/sec: 49.26 - lr: 0.100000
2022-05-02 15:46:35,204 epoch 4 - iter 16/26 - loss 0.56436807 - samples/sec: 37.07 - lr: 0.100000
2022-05-02 15:46:36,732 epoch 4 - iter 18/26 - loss 0.58195288 - samples/sec: 41.88 - lr: 0.100000
2022-05-02 15:46:38,109 epoch 4 - iter 20/26 - loss 0.58868604 - samples/sec: 46.53 - lr: 0.100000
2022-05-02 15:46:39,677 epoch 4 - iter 22/26 - loss 0.56758502 - samples/sec: 40.87 - lr: 0.100000
2022-05-02 15:46:41,433 epoch 4 - iter 24/26 - loss 0.55202777 - samples/sec: 36.45 - lr: 0.100000
2022-05-02 15:46:42,227 epoch 4 - iter 26/26 - loss 0.56373496 - samples/sec: 80.65 - lr: 0.100000
2022-05-02 15:46:42,227 ----------------------------------------------------------------------------------------------------
2022-05-02 15:46:42,227 EPOCH 4 done: loss 0.5637 - lr 0.1000000
2022-05-02 15:46:42,227 BAD EPOCHS (no improvement): 0
2022-05-02 15:46:42,228 ----------------------------------------------------------------------------------------------------
2022-05-02 15:46:43,715 epoch 5 - iter 2/26 - loss 0.45379848 - samples/sec: 43.04 - lr: 0.100000
2022-05-02 15:46:44,818 epoch 5 - iter 4/26 - loss 0.54424541 - samples/sec: 58.04 - lr: 0.100000
2022-05-02 15:46:46,795 epoch 5 - iter 6/26 - loss 0.55437849 - samples/sec: 32.38 - lr: 0.100000
2022-05-02 15:46:47,855 epoch 5 - iter 8/26 - loss 0.58448347 - samples/sec: 60.42 - lr: 0.100000
2022-05-02 15:46:49,017 epoch 5 - iter 10/26 - loss 0.57394500 - samples/sec: 55.10 - lr: 0.100000
2022-05-02 15:46:50,144 epoch 5 - iter 12/26 - loss 0.56309941 - samples/sec: 56.82 - lr: 0.100000
2022-05-02 15:46:51,022 epoch 5 - iter 14/26 - loss 0.56087045 - samples/sec: 72.92 - lr: 0.100000
2022-05-02 15:46:52,247 epoch 5 - iter 16/26 - loss 0.54126941 - samples/sec: 52.27 - lr: 0.100000
2022-05-02 15:46:53,517 epoch 5 - iter 18/26 - loss 0.54781672 - samples/sec: 50.41 - lr: 0.100000
2022-05-02 15:46:54,987 epoch 5 - iter 20/26 - loss 0.52409069 - samples/sec: 43.55 - lr: 0.100000
2022-05-02 15:46:56,416 epoch 5 - iter 22/26 - loss 0.51082819 - samples/sec: 44.84 - lr: 0.100000
2022-05-02 15:46:58,077 epoch 5 - iter 24/26 - loss 0.50232400 - samples/sec: 38.55 - lr: 0.100000
2022-05-02 15:46:58,995 epoch 5 - iter 26/26 - loss 0.48588470 - samples/sec: 69.78 - lr: 0.100000
2022-05-02 15:46:58,995 ----------------------------------------------------------------------------------------------------
2022-05-02 15:46:58,995 EPOCH 5 done: loss 0.4859 - lr 0.1000000
2022-05-02 15:46:58,995 BAD EPOCHS (no improvement): 0
2022-05-02 15:46:58,995 ----------------------------------------------------------------------------------------------------
2022-05-02 15:47:00,509 epoch 6 - iter 2/26 - loss 0.52375350 - samples/sec: 42.28 - lr: 0.100000
2022-05-02 15:47:02,103 epoch 6 - iter 4/26 - loss 0.41911038 - samples/sec: 40.18 - lr: 0.100000
2022-05-02 15:47:03,126 epoch 6 - iter 6/26 - loss 0.41424604 - samples/sec: 62.57 - lr: 0.100000
2022-05-02 15:47:04,316 epoch 6 - iter 8/26 - loss 0.39943972 - samples/sec: 53.82 - lr: 0.100000
2022-05-02 15:47:05,798 epoch 6 - iter 10/26 - loss 0.36462904 - samples/sec: 43.20 - lr: 0.100000
2022-05-02 15:47:06,774 epoch 6 - iter 12/26 - loss 0.37187295 - samples/sec: 65.60 - lr: 0.100000
2022-05-02 15:47:07,781 epoch 6 - iter 14/26 - loss 0.40622993 - samples/sec: 63.60 - lr: 0.100000
2022-05-02 15:47:08,846 epoch 6 - iter 16/26 - loss 0.42953310 - samples/sec: 60.13 - lr: 0.100000
2022-05-02 15:47:10,187 epoch 6 - iter 18/26 - loss 0.41096443 - samples/sec: 47.72 - lr: 0.100000
2022-05-02 15:47:11,212 epoch 6 - iter 20/26 - loss 0.42107760 - samples/sec: 62.50 - lr: 0.100000
2022-05-02 15:47:12,138 epoch 6 - iter 22/26 - loss 0.42309019 - samples/sec: 69.15 - lr: 0.100000
2022-05-02 15:47:13,311 epoch 6 - iter 24/26 - loss 0.42768651 - samples/sec: 54.57 - lr: 0.100000
2022-05-02 15:47:14,615 epoch 6 - iter 26/26 - loss 0.42251539 - samples/sec: 49.12 - lr: 0.100000
2022-05-02 15:47:14,615 ----------------------------------------------------------------------------------------------------
2022-05-02 15:47:14,615 EPOCH 6 done: loss 0.4225 - lr 0.1000000
2022-05-02 15:47:14,615 BAD EPOCHS (no improvement): 0
2022-05-02 15:47:14,615 ----------------------------------------------------------------------------------------------------
2022-05-02 15:47:15,953 epoch 7 - iter 2/26 - loss 0.42888915 - samples/sec: 47.86 - lr: 0.100000
2022-05-02 15:47:16,988 epoch 7 - iter 4/26 - loss 0.46386105 - samples/sec: 61.89 - lr: 0.100000
2022-05-02 15:47:17,972 epoch 7 - iter 6/26 - loss 0.45750826 - samples/sec: 65.04 - lr: 0.100000
2022-05-02 15:47:19,035 epoch 7 - iter 8/26 - loss 0.45111557 - samples/sec: 60.26 - lr: 0.100000
2022-05-02 15:47:20,138 epoch 7 - iter 10/26 - loss 0.44598492 - samples/sec: 58.08 - lr: 0.100000
2022-05-02 15:47:21,221 epoch 7 - iter 12/26 - loss 0.43062620 - samples/sec: 59.11 - lr: 0.100000
2022-05-02 15:47:22,486 epoch 7 - iter 14/26 - loss 0.43319146 - samples/sec: 50.61 - lr: 0.100000
2022-05-02 15:47:23,844 epoch 7 - iter 16/26 - loss 0.40657923 - samples/sec: 47.16 - lr: 0.100000
2022-05-02 15:47:25,007 epoch 7 - iter 18/26 - loss 0.41484192 - samples/sec: 55.05 - lr: 0.100000
2022-05-02 15:47:26,325 epoch 7 - iter 20/26 - loss 0.41555710 - samples/sec: 48.58 - lr: 0.100000
2022-05-02 15:47:27,600 epoch 7 - iter 22/26 - loss 0.40336973 - samples/sec: 50.21 - lr: 0.100000
2022-05-02 15:47:29,044 epoch 7 - iter 24/26 - loss 0.39532046 - samples/sec: 44.33 - lr: 0.100000
2022-05-02 15:47:30,078 epoch 7 - iter 26/26 - loss 0.38841035 - samples/sec: 61.93 - lr: 0.100000
2022-05-02 15:47:30,079 ----------------------------------------------------------------------------------------------------
2022-05-02 15:47:30,079 EPOCH 7 done: loss 0.3884 - lr 0.1000000
2022-05-02 15:47:30,079 BAD EPOCHS (no improvement): 0
2022-05-02 15:47:30,079 ----------------------------------------------------------------------------------------------------
2022-05-02 15:47:31,357 epoch 8 - iter 2/26 - loss 0.41543718 - samples/sec: 50.11 - lr: 0.100000
2022-05-02 15:47:32,538 epoch 8 - iter 4/26 - loss 0.32899498 - samples/sec: 54.19 - lr: 0.100000
2022-05-02 15:47:33,686 epoch 8 - iter 6/26 - loss 0.35113539 - samples/sec: 55.79 - lr: 0.100000
2022-05-02 15:47:34,725 epoch 8 - iter 8/26 - loss 0.38507402 - samples/sec: 61.58 - lr: 0.100000
2022-05-02 15:47:35,995 epoch 8 - iter 10/26 - loss 0.42831411 - samples/sec: 50.42 - lr: 0.100000
2022-05-02 15:47:37,049 epoch 8 - iter 12/26 - loss 0.39097058 - samples/sec: 60.79 - lr: 0.100000
2022-05-02 15:47:38,008 epoch 8 - iter 14/26 - loss 0.37596686 - samples/sec: 66.72 - lr: 0.100000
2022-05-02 15:47:39,462 epoch 8 - iter 16/26 - loss 0.37649604 - samples/sec: 44.05 - lr: 0.100000
2022-05-02 15:47:40,655 epoch 8 - iter 18/26 - loss 0.37892339 - samples/sec: 53.64 - lr: 0.100000
2022-05-02 15:47:42,031 epoch 8 - iter 20/26 - loss 0.35924042 - samples/sec: 46.54 - lr: 0.100000
2022-05-02 15:47:43,123 epoch 8 - iter 22/26 - loss 0.35480360 - samples/sec: 58.65 - lr: 0.100000
2022-05-02 15:47:44,286 epoch 8 - iter 24/26 - loss 0.34975662 - samples/sec: 55.03 - lr: 0.100000
2022-05-02 15:47:45,065 epoch 8 - iter 26/26 - loss 0.34695374 - samples/sec: 82.23 - lr: 0.100000
2022-05-02 15:47:45,065 ----------------------------------------------------------------------------------------------------
2022-05-02 15:47:45,065 EPOCH 8 done: loss 0.3470 - lr 0.1000000
2022-05-02 15:47:45,065 BAD EPOCHS (no improvement): 0
2022-05-02 15:47:45,066 ----------------------------------------------------------------------------------------------------
2022-05-02 15:47:46,190 epoch 9 - iter 2/26 - loss 0.25508414 - samples/sec: 56.93 - lr: 0.100000
2022-05-02 15:47:47,453 epoch 9 - iter 4/26 - loss 0.32180418 - samples/sec: 50.72 - lr: 0.100000
2022-05-02 15:47:48,461 epoch 9 - iter 6/26 - loss 0.40408790 - samples/sec: 63.48 - lr: 0.100000
2022-05-02 15:47:49,701 epoch 9 - iter 8/26 - loss 0.39779257 - samples/sec: 51.64 - lr: 0.100000
2022-05-02 15:47:51,048 epoch 9 - iter 10/26 - loss 0.36724150 - samples/sec: 47.52 - lr: 0.100000
2022-05-02 15:47:51,922 epoch 9 - iter 12/26 - loss 0.35932055 - samples/sec: 73.33 - lr: 0.100000
2022-05-02 15:47:53,117 epoch 9 - iter 14/26 - loss 0.34947437 - samples/sec: 53.57 - lr: 0.100000
2022-05-02 15:47:54,265 epoch 9 - iter 16/26 - loss 0.32652718 - samples/sec: 55.77 - lr: 0.100000
2022-05-02 15:47:55,487 epoch 9 - iter 18/26 - loss 0.32168879 - samples/sec: 52.41 - lr: 0.100000
2022-05-02 15:47:56,483 epoch 9 - iter 20/26 - loss 0.32835642 - samples/sec: 64.28 - lr: 0.100000
2022-05-02 15:47:57,790 epoch 9 - iter 22/26 - loss 0.33238740 - samples/sec: 48.98 - lr: 0.100000
2022-05-02 15:47:59,047 epoch 9 - iter 24/26 - loss 0.32465148 - samples/sec: 50.93 - lr: 0.100000
2022-05-02 15:48:00,176 epoch 9 - iter 26/26 - loss 0.30912264 - samples/sec: 56.73 - lr: 0.100000
2022-05-02 15:48:00,176 ----------------------------------------------------------------------------------------------------
2022-05-02 15:48:00,176 EPOCH 9 done: loss 0.3091 - lr 0.1000000
2022-05-02 15:48:00,176 BAD EPOCHS (no improvement): 0
2022-05-02 15:48:00,176 ----------------------------------------------------------------------------------------------------
2022-05-02 15:48:01,533 epoch 10 - iter 2/26 - loss 0.34254425 - samples/sec: 47.18 - lr: 0.100000
2022-05-02 15:48:02,801 epoch 10 - iter 4/26 - loss 0.37900189 - samples/sec: 50.52 - lr: 0.100000
2022-05-02 15:48:03,912 epoch 10 - iter 6/26 - loss 0.33156605 - samples/sec: 57.61 - lr: 0.100000
2022-05-02 15:48:05,257 epoch 10 - iter 8/26 - loss 0.30826664 - samples/sec: 47.58 - lr: 0.100000
2022-05-02 15:48:06,496 epoch 10 - iter 10/26 - loss 0.32724932 - samples/sec: 51.71 - lr: 0.100000
2022-05-02 15:48:07,790 epoch 10 - iter 12/26 - loss 0.30998078 - samples/sec: 49.46 - lr: 0.100000
2022-05-02 15:48:09,009 epoch 10 - iter 14/26 - loss 0.30504032 - samples/sec: 52.52 - lr: 0.100000
2022-05-02 15:48:10,539 epoch 10 - iter 16/26 - loss 0.28721872 - samples/sec: 41.87 - lr: 0.100000
2022-05-02 15:48:11,646 epoch 10 - iter 18/26 - loss 0.29072309 - samples/sec: 57.84 - lr: 0.100000
2022-05-02 15:48:12,706 epoch 10 - iter 20/26 - loss 0.30101217 - samples/sec: 60.40 - lr: 0.100000
2022-05-02 15:48:13,994 epoch 10 - iter 22/26 - loss 0.30494834 - samples/sec: 49.71 - lr: 0.100000
2022-05-02 15:48:15,298 epoch 10 - iter 24/26 - loss 0.31061478 - samples/sec: 49.09 - lr: 0.100000
2022-05-02 15:48:16,150 epoch 10 - iter 26/26 - loss 0.31141102 - samples/sec: 75.23 - lr: 0.100000
2022-05-02 15:48:16,150 ----------------------------------------------------------------------------------------------------
2022-05-02 15:48:16,150 EPOCH 10 done: loss 0.3114 - lr 0.1000000
2022-05-02 15:48:16,150 BAD EPOCHS (no improvement): 1
2022-05-02 15:48:46,752 ----------------------------------------------------------------------------------------------------
2022-05-02 15:48:46,768 Testing using last state of model ...
2022-05-02 15:49:06,966 0.3 0.1846 0.2286 0.1364
2022-05-02 15:49:06,967
Results:
- F-score (micro) 0.2286
- F-score (macro) 0.1296
- Accuracy 0.1364
By class:
precision recall f1-score support
quantity 0.3571 0.8333 0.5000 6
title 0.2857 0.2000 0.2353 10
goal 0.0000 0.0000 0.0000 10
time 0.4000 0.2222 0.2857 9
date 0.6667 0.5000 0.5714 4
area 0.0000 0.0000 0.0000 3
interval 0.0000 0.0000 0.0000 1
movie 0.0000 0.0000 0.0000 3
phone 0.0000 0.0000 0.0000 3
seat 0.0000 0.0000 0.0000 2
hour 0.0000 0.0000 0.0000 2
row 0.0000 0.0000 0.0000 2
ticketnumber 0.0000 0.0000 0.0000 2
name 1.0000 1.0000 1.0000 1
e-mail 0.0000 0.0000 0.0000 2
ticketsnumber 0.0000 0.0000 0.0000 1
sit_place 0.0000 0.0000 0.0000 1
email 0.0000 0.0000 0.0000 1
bankAccountNumber 0.0000 0.0000 0.0000 1
issue 0.0000 0.0000 0.0000 1
micro avg 0.3000 0.1846 0.2286 65
macro avg 0.1355 0.1378 0.1296 65
weighted avg 0.1887 0.1846 0.1725 65
samples avg 0.1364 0.1364 0.1364 65
2022-05-02 15:49:06,967 ----------------------------------------------------------------------------------------------------

0
NLU_lab_7-8/slot-model-pl/weights.txt
View File

1471
NLU_lab_7-8/test-pl-full.conllu
View File

File diff suppressed because it is too large Load Diff

1020
NLU_lab_7-8/test-pl.conllu
View File

File diff suppressed because it is too large Load Diff

5368
NLU_lab_7-8/train-pl-full.conllu
View File

File diff suppressed because it is too large Load Diff

3445
NLU_lab_7-8/train-pl.conllu
View File

File diff suppressed because it is too large Load Diff