id	sid	tid	token	lemma	pos
bracis-19080	1	1	comparing	compare	VERB
bracis-19080	1	2	contextual	contextual	ADJ
bracis-19080	1	3	embeddings	embedding	NOUN
bracis-19080	1	4	for	for	ADP
bracis-19080	1	5	semantic	semantic	ADJ
bracis-19080	1	6	textual	textual	ADJ
bracis-19080	1	7	similarity	similarity	NOUN
bracis-19080	1	8	in	in	ADP
bracis-19080	1	9	portuguese	portuguese	ADJ
bracis-19080	1	10	|	|	ADV
bracis-19080	1	11	springer	springer	NOUN
bracis-19080	1	12	nature	nature	PROPN
bracis-19080	1	13	link	link	PROPN
bracis-19080	1	14	(	(	PUNCT
bracis-19080	1	15	formerly	formerly	ADV
bracis-19080	1	16	springerlink	springerlink	NOUN
bracis-19080	1	17	)	)	PUNCT
bracis-19080	1	18	skip	skip	VERB
bracis-19080	1	19	to	to	ADP
bracis-19080	1	20	main	main	ADJ
bracis-19080	1	21	content	content	NOUN
bracis-19080	1	22	advertisement	advertisement	NOUN
bracis-19080	1	23	log	log	NOUN
bracis-19080	1	24	in	in	ADP
bracis-19080	1	25	menu	menu	NOUN
bracis-19080	1	26	find	find	VERB
bracis-19080	1	27	a	a	DET
bracis-19080	1	28	journal	journal	NOUN
bracis-19080	1	29	publish	publish	VERB
bracis-19080	1	30	with	with	ADP
bracis-19080	1	31	us	we	PRON
bracis-19080	1	32	track	track	VERB
bracis-19080	1	33	your	your	PRON
bracis-19080	1	34	research	research	NOUN
bracis-19080	1	35	search	search	NOUN
bracis-19080	1	36	cart	cart	NOUN
bracis-19080	1	37	home	home	NOUN
bracis-19080	1	38	intelligent	intelligent	ADJ
bracis-19080	1	39	systems	system	NOUN
bracis-19080	1	40	conference	conference	NOUN
bracis-19080	1	41	paper	paper	NOUN
bracis-19080	1	42	comparing	compare	VERB
bracis-19080	1	43	contextual	contextual	ADJ
bracis-19080	1	44	embeddings	embedding	NOUN
bracis-19080	1	45	for	for	ADP
bracis-19080	1	46	semantic	semantic	ADJ
bracis-19080	1	47	textual	textual	ADJ
bracis-19080	1	48	similarity	similarity	NOUN
bracis-19080	1	49	in	in	ADP
bracis-19080	1	50	portuguese	portuguese	ADJ
bracis-19080	1	51	conference	conference	NOUN
bracis-19080	1	52	paper	paper	NOUN
bracis-19080	1	53	first	first	ADV
bracis-19080	1	54	online	online	ADV
bracis-19080	1	55	:	:	PUNCT
bracis-19080	1	56	28	28	NUM
bracis-19080	1	57	november	november	PROPN
bracis-19080	1	58	2021	2021	NUM
bracis-19080	2	1	pp	pp	CCONJ
bracis-19080	2	2	389–404	389–404	NUM
bracis-19080	2	3	cite	cite	VERB
bracis-19080	2	4	this	this	DET
bracis-19080	2	5	conference	conference	NOUN
bracis-19080	2	6	paper	paper	NOUN
bracis-19080	2	7	access	access	NOUN
bracis-19080	2	8	provided	provide	VERB
bracis-19080	2	9	by	by	ADP
bracis-19080	2	10	university	university	PROPN
bracis-19080	2	11	of	of	ADP
bracis-19080	2	12	notre	notre	PROPN
bracis-19080	2	13	dame	dame	PROPN
bracis-19080	2	14	hesburgh	hesburgh	PROPN
bracis-19080	2	15	library	library	PROPN
bracis-19080	2	16	download	download	PROPN
bracis-19080	2	17	book	book	NOUN
bracis-19080	2	18	pdf	pdf	PROPN
bracis-19080	2	19	download	download	NOUN
bracis-19080	2	20	book	book	NOUN
bracis-19080	2	21	epub	epub	PROPN
bracis-19080	2	22	intelligent	intelligent	ADJ
bracis-19080	2	23	systems	system	NOUN
bracis-19080	2	24	(	(	PUNCT
bracis-19080	2	25	bracis	bracis	NOUN
bracis-19080	2	26	2021	2021	NUM
bracis-19080	2	27	)	)	PUNCT
bracis-19080	2	28	comparing	compare	VERB
bracis-19080	2	29	contextual	contextual	ADJ
bracis-19080	2	30	embeddings	embedding	NOUN
bracis-19080	2	31	for	for	ADP
bracis-19080	2	32	semantic	semantic	ADJ
bracis-19080	2	33	textual	textual	ADJ
bracis-19080	2	34	similarity	similarity	NOUN
bracis-19080	2	35	in	in	ADP
bracis-19080	2	36	portuguese	portuguese	ADJ
bracis-19080	2	37	download	download	NOUN
bracis-19080	2	38	book	book	NOUN
bracis-19080	2	39	pdf	pdf	PROPN
bracis-19080	2	40	download	download	NOUN
bracis-19080	2	41	book	book	PROPN
bracis-19080	2	42	epub	epub	PROPN
bracis-19080	2	43	josé	josé	PROPN
bracis-19080	2	44	e.	e.	PROPN
bracis-19080	2	45	andrade	andrade	PROPN
bracis-19080	2	46	junior10,11	junior10,11	PROPN
bracis-19080	2	47	,	,	PUNCT
bracis-19080	2	48	jonathan	jonathan	PROPN
bracis-19080	2	49	cardoso	cardoso	PROPN
bracis-19080	2	50	-	-	PUNCT
bracis-19080	2	51	silva12,13	silva12,13	PROPN
bracis-19080	2	52	&	&	CCONJ
bracis-19080	2	53	leonardo	leonardo	PROPN
bracis-19080	2	54	c.	c.	PROPN
bracis-19080	2	55	t.	t.	PROPN
bracis-19080	2	56	bezerra10	bezerra10	PROPN
bracis-19080	2	57	  	  	SPACE
bracis-19080	2	58	part	part	NOUN
bracis-19080	2	59	of	of	ADP
bracis-19080	2	60	the	the	DET
bracis-19080	2	61	book	book	NOUN
bracis-19080	2	62	series	series	NOUN
bracis-19080	2	63	:	:	PUNCT
bracis-19080	2	64	lecture	lecture	NOUN
bracis-19080	2	65	notes	note	NOUN
bracis-19080	2	66	in	in	ADP
bracis-19080	2	67	computer	computer	NOUN
bracis-19080	2	68	science	science	NOUN
bracis-19080	2	69	(	(	PUNCT
bracis-19080	2	70	(	(	PUNCT
bracis-19080	2	71	lnai	lnai	ADJ
bracis-19080	2	72	,	,	PUNCT
bracis-19080	2	73	volume	volume	NOUN
bracis-19080	2	74	13074	13074	NUM
bracis-19080	2	75	)	)	PUNCT
bracis-19080	2	76	)	)	PUNCT
bracis-19080	2	77	included	include	VERB
bracis-19080	2	78	in	in	ADP
bracis-19080	2	79	the	the	DET
bracis-19080	2	80	following	follow	VERB
bracis-19080	2	81	conference	conference	NOUN
bracis-19080	2	82	series	series	NOUN
bracis-19080	2	83	:	:	PUNCT
bracis-19080	2	84	brazilian	brazilian	ADJ
bracis-19080	2	85	conference	conference	NOUN
bracis-19080	2	86	on	on	ADP
bracis-19080	2	87	intelligent	intelligent	ADJ
bracis-19080	2	88	systems	system	NOUN
bracis-19080	2	89	1281	1281	NUM
bracis-19080	2	90	accesses	access	VERB
bracis-19080	2	91	1	1	NUM
bracis-19080	2	92	citation	citation	NOUN
bracis-19080	2	93	abstract	abstract	ADJ
bracis-19080	2	94	semantic	semantic	ADJ
bracis-19080	2	95	textual	textual	ADJ
bracis-19080	2	96	similarity	similarity	NOUN
bracis-19080	2	97	(	(	PUNCT
bracis-19080	2	98	sts	st	NOUN
bracis-19080	2	99	)	)	PUNCT
bracis-19080	2	100	measures	measure	NOUN
bracis-19080	2	101	how	how	SCONJ
bracis-19080	2	102	semantically	semantically	ADV
bracis-19080	2	103	similar	similar	ADJ
bracis-19080	2	104	two	two	NUM
bracis-19080	2	105	sentences	sentence	NOUN
bracis-19080	2	106	are	be	AUX
bracis-19080	2	107	.	.	PUNCT
bracis-19080	3	1	in	in	ADP
bracis-19080	3	2	the	the	DET
bracis-19080	3	3	context	context	NOUN
bracis-19080	3	4	of	of	ADP
bracis-19080	3	5	the	the	DET
bracis-19080	3	6	portuguese	portuguese	ADJ
bracis-19080	3	7	language	language	NOUN
bracis-19080	3	8	,	,	PUNCT
bracis-19080	3	9	sts	st	NOUN
bracis-19080	3	10	literature	literature	NOUN
bracis-19080	3	11	is	be	AUX
bracis-19080	3	12	still	still	ADV
bracis-19080	3	13	incipient	incipient	ADJ
bracis-19080	3	14	but	but	CCONJ
bracis-19080	3	15	includes	include	VERB
bracis-19080	3	16	important	important	ADJ
bracis-19080	3	17	initiatives	initiative	NOUN
bracis-19080	3	18	like	like	ADP
bracis-19080	3	19	the	the	DET
bracis-19080	3	20	assin	assin	PROPN
bracis-19080	3	21	and	and	CCONJ
bracis-19080	3	22	assin	assin	PROPN
bracis-19080	3	23	 	 	SPACE
bracis-19080	3	24	2	2	NUM
bracis-19080	3	25	shared	share	VERB
bracis-19080	3	26	tasks	task	NOUN
bracis-19080	3	27	.	.	PUNCT
bracis-19080	4	1	the	the	DET
bracis-19080	4	2	state	state	NOUN
bracis-19080	4	3	-	-	PUNCT
bracis-19080	4	4	of	of	ADP
bracis-19080	4	5	-	-	PUNCT
bracis-19080	4	6	the	the	DET
bracis-19080	4	7	-	-	PUNCT
bracis-19080	4	8	art	art	NOUN
bracis-19080	4	9	for	for	ADP
bracis-19080	4	10	those	those	DET
bracis-19080	4	11	datasets	dataset	NOUN
bracis-19080	4	12	is	be	AUX
bracis-19080	4	13	a	a	DET
bracis-19080	4	14	contextual	contextual	ADJ
bracis-19080	4	15	embedding	embed	VERB
bracis-19080	4	16	produced	produce	VERB
bracis-19080	4	17	by	by	ADP
bracis-19080	4	18	a	a	DET
bracis-19080	4	19	portuguese	portuguese	ADJ
bracis-19080	4	20	pre	pre	VERB
bracis-19080	4	21	-	-	VERB
bracis-19080	4	22	trained	trained	ADJ
bracis-19080	4	23	and	and	CCONJ
bracis-19080	4	24	fine	fine	ADV
bracis-19080	4	25	-	-	PUNCT
bracis-19080	4	26	tuned	tune	VERB
bracis-19080	4	27	bert	bert	PROPN
bracis-19080	4	28	model	model	NOUN
bracis-19080	4	29	.	.	PUNCT
bracis-19080	5	1	in	in	ADP
bracis-19080	5	2	this	this	DET
bracis-19080	5	3	work	work	NOUN
bracis-19080	5	4	,	,	PUNCT
bracis-19080	5	5	we	we	PRON
bracis-19080	5	6	investigate	investigate	VERB
bracis-19080	5	7	the	the	DET
bracis-19080	5	8	application	application	NOUN
bracis-19080	5	9	of	of	ADP
bracis-19080	5	10	sentence	sentence	NOUN
bracis-19080	5	11	-	-	PUNCT
bracis-19080	5	12	bert	bert	NOUN
bracis-19080	5	13	 	 	SPACE
bracis-19080	5	14	(	(	PUNCT
bracis-19080	5	15	sbert	sbert	NOUN
bracis-19080	5	16	)	)	PUNCT
bracis-19080	5	17	contextual	contextual	ADJ
bracis-19080	5	18	embeddings	embedding	NOUN
bracis-19080	5	19	to	to	ADP
bracis-19080	5	20	these	these	DET
bracis-19080	5	21	datasets	dataset	NOUN
bracis-19080	5	22	.	.	PUNCT
bracis-19080	6	1	compared	compare	VERB
bracis-19080	6	2	to	to	ADP
bracis-19080	6	3	bert	bert	PROPN
bracis-19080	6	4	,	,	PUNCT
bracis-19080	6	5	sbert	sbert	NOUN
bracis-19080	6	6	is	be	AUX
bracis-19080	6	7	a	a	DET
bracis-19080	6	8	more	more	ADV
bracis-19080	6	9	computationally	computationally	ADV
bracis-19080	6	10	efficient	efficient	ADJ
bracis-19080	6	11	approach	approach	NOUN
bracis-19080	6	12	,	,	PUNCT
bracis-19080	6	13	enabling	enable	VERB
bracis-19080	6	14	its	its	PRON
bracis-19080	6	15	application	application	NOUN
bracis-19080	6	16	to	to	ADP
bracis-19080	6	17	scalable	scalable	ADJ
bracis-19080	6	18	unsupervised	unsupervised	ADJ
bracis-19080	6	19	learning	learning	NOUN
bracis-19080	6	20	problems	problem	NOUN
bracis-19080	6	21	.	.	PUNCT
bracis-19080	7	1	given	give	VERB
bracis-19080	7	2	the	the	DET
bracis-19080	7	3	absence	absence	NOUN
bracis-19080	7	4	of	of	ADP
bracis-19080	7	5	sbert	sbert	NOUN
bracis-19080	7	6	models	model	NOUN
bracis-19080	7	7	pre	pre	VERB
bracis-19080	7	8	-	-	VERB
bracis-19080	7	9	trained	train	VERB
bracis-19080	7	10	in	in	ADP
bracis-19080	7	11	portuguese	portuguese	PROPN
bracis-19080	7	12	and	and	CCONJ
bracis-19080	7	13	the	the	DET
bracis-19080	7	14	computational	computational	ADJ
bracis-19080	7	15	cost	cost	NOUN
bracis-19080	7	16	for	for	ADP
bracis-19080	7	17	such	such	ADJ
bracis-19080	7	18	training	training	NOUN
bracis-19080	7	19	,	,	PUNCT
bracis-19080	7	20	we	we	PRON
bracis-19080	7	21	adopt	adopt	VERB
bracis-19080	7	22	multilingual	multilingual	ADJ
bracis-19080	7	23	models	model	NOUN
bracis-19080	7	24	and	and	CCONJ
bracis-19080	7	25	also	also	ADV
bracis-19080	7	26	fine	fine	ADJ
bracis-19080	7	27	-	-	PUNCT
bracis-19080	7	28	tune	tune	VERB
bracis-19080	7	29	them	they	PRON
bracis-19080	7	30	for	for	ADP
bracis-19080	7	31	portuguese	portuguese	PROPN
bracis-19080	7	32	.	.	PUNCT
bracis-19080	8	1	results	result	NOUN
bracis-19080	8	2	showed	show	VERB
bracis-19080	8	3	that	that	SCONJ
bracis-19080	8	4	sbert	sbert	NOUN
bracis-19080	8	5	embeddings	embedding	NOUN
bracis-19080	8	6	were	be	AUX
bracis-19080	8	7	competitive	competitive	ADJ
bracis-19080	8	8	especially	especially	ADV
bracis-19080	8	9	after	after	ADP
bracis-19080	8	10	fine	fine	ADV
bracis-19080	8	11	-	-	PUNCT
bracis-19080	8	12	tuning	tuning	NOUN
bracis-19080	8	13	,	,	PUNCT
bracis-19080	8	14	numerically	numerically	ADV
bracis-19080	8	15	surpassing	surpass	VERB
bracis-19080	8	16	the	the	DET
bracis-19080	8	17	results	result	NOUN
bracis-19080	8	18	of	of	ADP
bracis-19080	8	19	bert	bert	PROPN
bracis-19080	8	20	on	on	ADP
bracis-19080	8	21	assin	assin	PROPN
bracis-19080	8	22	 	 	SPACE
bracis-19080	8	23	2	2	NUM
bracis-19080	8	24	and	and	CCONJ
bracis-19080	8	25	the	the	DET
bracis-19080	8	26	results	result	NOUN
bracis-19080	8	27	observed	observe	VERB
bracis-19080	8	28	during	during	ADP
bracis-19080	8	29	the	the	DET
bracis-19080	8	30	shared	share	VERB
bracis-19080	8	31	tasks	task	NOUN
bracis-19080	8	32	for	for	ADP
bracis-19080	8	33	all	all	DET
bracis-19080	8	34	datasets	dataset	NOUN
bracis-19080	8	35	considered	consider	VERB
bracis-19080	8	36	.	.	PUNCT
bracis-19080	9	1	access	access	NOUN
bracis-19080	9	2	provided	provide	VERB
bracis-19080	9	3	by	by	ADP
bracis-19080	9	4	university	university	PROPN
bracis-19080	9	5	of	of	ADP
bracis-19080	9	6	notre	notre	PROPN
bracis-19080	9	7	dame	dame	PROPN
bracis-19080	9	8	hesburgh	hesburgh	PROPN
bracis-19080	9	9	library	library	PROPN
bracis-19080	9	10	.	.	PUNCT
bracis-19080	10	1	download	download	PROPN
bracis-19080	10	2	conference	conference	NOUN
bracis-19080	10	3	paper	paper	NOUN
bracis-19080	10	4	pdf	pdf	NOUN
bracis-19080	10	5	similar	similar	ADJ
bracis-19080	10	6	content	content	NOUN
bracis-19080	10	7	being	be	AUX
bracis-19080	10	8	viewed	view	VERB
bracis-19080	10	9	by	by	ADP
bracis-19080	10	10	others	other	NOUN
bracis-19080	10	11	sentiment	sentiment	NOUN
bracis-19080	10	12	analysis	analysis	NOUN
bracis-19080	10	13	in	in	ADP
bracis-19080	10	14	portuguese	portuguese	ADJ
bracis-19080	10	15	tweets	tweet	NOUN
bracis-19080	10	16	:	:	PUNCT
bracis-19080	10	17	an	an	DET
bracis-19080	10	18	evaluation	evaluation	NOUN
bracis-19080	10	19	of	of	ADP
bracis-19080	10	20	diverse	diverse	ADJ
bracis-19080	10	21	word	word	NOUN
bracis-19080	10	22	representation	representation	NOUN
bracis-19080	10	23	models	model	NOUN
bracis-19080	10	24	article	article	NOUN
bracis-19080	10	25	28	28	NUM
bracis-19080	10	26	june	june	PROPN
bracis-19080	10	27	2023	2023	NUM
bracis-19080	10	28	cosbert	cosbert	PROPN
bracis-19080	10	29	:	:	PUNCT
bracis-19080	10	30	a	a	DET
bracis-19080	10	31	cosine	cosine	NOUN
bracis-19080	10	32	-	-	PUNCT
bracis-19080	10	33	based	base	VERB
bracis-19080	10	34	siamese	siamese	ADJ
bracis-19080	10	35	bert	bert	NOUN
bracis-19080	10	36	-	-	PUNCT
bracis-19080	10	37	networks	network	NOUN
bracis-19080	10	38	using	use	VERB
bracis-19080	10	39	for	for	ADP
bracis-19080	10	40	 	 	SPACE
bracis-19080	10	41	semantic	semantic	ADJ
bracis-19080	10	42	textual	textual	ADJ
bracis-19080	10	43	similarity	similarity	NOUN
bracis-19080	10	44	chapter	chapter	NOUN
bracis-19080	10	45	©	©	PROPN
bracis-19080	10	46	2023	2023	NUM
bracis-19080	10	47	sense	sense	NOUN
bracis-19080	10	48	representations	representation	NOUN
bracis-19080	10	49	for	for	ADP
bracis-19080	10	50	portuguese	portuguese	ADJ
bracis-19080	10	51	:	:	PUNCT
bracis-19080	10	52	experiments	experiment	NOUN
bracis-19080	10	53	with	with	ADP
bracis-19080	10	54	sense	sense	NOUN
bracis-19080	10	55	embeddings	embedding	NOUN
bracis-19080	10	56	and	and	CCONJ
bracis-19080	10	57	deep	deep	ADJ
bracis-19080	10	58	neural	neural	ADJ
bracis-19080	10	59	language	language	NOUN
bracis-19080	10	60	models	model	NOUN
bracis-19080	10	61	article	article	NOUN
bracis-19080	10	62	28	28	NUM
bracis-19080	10	63	february	february	PROPN
bracis-19080	10	64	2021	2021	NUM
bracis-19080	10	65	explore	explore	VERB
bracis-19080	10	66	related	relate	VERB
bracis-19080	10	67	subjects	subject	NOUN
bracis-19080	10	68	discover	discover	VERB
bracis-19080	10	69	the	the	DET
bracis-19080	10	70	latest	late	ADJ
bracis-19080	10	71	articles	article	NOUN
bracis-19080	10	72	,	,	PUNCT
bracis-19080	10	73	books	book	NOUN
bracis-19080	10	74	and	and	CCONJ
bracis-19080	10	75	news	news	NOUN
bracis-19080	10	76	in	in	ADP
bracis-19080	10	77	related	related	ADJ
bracis-19080	10	78	subjects	subject	NOUN
bracis-19080	10	79	,	,	PUNCT
bracis-19080	10	80	suggested	suggest	VERB
bracis-19080	10	81	using	use	VERB
bracis-19080	10	82	machine	machine	NOUN
bracis-19080	10	83	learning	learning	NOUN
bracis-19080	10	84	.	.	PUNCT
bracis-19080	11	1	computational	computational	ADJ
bracis-19080	11	2	linguistics	linguistics	PROPN
bracis-19080	11	3	corpus	corpus	PROPN
bracis-19080	11	4	linguistics	linguistics	PROPN
bracis-19080	11	5	machine	machine	NOUN
bracis-19080	11	6	translation	translation	NOUN
bracis-19080	11	7	machine	machine	NOUN
bracis-19080	11	8	learning	learn	VERB
bracis-19080	11	9	portuguese	portuguese	ADJ
bracis-19080	11	10	semantic	semantic	ADJ
bracis-19080	11	11	memory	memory	NOUN
bracis-19080	11	12	1	1	NUM
bracis-19080	11	13	introduction	introduction	NOUN
bracis-19080	11	14	semantic	semantic	ADJ
bracis-19080	11	15	textual	textual	ADJ
bracis-19080	11	16	similarity	similarity	NOUN
bracis-19080	11	17	(	(	PUNCT
bracis-19080	11	18	sts	st	NOUN
bracis-19080	11	19	)	)	PUNCT
bracis-19080	11	20	is	be	AUX
bracis-19080	11	21	the	the	DET
bracis-19080	11	22	task	task	NOUN
bracis-19080	11	23	of	of	ADP
bracis-19080	11	24	measuring	measure	VERB
bracis-19080	11	25	how	how	SCONJ
bracis-19080	11	26	semantically	semantically	ADV
bracis-19080	11	27	similar	similar	ADJ
bracis-19080	11	28	a	a	DET
bracis-19080	11	29	pair	pair	NOUN
bracis-19080	11	30	of	of	ADP
bracis-19080	11	31	sentences	sentence	NOUN
bracis-19080	11	32	is	be	AUX
bracis-19080	11	33	 	 	SPACE
bracis-19080	11	34	[	[	PUNCT
bracis-19080	11	35	11	11	NUM
bracis-19080	11	36	]	]	PUNCT
bracis-19080	11	37	.	.	PUNCT
bracis-19080	12	1	the	the	DET
bracis-19080	12	2	importance	importance	NOUN
bracis-19080	12	3	of	of	ADP
bracis-19080	12	4	this	this	DET
bracis-19080	12	5	task	task	NOUN
bracis-19080	12	6	to	to	ADP
bracis-19080	12	7	the	the	DET
bracis-19080	12	8	natural	natural	ADJ
bracis-19080	12	9	language	language	NOUN
bracis-19080	12	10	processing	processing	NOUN
bracis-19080	12	11	(	(	PUNCT
bracis-19080	12	12	nlp	nlp	NOUN
bracis-19080	12	13	)	)	PUNCT
bracis-19080	12	14	field	field	NOUN
bracis-19080	12	15	is	be	AUX
bracis-19080	12	16	endorsed	endorse	VERB
bracis-19080	12	17	by	by	ADP
bracis-19080	12	18	the	the	DET
bracis-19080	12	19	creation	creation	NOUN
bracis-19080	12	20	of	of	ADP
bracis-19080	12	21	sts	st	NOUN
bracis-19080	12	22	shared	share	VERB
bracis-19080	12	23	tasks	task	NOUN
bracis-19080	12	24	,	,	PUNCT
bracis-19080	12	25	such	such	ADJ
bracis-19080	12	26	as	as	ADP
bracis-19080	12	27	the	the	DET
bracis-19080	12	28	ones	one	NOUN
bracis-19080	12	29	proposed	propose	VERB
bracis-19080	12	30	by	by	ADP
bracis-19080	12	31	the	the	DET
bracis-19080	12	32	international	international	ADJ
bracis-19080	12	33	workshop	workshop	NOUN
bracis-19080	12	34	on	on	ADP
bracis-19080	12	35	semantic	semantic	ADJ
bracis-19080	12	36	evaluation	evaluation	NOUN
bracis-19080	12	37	 	 	SPACE
bracis-19080	12	38	(	(	PUNCT
bracis-19080	12	39	semeval	semeval	NOUN
bracis-19080	12	40	 	 	SPACE
bracis-19080	13	1	[	[	X
bracis-19080	13	2	5	5	NUM
bracis-19080	13	3	,	,	PUNCT
bracis-19080	13	4	12	12	NUM
bracis-19080	13	5	]	]	PUNCT
bracis-19080	13	6	)	)	PUNCT
bracis-19080	13	7	.	.	PUNCT
bracis-19080	14	1	shared	share	VERB
bracis-19080	14	2	tasks	task	NOUN
bracis-19080	14	3	like	like	ADP
bracis-19080	14	4	semeval	semeval	NOUN
bracis-19080	14	5	led	lead	VERB
bracis-19080	14	6	to	to	ADP
bracis-19080	14	7	great	great	ADJ
bracis-19080	14	8	advancements	advancement	NOUN
bracis-19080	14	9	in	in	ADP
bracis-19080	14	10	sts	st	NOUN
bracis-19080	14	11	,	,	PUNCT
bracis-19080	14	12	but	but	CCONJ
bracis-19080	14	13	there	there	PRON
bracis-19080	14	14	is	be	VERB
bracis-19080	14	15	still	still	ADV
bracis-19080	14	16	a	a	DET
bracis-19080	14	17	reduced	reduce	VERB
bracis-19080	14	18	number	number	NOUN
bracis-19080	14	19	of	of	ADP
bracis-19080	14	20	studies	study	NOUN
bracis-19080	14	21	in	in	ADP
bracis-19080	14	22	the	the	DET
bracis-19080	14	23	context	context	NOUN
bracis-19080	14	24	of	of	ADP
bracis-19080	14	25	the	the	DET
bracis-19080	14	26	portuguese	portuguese	ADJ
bracis-19080	14	27	language	language	NOUN
bracis-19080	14	28	.	.	PUNCT
bracis-19080	15	1	to	to	PART
bracis-19080	15	2	foster	foster	VERB
bracis-19080	15	3	this	this	DET
bracis-19080	15	4	research	research	NOUN
bracis-19080	15	5	,	,	PUNCT
bracis-19080	15	6	the	the	DET
bracis-19080	15	7	assin	assin	PROPN
bracis-19080	15	8	 	 	SPACE
bracis-19080	16	1	[	[	X
bracis-19080	16	2	9	9	NUM
bracis-19080	16	3	]	]	PUNCT
bracis-19080	16	4	and	and	CCONJ
bracis-19080	16	5	assin	assin	PROPN
bracis-19080	16	6	 	 	SPACE
bracis-19080	16	7	2	2	NUM
bracis-19080	16	8	 	 	SPACE
bracis-19080	17	1	[	[	X
bracis-19080	17	2	15	15	NUM
bracis-19080	17	3	]	]	X
bracis-19080	17	4	workshops	workshop	NOUN
bracis-19080	17	5	hosted	host	VERB
bracis-19080	17	6	shared	share	VERB
bracis-19080	17	7	tasks	task	NOUN
bracis-19080	17	8	for	for	ADP
bracis-19080	17	9	sts	st	NOUN
bracis-19080	17	10	and	and	CCONJ
bracis-19080	17	11	natural	natural	ADJ
bracis-19080	17	12	language	language	NOUN
bracis-19080	17	13	inference	inference	NOUN
bracis-19080	17	14	 	 	SPACE
bracis-19080	17	15	(	(	PUNCT
bracis-19080	17	16	nli	nli	PROPN
bracis-19080	17	17	)	)	PUNCT
bracis-19080	17	18	in	in	ADP
bracis-19080	17	19	portuguese	portuguese	PROPN
bracis-19080	17	20	.	.	PUNCT
bracis-19080	18	1	the	the	DET
bracis-19080	18	2	models	model	NOUN
bracis-19080	18	3	developed	develop	VERB
bracis-19080	18	4	during	during	ADP
bracis-19080	18	5	assin	assin	PROPN
bracis-19080	18	6	 	 	SPACE
bracis-19080	18	7	2	2	NUM
bracis-19080	18	8	used	use	VERB
bracis-19080	18	9	more	more	ADV
bracis-19080	18	10	recent	recent	ADJ
bracis-19080	18	11	nlp	nlp	ADJ
bracis-19080	18	12	approaches	approach	NOUN
bracis-19080	18	13	,	,	PUNCT
bracis-19080	18	14	including	include	VERB
bracis-19080	18	15	contextual	contextual	ADJ
bracis-19080	18	16	embeddings	embedding	NOUN
bracis-19080	18	17	like	like	ADP
bracis-19080	18	18	bert	bert	NOUN
bracis-19080	18	19	 	 	SPACE
bracis-19080	19	1	[	[	X
bracis-19080	19	2	7	7	NUM
bracis-19080	19	3	]	]	PUNCT
bracis-19080	19	4	.	.	PUNCT
bracis-19080	20	1	recent	recent	ADJ
bracis-19080	20	2	works	work	NOUN
bracis-19080	20	3	addressing	address	VERB
bracis-19080	20	4	these	these	DET
bracis-19080	20	5	datasets	dataset	NOUN
bracis-19080	20	6	have	have	AUX
bracis-19080	20	7	since	since	ADV
bracis-19080	20	8	been	be	AUX
bracis-19080	20	9	continuously	continuously	ADV
bracis-19080	20	10	proposed	propose	VERB
bracis-19080	20	11	,	,	PUNCT
bracis-19080	20	12	with	with	ADP
bracis-19080	20	13	the	the	DET
bracis-19080	20	14	state	state	NOUN
bracis-19080	20	15	-	-	PUNCT
bracis-19080	20	16	of	of	ADP
bracis-19080	20	17	-	-	PUNCT
bracis-19080	20	18	the	the	DET
bracis-19080	20	19	-	-	PUNCT
bracis-19080	20	20	art	art	NOUN
bracis-19080	20	21	being	be	AUX
bracis-19080	20	22	a	a	DET
bracis-19080	20	23	bert	bert	NOUN
bracis-19080	20	24	model	model	NOUN
bracis-19080	20	25	pre	pre	VERB
bracis-19080	20	26	-	-	VERB
bracis-19080	20	27	trained	train	VERB
bracis-19080	20	28	in	in	ADP
bracis-19080	20	29	portuguese	portuguese	ADJ
bracis-19080	20	30	fine	fine	ADV
bracis-19080	20	31	-	-	PUNCT
bracis-19080	20	32	tuned	tune	VERB
bracis-19080	20	33	for	for	ADP
bracis-19080	20	34	sts	st	NOUN
bracis-19080	20	35	 	 	SPACE
bracis-19080	20	36	[	[	X
bracis-19080	20	37	8	8	NUM
bracis-19080	20	38	]	]	PUNCT
bracis-19080	20	39	.	.	PUNCT
bracis-19080	21	1	though	though	SCONJ
bracis-19080	21	2	competitive	competitive	ADJ
bracis-19080	21	3	,	,	PUNCT
bracis-19080	21	4	training	train	VERB
bracis-19080	21	5	bert	bert	NOUN
bracis-19080	21	6	models	model	NOUN
bracis-19080	21	7	on	on	ADP
bracis-19080	21	8	sts	sts	ADJ
bracis-19080	21	9	demands	demand	NOUN
bracis-19080	21	10	that	that	SCONJ
bracis-19080	21	11	all	all	DET
bracis-19080	21	12	pairs	pair	NOUN
bracis-19080	21	13	of	of	ADP
bracis-19080	21	14	sentences	sentence	NOUN
bracis-19080	21	15	be	be	AUX
bracis-19080	21	16	used	use	VERB
bracis-19080	21	17	as	as	ADP
bracis-19080	21	18	input	input	NOUN
bracis-19080	21	19	to	to	ADP
bracis-19080	21	20	the	the	DET
bracis-19080	21	21	network	network	NOUN
bracis-19080	21	22	.	.	PUNCT
bracis-19080	22	1	this	this	DET
bracis-19080	22	2	approach	approach	NOUN
bracis-19080	22	3	can	can	AUX
bracis-19080	22	4	cause	cause	VERB
bracis-19080	22	5	a	a	DET
bracis-19080	22	6	massive	massive	ADJ
bracis-19080	22	7	computational	computational	ADJ
bracis-19080	22	8	overhead	overhead	NOUN
bracis-19080	22	9	,	,	PUNCT
bracis-19080	22	10	even	even	ADV
bracis-19080	22	11	for	for	ADP
bracis-19080	22	12	a	a	DET
bracis-19080	22	13	moderate	moderate	ADJ
bracis-19080	22	14	-	-	PUNCT
bracis-19080	22	15	size	size	NOUN
bracis-19080	22	16	corpus	corpus	NOUN
bracis-19080	22	17	.	.	PUNCT
bracis-19080	23	1	to	to	PART
bracis-19080	23	2	address	address	VERB
bracis-19080	23	3	this	this	DET
bracis-19080	23	4	problem	problem	NOUN
bracis-19080	23	5	,	,	PUNCT
bracis-19080	23	6	reimers	reimer	NOUN
bracis-19080	23	7	and	and	CCONJ
bracis-19080	23	8	gurevych	gurevych	NOUN
bracis-19080	23	9	 	 	SPACE
bracis-19080	24	1	[	[	X
bracis-19080	24	2	17	17	NUM
bracis-19080	24	3	]	]	PUNCT
bracis-19080	24	4	proposed	propose	VERB
bracis-19080	24	5	sentence	sentence	NOUN
bracis-19080	24	6	-	-	PUNCT
bracis-19080	24	7	bert	bert	PROPN
bracis-19080	24	8	(	(	PUNCT
bracis-19080	24	9	sbert	sbert	NOUN
bracis-19080	24	10	)	)	PUNCT
bracis-19080	24	11	,	,	PUNCT
bracis-19080	24	12	a	a	DET
bracis-19080	24	13	siamese	siamese	ADJ
bracis-19080	24	14	architecture	architecture	NOUN
bracis-19080	24	15	with	with	ADP
bracis-19080	24	16	shared	share	VERB
bracis-19080	24	17	weights	weight	NOUN
bracis-19080	24	18	that	that	PRON
bracis-19080	24	19	can	can	AUX
bracis-19080	24	20	reduce	reduce	VERB
bracis-19080	24	21	the	the	DET
bracis-19080	24	22	cost	cost	NOUN
bracis-19080	24	23	of	of	ADP
bracis-19080	24	24	bert	bert	PROPN
bracis-19080	24	25	models	model	NOUN
bracis-19080	24	26	.	.	PUNCT
bracis-19080	25	1	since	since	SCONJ
bracis-19080	25	2	its	its	PRON
bracis-19080	25	3	proposition	proposition	NOUN
bracis-19080	25	4	,	,	PUNCT
bracis-19080	25	5	sbert	sbert	NOUN
bracis-19080	25	6	has	have	AUX
bracis-19080	25	7	been	be	AUX
bracis-19080	25	8	used	use	VERB
bracis-19080	25	9	successfully	successfully	ADV
bracis-19080	25	10	in	in	ADP
bracis-19080	25	11	tasks	task	NOUN
bracis-19080	25	12	where	where	SCONJ
bracis-19080	25	13	bert	bert	PROPN
bracis-19080	25	14	had	have	AUX
bracis-19080	25	15	obtained	obtain	VERB
bracis-19080	25	16	good	good	ADJ
bracis-19080	25	17	results	result	NOUN
bracis-19080	25	18	 	 	SPACE
bracis-19080	25	19	[	[	X
bracis-19080	25	20	17	17	NUM
bracis-19080	25	21	,	,	PUNCT
bracis-19080	25	22	18	18	NUM
bracis-19080	25	23	]	]	PUNCT
bracis-19080	25	24	.	.	PUNCT
bracis-19080	26	1	however	however	ADV
bracis-19080	26	2	,	,	PUNCT
bracis-19080	26	3	to	to	ADP
bracis-19080	26	4	the	the	DET
bracis-19080	26	5	best	good	ADJ
bracis-19080	26	6	of	of	ADP
bracis-19080	26	7	our	our	PRON
bracis-19080	26	8	knowledge	knowledge	NOUN
bracis-19080	26	9	,	,	PUNCT
bracis-19080	26	10	these	these	DET
bracis-19080	26	11	applications	application	NOUN
bracis-19080	26	12	do	do	AUX
bracis-19080	26	13	not	not	PART
bracis-19080	26	14	yet	yet	ADV
bracis-19080	26	15	include	include	VERB
bracis-19080	26	16	sts	st	NOUN
bracis-19080	26	17	in	in	ADP
bracis-19080	26	18	portuguese	portuguese	PROPN
bracis-19080	26	19	.	.	PUNCT
bracis-19080	27	1	the	the	DET
bracis-19080	27	2	goal	goal	NOUN
bracis-19080	27	3	of	of	ADP
bracis-19080	27	4	this	this	DET
bracis-19080	27	5	work	work	NOUN
bracis-19080	27	6	is	be	AUX
bracis-19080	27	7	to	to	PART
bracis-19080	27	8	evaluate	evaluate	VERB
bracis-19080	27	9	contextual	contextual	ADJ
bracis-19080	27	10	embeddings	embedding	NOUN
bracis-19080	27	11	generated	generate	VERB
bracis-19080	27	12	by	by	ADP
bracis-19080	27	13	sbert	sbert	NOUN
bracis-19080	27	14	models	model	NOUN
bracis-19080	27	15	for	for	ADP
bracis-19080	27	16	sts	st	NOUN
bracis-19080	27	17	in	in	ADP
bracis-19080	27	18	portuguese	portuguese	NOUN
bracis-19080	27	19	,	,	PUNCT
bracis-19080	27	20	which	which	PRON
bracis-19080	27	21	we	we	PRON
bracis-19080	27	22	investigate	investigate	VERB
bracis-19080	27	23	in	in	ADP
bracis-19080	27	24	two	two	NUM
bracis-19080	27	25	stages	stage	NOUN
bracis-19080	27	26	.	.	PUNCT
bracis-19080	28	1	first	first	ADV
bracis-19080	28	2	,	,	PUNCT
bracis-19080	28	3	we	we	PRON
bracis-19080	28	4	compare	compare	VERB
bracis-19080	28	5	the	the	DET
bracis-19080	28	6	performance	performance	NOUN
bracis-19080	28	7	of	of	ADP
bracis-19080	28	8	pre	pre	ADJ
bracis-19080	28	9	-	-	ADJ
bracis-19080	28	10	trained	train	VERB
bracis-19080	28	11	sbert	sbert	NOUN
bracis-19080	28	12	models	model	NOUN
bracis-19080	28	13	with	with	ADP
bracis-19080	28	14	the	the	DET
bracis-19080	28	15	state	state	NOUN
bracis-19080	28	16	-	-	PUNCT
bracis-19080	28	17	of	of	ADP
bracis-19080	28	18	-	-	PUNCT
bracis-19080	28	19	the	the	DET
bracis-19080	28	20	-	-	PUNCT
bracis-19080	28	21	art	art	NOUN
bracis-19080	28	22	bert	bert	NOUN
bracis-19080	28	23	models	model	NOUN
bracis-19080	28	24	for	for	ADP
bracis-19080	28	25	the	the	DET
bracis-19080	28	26	assin	assin	PROPN
bracis-19080	28	27	datasets	dataset	NOUN
bracis-19080	28	28	 	 	SPACE
bracis-19080	29	1	[	[	X
bracis-19080	29	2	8	8	NUM
bracis-19080	29	3	]	]	PUNCT
bracis-19080	29	4	.	.	PUNCT
bracis-19080	30	1	in	in	ADP
bracis-19080	30	2	addition	addition	NOUN
bracis-19080	30	3	,	,	PUNCT
bracis-19080	30	4	we	we	PRON
bracis-19080	30	5	include	include	VERB
bracis-19080	30	6	other	other	ADJ
bracis-19080	30	7	baseline	baseline	NOUN
bracis-19080	30	8	models	model	NOUN
bracis-19080	30	9	,	,	PUNCT
bracis-19080	30	10	such	such	ADJ
bracis-19080	30	11	as	as	ADP
bracis-19080	30	12	the	the	DET
bracis-19080	30	13	best	well	ADV
bracis-19080	30	14	-	-	PUNCT
bracis-19080	30	15	performing	perform	VERB
bracis-19080	30	16	works	work	NOUN
bracis-19080	30	17	assessed	assess	VERB
bracis-19080	30	18	in	in	ADP
bracis-19080	30	19	the	the	DET
bracis-19080	30	20	workshops	workshop	NOUN
bracis-19080	30	21	and	and	CCONJ
bracis-19080	30	22	other	other	ADJ
bracis-19080	30	23	multilingual	multilingual	ADJ
bracis-19080	30	24	contextual	contextual	ADJ
bracis-19080	30	25	embeddings	embedding	NOUN
bracis-19080	30	26	.	.	PUNCT
bracis-19080	31	1	later	later	ADV
bracis-19080	31	2	,	,	PUNCT
bracis-19080	31	3	we	we	PRON
bracis-19080	31	4	evaluate	evaluate	VERB
bracis-19080	31	5	the	the	DET
bracis-19080	31	6	benefits	benefit	NOUN
bracis-19080	31	7	of	of	ADP
bracis-19080	31	8	fine	fine	ADV
bracis-19080	31	9	-	-	PUNCT
bracis-19080	31	10	tuning	tune	VERB
bracis-19080	31	11	sbert	sbert	NOUN
bracis-19080	31	12	models	model	NOUN
bracis-19080	31	13	for	for	ADP
bracis-19080	31	14	sts	st	NOUN
bracis-19080	31	15	in	in	ADP
bracis-19080	31	16	portuguese	portuguese	NOUN
bracis-19080	31	17	,	,	PUNCT
bracis-19080	31	18	also	also	ADV
bracis-19080	31	19	comparing	compare	VERB
bracis-19080	31	20	them	they	PRON
bracis-19080	31	21	with	with	ADP
bracis-19080	31	22	all	all	DET
bracis-19080	31	23	baseline	baseline	ADJ
bracis-19080	31	24	and	and	CCONJ
bracis-19080	31	25	state	state	NOUN
bracis-19080	31	26	-	-	PUNCT
bracis-19080	31	27	of	of	ADP
bracis-19080	31	28	-	-	PUNCT
bracis-19080	31	29	the	the	DET
bracis-19080	31	30	-	-	PUNCT
bracis-19080	31	31	art	art	NOUN
bracis-19080	31	32	models	model	NOUN
bracis-19080	31	33	.	.	PUNCT
bracis-19080	32	1	results	result	NOUN
bracis-19080	32	2	from	from	ADP
bracis-19080	32	3	the	the	DET
bracis-19080	32	4	first	first	ADJ
bracis-19080	32	5	part	part	NOUN
bracis-19080	32	6	of	of	ADP
bracis-19080	32	7	our	our	PRON
bracis-19080	32	8	investigation	investigation	NOUN
bracis-19080	32	9	showed	show	VERB
bracis-19080	32	10	that	that	SCONJ
bracis-19080	32	11	multilingual	multilingual	ADJ
bracis-19080	32	12	sbert	sbert	NOUN
bracis-19080	32	13	models	model	NOUN
bracis-19080	32	14	are	be	AUX
bracis-19080	32	15	competitive	competitive	ADJ
bracis-19080	32	16	,	,	PUNCT
bracis-19080	32	17	outperforming	outperform	VERB
bracis-19080	32	18	the	the	DET
bracis-19080	32	19	best	good	ADJ
bracis-19080	32	20	results	result	NOUN
bracis-19080	32	21	of	of	ADP
bracis-19080	32	22	the	the	DET
bracis-19080	32	23	assin	assin	PROPN
bracis-19080	32	24	shared	share	VERB
bracis-19080	32	25	task	task	NOUN
bracis-19080	32	26	.	.	PUNCT
bracis-19080	33	1	even	even	ADV
bracis-19080	33	2	if	if	SCONJ
bracis-19080	33	3	at	at	ADP
bracis-19080	33	4	this	this	DET
bracis-19080	33	5	stage	stage	NOUN
bracis-19080	33	6	fine	fine	ADV
bracis-19080	33	7	-	-	PUNCT
bracis-19080	33	8	tuning	tuning	NOUN
bracis-19080	33	9	was	be	AUX
bracis-19080	33	10	not	not	PART
bracis-19080	33	11	considered	consider	VERB
bracis-19080	33	12	,	,	PUNCT
bracis-19080	33	13	sbert	sbert	NOUN
bracis-19080	33	14	results	result	NOUN
bracis-19080	33	15	were	be	AUX
bracis-19080	33	16	second	second	ADJ
bracis-19080	33	17	only	only	ADV
bracis-19080	33	18	to	to	ADP
bracis-19080	33	19	the	the	DET
bracis-19080	33	20	results	result	NOUN
bracis-19080	33	21	achieved	achieve	VERB
bracis-19080	33	22	by	by	ADP
bracis-19080	33	23	the	the	DET
bracis-19080	33	24	state	state	NOUN
bracis-19080	33	25	-	-	PUNCT
bracis-19080	33	26	of	of	ADP
bracis-19080	33	27	-	-	PUNCT
bracis-19080	33	28	the	the	DET
bracis-19080	33	29	-	-	PUNCT
bracis-19080	33	30	art	art	NOUN
bracis-19080	33	31	bert	bert	PROPN
bracis-19080	33	32	model	model	NOUN
bracis-19080	33	33	 	 	SPACE
bracis-19080	34	1	[	[	X
bracis-19080	34	2	8	8	NUM
bracis-19080	34	3	]	]	PUNCT
bracis-19080	34	4	.	.	PUNCT
bracis-19080	35	1	in	in	ADP
bracis-19080	35	2	the	the	DET
bracis-19080	35	3	second	second	ADJ
bracis-19080	35	4	part	part	NOUN
bracis-19080	35	5	of	of	ADP
bracis-19080	35	6	our	our	PRON
bracis-19080	35	7	work	work	NOUN
bracis-19080	35	8	,	,	PUNCT
bracis-19080	35	9	results	result	NOUN
bracis-19080	35	10	were	be	AUX
bracis-19080	35	11	improved	improve	VERB
bracis-19080	35	12	with	with	ADP
bracis-19080	35	13	the	the	DET
bracis-19080	35	14	fine	fine	ADV
bracis-19080	35	15	-	-	PUNCT
bracis-19080	35	16	tuning	tuning	NOUN
bracis-19080	35	17	,	,	PUNCT
bracis-19080	35	18	numerically	numerically	ADV
bracis-19080	35	19	surpassing	surpass	VERB
bracis-19080	35	20	the	the	DET
bracis-19080	35	21	performance	performance	NOUN
bracis-19080	35	22	of	of	ADP
bracis-19080	35	23	the	the	DET
bracis-19080	35	24	state	state	NOUN
bracis-19080	35	25	-	-	PUNCT
bracis-19080	35	26	of	of	ADP
bracis-19080	35	27	-	-	PUNCT
bracis-19080	35	28	the	the	DET
bracis-19080	35	29	-	-	PUNCT
bracis-19080	35	30	art	art	NOUN
bracis-19080	35	31	model	model	NOUN
bracis-19080	35	32	for	for	ADP
bracis-19080	35	33	the	the	DET
bracis-19080	35	34	assin	assin	PROPN
bracis-19080	35	35	 	 	SPACE
bracis-19080	35	36	2	2	NUM
bracis-19080	35	37	dataset	dataset	NOUN
bracis-19080	35	38	,	,	PUNCT
bracis-19080	35	39	at	at	ADP
bracis-19080	35	40	a	a	DET
bracis-19080	35	41	much	much	ADV
bracis-19080	35	42	lower	low	ADJ
bracis-19080	35	43	computational	computational	ADJ
bracis-19080	35	44	cost	cost	NOUN
bracis-19080	35	45	.	.	PUNCT
bracis-19080	36	1	for	for	ADP
bracis-19080	36	2	the	the	DET
bracis-19080	36	3	assin	assin	PROPN
bracis-19080	36	4	datasets	dataset	NOUN
bracis-19080	36	5	,	,	PUNCT
bracis-19080	36	6	even	even	ADV
bracis-19080	36	7	if	if	SCONJ
bracis-19080	36	8	the	the	DET
bracis-19080	36	9	state	state	NOUN
bracis-19080	36	10	-	-	PUNCT
bracis-19080	36	11	of	of	ADP
bracis-19080	36	12	-	-	PUNCT
bracis-19080	36	13	the	the	DET
bracis-19080	36	14	-	-	PUNCT
bracis-19080	36	15	art	art	NOUN
bracis-19080	36	16	results	result	NOUN
bracis-19080	36	17	were	be	AUX
bracis-19080	36	18	not	not	PART
bracis-19080	36	19	matched	match	VERB
bracis-19080	36	20	,	,	PUNCT
bracis-19080	36	21	the	the	DET
bracis-19080	36	22	contextual	contextual	ADJ
bracis-19080	36	23	embeddings	embedding	NOUN
bracis-19080	36	24	generated	generate	VERB
bracis-19080	36	25	by	by	ADP
bracis-19080	36	26	multilingual	multilingual	ADJ
bracis-19080	36	27	sbert	sbert	NOUN
bracis-19080	36	28	models	model	NOUN
bracis-19080	36	29	with	with	ADP
bracis-19080	36	30	fine	fine	ADV
bracis-19080	36	31	-	-	PUNCT
bracis-19080	36	32	tuning	tuning	NOUN
bracis-19080	36	33	remained	remain	VERB
bracis-19080	36	34	competitive	competitive	ADJ
bracis-19080	36	35	and	and	CCONJ
bracis-19080	36	36	,	,	PUNCT
bracis-19080	36	37	as	as	SCONJ
bracis-19080	36	38	mentioned	mention	VERB
bracis-19080	36	39	before	before	ADV
bracis-19080	36	40	,	,	PUNCT
bracis-19080	36	41	required	require	VERB
bracis-19080	36	42	a	a	DET
bracis-19080	36	43	reduced	reduced	ADJ
bracis-19080	36	44	computational	computational	ADJ
bracis-19080	36	45	cost	cost	NOUN
bracis-19080	36	46	.	.	PUNCT
bracis-19080	37	1	finally	finally	ADV
bracis-19080	37	2	,	,	PUNCT
bracis-19080	37	3	we	we	PRON
bracis-19080	37	4	discuss	discuss	VERB
bracis-19080	37	5	the	the	DET
bracis-19080	37	6	impacts	impact	NOUN
bracis-19080	37	7	of	of	ADP
bracis-19080	37	8	fine	fine	ADV
bracis-19080	37	9	-	-	PUNCT
bracis-19080	37	10	tuning	tuning	NOUN
bracis-19080	37	11	and	and	CCONJ
bracis-19080	37	12	language	language	NOUN
bracis-19080	37	13	variants	variant	NOUN
bracis-19080	37	14	with	with	ADP
bracis-19080	37	15	a	a	DET
bracis-19080	37	16	qualitative	qualitative	ADJ
bracis-19080	37	17	assessment	assessment	NOUN
bracis-19080	37	18	that	that	PRON
bracis-19080	37	19	demonstrates	demonstrate	VERB
bracis-19080	37	20	that	that	SCONJ
bracis-19080	37	21	results	result	NOUN
bracis-19080	37	22	can	can	AUX
bracis-19080	37	23	be	be	AUX
bracis-19080	37	24	further	far	ADV
bracis-19080	37	25	improved	improve	VERB
bracis-19080	37	26	in	in	ADP
bracis-19080	37	27	the	the	DET
bracis-19080	37	28	future	future	NOUN
bracis-19080	37	29	.	.	PUNCT
bracis-19080	38	1	the	the	DET
bracis-19080	38	2	remainder	remainder	NOUN
bracis-19080	38	3	of	of	ADP
bracis-19080	38	4	this	this	DET
bracis-19080	38	5	work	work	NOUN
bracis-19080	38	6	is	be	AUX
bracis-19080	38	7	structured	structure	VERB
bracis-19080	38	8	as	as	SCONJ
bracis-19080	38	9	follows	follow	VERB
bracis-19080	38	10	.	.	PUNCT
bracis-19080	39	1	section	section	NOUN
bracis-19080	39	2	 	 	SPACE
bracis-19080	39	3	2	2	NUM
bracis-19080	39	4	briefly	briefly	NOUN
bracis-19080	39	5	reviews	review	VERB
bracis-19080	39	6	preliminary	preliminary	ADJ
bracis-19080	39	7	concepts	concept	NOUN
bracis-19080	39	8	,	,	PUNCT
bracis-19080	39	9	namely	namely	ADV
bracis-19080	39	10	contextual	contextual	ADJ
bracis-19080	39	11	embeddings	embedding	NOUN
bracis-19080	39	12	and	and	CCONJ
bracis-19080	39	13	the	the	DET
bracis-19080	39	14	most	most	ADV
bracis-19080	39	15	relevant	relevant	ADJ
bracis-19080	39	16	architectures	architecture	NOUN
bracis-19080	39	17	used	use	VERB
bracis-19080	39	18	to	to	PART
bracis-19080	39	19	produce	produce	VERB
bracis-19080	39	20	them	they	PRON
bracis-19080	39	21	,	,	PUNCT
bracis-19080	39	22	and	and	CCONJ
bracis-19080	39	23	deep	deep	ADJ
bracis-19080	39	24	learning	learning	NOUN
bracis-19080	39	25	training	training	NOUN
bracis-19080	39	26	approaches	approach	NOUN
bracis-19080	39	27	such	such	ADJ
bracis-19080	39	28	as	as	ADP
bracis-19080	39	29	fine	fine	ADV
bracis-19080	39	30	-	-	PUNCT
bracis-19080	39	31	tuning	tuning	NOUN
bracis-19080	39	32	and	and	CCONJ
bracis-19080	39	33	knowledge	knowledge	NOUN
bracis-19080	39	34	distillation	distillation	NOUN
bracis-19080	39	35	.	.	PUNCT
bracis-19080	40	1	next	next	ADJ
bracis-19080	40	2	,	,	PUNCT
bracis-19080	40	3	sect	sect	NOUN
bracis-19080	40	4	.	.	PUNCT
bracis-19080	40	5	 	 	SPACE
bracis-19080	41	1	3	3	NUM
bracis-19080	41	2	defines	define	VERB
bracis-19080	41	3	the	the	DET
bracis-19080	41	4	sts	st	NOUN
bracis-19080	41	5	problem	problem	NOUN
bracis-19080	41	6	,	,	PUNCT
bracis-19080	41	7	details	detail	VERB
bracis-19080	41	8	the	the	DET
bracis-19080	41	9	assin	assin	PROPN
bracis-19080	41	10	and	and	CCONJ
bracis-19080	41	11	assin	assin	PROPN
bracis-19080	41	12	 	 	SPACE
bracis-19080	41	13	2	2	NUM
bracis-19080	41	14	shared	share	VERB
bracis-19080	41	15	tasks	task	NOUN
bracis-19080	41	16	,	,	PUNCT
bracis-19080	41	17	and	and	CCONJ
bracis-19080	41	18	briefly	briefly	ADV
bracis-19080	41	19	discusses	discuss	VERB
bracis-19080	41	20	the	the	DET
bracis-19080	41	21	state	state	NOUN
bracis-19080	41	22	-	-	PUNCT
bracis-19080	41	23	of	of	ADP
bracis-19080	41	24	-	-	PUNCT
bracis-19080	41	25	the	the	DET
bracis-19080	41	26	-	-	PUNCT
bracis-19080	41	27	art	art	NOUN
bracis-19080	41	28	for	for	ADP
bracis-19080	41	29	these	these	DET
bracis-19080	41	30	datasets	dataset	NOUN
bracis-19080	41	31	.	.	PUNCT
bracis-19080	42	1	in	in	ADP
bracis-19080	42	2	sect	sect	NOUN
bracis-19080	42	3	.	.	PUNCT
bracis-19080	42	4	 	 	SPACE
bracis-19080	42	5	4	4	NUM
bracis-19080	42	6	,	,	PUNCT
bracis-19080	42	7	we	we	PRON
bracis-19080	42	8	detail	detail	VERB
bracis-19080	42	9	the	the	DET
bracis-19080	42	10	experimental	experimental	ADJ
bracis-19080	42	11	setup	setup	NOUN
bracis-19080	42	12	adopted	adopt	VERB
bracis-19080	42	13	in	in	ADP
bracis-19080	42	14	this	this	DET
bracis-19080	42	15	investigation	investigation	NOUN
bracis-19080	42	16	,	,	PUNCT
bracis-19080	42	17	and	and	CCONJ
bracis-19080	42	18	discuss	discuss	VERB
bracis-19080	42	19	results	result	NOUN
bracis-19080	42	20	in	in	ADP
bracis-19080	42	21	sects	sect	NOUN
bracis-19080	42	22	.	.	PUNCT
bracis-19080	42	23	 	 	SPACE
bracis-19080	43	1	5	5	NUM
bracis-19080	43	2	and	and	CCONJ
bracis-19080	43	3	 	 	SPACE
bracis-19080	43	4	6	6	NUM
bracis-19080	43	5	.	.	PUNCT
bracis-19080	44	1	we	we	PRON
bracis-19080	44	2	conclude	conclude	VERB
bracis-19080	44	3	and	and	CCONJ
bracis-19080	44	4	discuss	discuss	VERB
bracis-19080	44	5	future	future	ADJ
bracis-19080	44	6	work	work	NOUN
bracis-19080	44	7	in	in	ADP
bracis-19080	44	8	sect	sect	NOUN
bracis-19080	44	9	.	.	PUNCT
bracis-19080	44	10	 	 	SPACE
bracis-19080	45	1	7	7	NUM
bracis-19080	45	2	.	.	SYM
bracis-19080	45	3	2	2	NUM
bracis-19080	45	4	background	background	NOUN
bracis-19080	45	5	in	in	ADP
bracis-19080	45	6	this	this	DET
bracis-19080	45	7	section	section	NOUN
bracis-19080	45	8	,	,	PUNCT
bracis-19080	45	9	we	we	PRON
bracis-19080	45	10	briefly	briefly	ADV
bracis-19080	45	11	review	review	VERB
bracis-19080	45	12	the	the	DET
bracis-19080	45	13	main	main	ADJ
bracis-19080	45	14	preliminary	preliminary	ADJ
bracis-19080	45	15	concepts	concept	NOUN
bracis-19080	45	16	required	require	VERB
bracis-19080	45	17	to	to	PART
bracis-19080	45	18	understand	understand	VERB
bracis-19080	45	19	contextual	contextual	ADJ
bracis-19080	45	20	embeddings	embedding	NOUN
bracis-19080	45	21	,	,	PUNCT
bracis-19080	45	22	an	an	DET
bracis-19080	45	23	important	important	ADJ
bracis-19080	45	24	tool	tool	NOUN
bracis-19080	45	25	for	for	ADP
bracis-19080	45	26	problem	problem	NOUN
bracis-19080	45	27	solving	solve	VERB
bracis-19080	45	28	in	in	ADP
bracis-19080	45	29	nlp	nlp	NOUN
bracis-19080	45	30	.	.	PUNCT
bracis-19080	46	1	initially	initially	ADV
bracis-19080	46	2	,	,	PUNCT
bracis-19080	46	3	we	we	PRON
bracis-19080	46	4	discuss	discuss	VERB
bracis-19080	46	5	transfer	transfer	NOUN
bracis-19080	46	6	learning	learning	NOUN
bracis-19080	46	7	 	 	SPACE
bracis-19080	47	1	[	[	X
bracis-19080	47	2	25	25	NUM
bracis-19080	47	3	]	]	PUNCT
bracis-19080	47	4	,	,	PUNCT
bracis-19080	47	5	the	the	DET
bracis-19080	47	6	training	training	NOUN
bracis-19080	47	7	paradigm	paradigm	NOUN
bracis-19080	47	8	that	that	PRON
bracis-19080	47	9	motivated	motivate	VERB
bracis-19080	47	10	the	the	DET
bracis-19080	47	11	proposal	proposal	NOUN
bracis-19080	47	12	of	of	ADP
bracis-19080	47	13	embeddings	embedding	NOUN
bracis-19080	47	14	in	in	ADP
bracis-19080	47	15	general	general	ADJ
bracis-19080	47	16	.	.	PUNCT
bracis-19080	48	1	next	next	ADV
bracis-19080	48	2	,	,	PUNCT
bracis-19080	48	3	we	we	PRON
bracis-19080	48	4	discuss	discuss	VERB
bracis-19080	48	5	contextual	contextual	ADJ
bracis-19080	48	6	embeddings	embedding	NOUN
bracis-19080	48	7	and	and	CCONJ
bracis-19080	48	8	the	the	DET
bracis-19080	48	9	main	main	ADJ
bracis-19080	48	10	algorithms	algorithm	NOUN
bracis-19080	48	11	adopted	adopt	VERB
bracis-19080	48	12	in	in	ADP
bracis-19080	48	13	this	this	DET
bracis-19080	48	14	work	work	NOUN
bracis-19080	48	15	,	,	PUNCT
bracis-19080	48	16	namely	namely	ADV
bracis-19080	48	17	bert	bert	VERB
bracis-19080	48	18	 	 	SPACE
bracis-19080	49	1	[	[	X
bracis-19080	49	2	7	7	NUM
bracis-19080	49	3	]	]	PUNCT
bracis-19080	49	4	and	and	CCONJ
bracis-19080	49	5	sentence	sentence	NOUN
bracis-19080	49	6	-	-	PUNCT
bracis-19080	49	7	bert	bert	NOUN
bracis-19080	49	8	 	 	SPACE
bracis-19080	50	1	[	[	X
bracis-19080	50	2	17	17	NUM
bracis-19080	50	3	]	]	PUNCT
bracis-19080	50	4	.	.	PUNCT
bracis-19080	51	1	finally	finally	ADV
bracis-19080	51	2	,	,	PUNCT
bracis-19080	51	3	we	we	PRON
bracis-19080	51	4	briefly	briefly	ADV
bracis-19080	51	5	detail	detail	VERB
bracis-19080	51	6	knowledge	knowledge	NOUN
bracis-19080	51	7	distillation	distillation	NOUN
bracis-19080	51	8	 	 	SPACE
bracis-19080	52	1	[	[	X
bracis-19080	52	2	10	10	NUM
bracis-19080	52	3	]	]	PUNCT
bracis-19080	52	4	,	,	PUNCT
bracis-19080	52	5	the	the	DET
bracis-19080	52	6	deep	deep	ADJ
bracis-19080	52	7	learning	learning	NOUN
bracis-19080	52	8	training	training	NOUN
bracis-19080	52	9	paradigm	paradigm	NOUN
bracis-19080	52	10	that	that	PRON
bracis-19080	52	11	enables	enable	VERB
bracis-19080	52	12	the	the	DET
bracis-19080	52	13	multilingual	multilingual	ADJ
bracis-19080	52	14	training	training	NOUN
bracis-19080	52	15	used	use	VERB
bracis-19080	52	16	to	to	PART
bracis-19080	52	17	produce	produce	VERB
bracis-19080	52	18	the	the	DET
bracis-19080	52	19	models	model	NOUN
bracis-19080	52	20	we	we	PRON
bracis-19080	52	21	adopt	adopt	VERB
bracis-19080	52	22	in	in	ADP
bracis-19080	52	23	this	this	DET
bracis-19080	52	24	work	work	NOUN
bracis-19080	52	25	.	.	PUNCT
bracis-19080	53	1	2.1	2.1	NUM
bracis-19080	53	2	transfer	transfer	NOUN
bracis-19080	53	3	learning	learning	NOUN
bracis-19080	53	4	transfer	transfer	NOUN
bracis-19080	53	5	learning	learning	NOUN
bracis-19080	53	6	is	be	AUX
bracis-19080	53	7	a	a	DET
bracis-19080	53	8	training	training	NOUN
bracis-19080	53	9	paradigm	paradigm	NOUN
bracis-19080	53	10	that	that	PRON
bracis-19080	53	11	enables	enable	VERB
bracis-19080	53	12	models	model	NOUN
bracis-19080	53	13	fit	fit	ADJ
bracis-19080	53	14	to	to	ADP
bracis-19080	53	15	a	a	DET
bracis-19080	53	16	general	general	ADJ
bracis-19080	53	17	problem	problem	NOUN
bracis-19080	53	18	to	to	PART
bracis-19080	53	19	be	be	AUX
bracis-19080	53	20	reused	reuse	VERB
bracis-19080	53	21	for	for	ADP
bracis-19080	53	22	a	a	DET
bracis-19080	53	23	separate	separate	ADJ
bracis-19080	53	24	,	,	PUNCT
bracis-19080	53	25	and	and	CCONJ
bracis-19080	53	26	usually	usually	ADV
bracis-19080	53	27	more	more	ADV
bracis-19080	53	28	specific	specific	ADJ
bracis-19080	53	29	problem	problem	NOUN
bracis-19080	53	30	 	 	SPACE
bracis-19080	54	1	[	[	X
bracis-19080	54	2	25	25	NUM
bracis-19080	54	3	]	]	PUNCT
bracis-19080	54	4	.	.	PUNCT
bracis-19080	55	1	when	when	SCONJ
bracis-19080	55	2	solving	solve	VERB
bracis-19080	55	3	the	the	DET
bracis-19080	55	4	new	new	ADJ
bracis-19080	55	5	problem	problem	NOUN
bracis-19080	55	6	,	,	PUNCT
bracis-19080	55	7	two	two	NUM
bracis-19080	55	8	main	main	ADJ
bracis-19080	55	9	strategies	strategy	NOUN
bracis-19080	55	10	can	can	AUX
bracis-19080	55	11	be	be	AUX
bracis-19080	55	12	adopted	adopt	VERB
bracis-19080	55	13	.	.	PUNCT
bracis-19080	56	1	pre	pre	VERB
bracis-19080	56	2	-	-	ADJ
bracis-19080	56	3	trained	train	VERB
bracis-19080	56	4	models	model	NOUN
bracis-19080	56	5	can	can	AUX
bracis-19080	56	6	be	be	AUX
bracis-19080	56	7	applied	apply	VERB
bracis-19080	56	8	directly	directly	ADV
bracis-19080	56	9	to	to	ADP
bracis-19080	56	10	the	the	DET
bracis-19080	56	11	problem	problem	NOUN
bracis-19080	56	12	at	at	ADP
bracis-19080	56	13	hand	hand	NOUN
bracis-19080	56	14	when	when	SCONJ
bracis-19080	56	15	adjustments	adjustment	NOUN
bracis-19080	56	16	to	to	PART
bracis-19080	56	17	match	match	VERB
bracis-19080	56	18	different	different	ADJ
bracis-19080	56	19	input	input	NOUN
bracis-19080	56	20	and/or	and/or	CCONJ
bracis-19080	56	21	output	output	NOUN
bracis-19080	56	22	dimensionalities	dimensionality	NOUN
bracis-19080	56	23	are	be	AUX
bracis-19080	56	24	feasible	feasible	ADJ
bracis-19080	56	25	.	.	PUNCT
bracis-19080	57	1	though	though	SCONJ
bracis-19080	57	2	pre	pre	ADJ
bracis-19080	57	3	-	-	ADJ
bracis-19080	57	4	trained	train	VERB
bracis-19080	57	5	models	model	NOUN
bracis-19080	57	6	are	be	AUX
bracis-19080	57	7	not	not	PART
bracis-19080	57	8	expected	expect	VERB
bracis-19080	57	9	to	to	PART
bracis-19080	57	10	perform	perform	VERB
bracis-19080	57	11	as	as	ADV
bracis-19080	57	12	well	well	ADV
bracis-19080	57	13	as	as	ADP
bracis-19080	57	14	problem	problem	NOUN
bracis-19080	57	15	-	-	PUNCT
bracis-19080	57	16	specific	specific	ADJ
bracis-19080	57	17	models	model	NOUN
bracis-19080	57	18	,	,	PUNCT
bracis-19080	57	19	they	they	PRON
bracis-19080	57	20	are	be	AUX
bracis-19080	57	21	reusable	reusable	ADJ
bracis-19080	57	22	across	across	ADP
bracis-19080	57	23	different	different	ADJ
bracis-19080	57	24	problems	problem	NOUN
bracis-19080	57	25	.	.	PUNCT
bracis-19080	58	1	fine	fine	ADJ
bracis-19080	58	2	-	-	PUNCT
bracis-19080	58	3	tuned	tune	VERB
bracis-19080	58	4	models	model	NOUN
bracis-19080	58	5	are	be	AUX
bracis-19080	58	6	pre	pre	ADJ
bracis-19080	58	7	-	-	ADJ
bracis-19080	58	8	trained	train	VERB
bracis-19080	58	9	models	model	NOUN
bracis-19080	58	10	that	that	PRON
bracis-19080	58	11	are	be	AUX
bracis-19080	58	12	subject	subject	ADJ
bracis-19080	58	13	to	to	ADP
bracis-19080	58	14	additional	additional	ADJ
bracis-19080	58	15	training	training	NOUN
bracis-19080	58	16	on	on	ADP
bracis-19080	58	17	a	a	DET
bracis-19080	58	18	target	target	NOUN
bracis-19080	58	19	specific	specific	ADJ
bracis-19080	58	20	problem	problem	NOUN
bracis-19080	58	21	.	.	PUNCT
bracis-19080	59	1	in	in	ADP
bracis-19080	59	2	detail	detail	NOUN
bracis-19080	59	3	,	,	PUNCT
bracis-19080	59	4	fine	fine	ADV
bracis-19080	59	5	-	-	PUNCT
bracis-19080	59	6	tuning	tuning	NOUN
bracis-19080	59	7	uses	use	VERB
bracis-19080	59	8	the	the	DET
bracis-19080	59	9	pre	pre	ADJ
bracis-19080	59	10	-	-	ADJ
bracis-19080	59	11	trained	train	VERB
bracis-19080	59	12	model	model	NOUN
bracis-19080	59	13	as	as	ADP
bracis-19080	59	14	a	a	DET
bracis-19080	59	15	starting	starting	NOUN
bracis-19080	59	16	point	point	NOUN
bracis-19080	59	17	and	and	CCONJ
bracis-19080	59	18	runs	run	VERB
bracis-19080	59	19	additional	additional	ADJ
bracis-19080	59	20	training	training	NOUN
bracis-19080	59	21	iterations	iteration	NOUN
bracis-19080	59	22	on	on	ADP
bracis-19080	59	23	the	the	DET
bracis-19080	59	24	new	new	ADJ
bracis-19080	59	25	data	datum	NOUN
bracis-19080	59	26	.	.	PUNCT
bracis-19080	60	1	given	give	VERB
bracis-19080	60	2	the	the	DET
bracis-19080	60	3	benefits	benefit	NOUN
bracis-19080	60	4	of	of	ADP
bracis-19080	60	5	transfer	transfer	NOUN
bracis-19080	60	6	learning	learning	NOUN
bracis-19080	60	7	,	,	PUNCT
bracis-19080	60	8	pre	pre	ADJ
bracis-19080	60	9	-	-	ADJ
bracis-19080	60	10	trained	train	VERB
bracis-19080	60	11	models	model	NOUN
bracis-19080	60	12	have	have	AUX
bracis-19080	60	13	become	become	VERB
bracis-19080	60	14	increasingly	increasingly	ADV
bracis-19080	60	15	publicly	publicly	ADV
bracis-19080	60	16	available	available	ADJ
bracis-19080	60	17	.	.	PUNCT
bracis-19080	61	1	combined	combine	VERB
bracis-19080	61	2	with	with	ADP
bracis-19080	61	3	fine	fine	ADV
bracis-19080	61	4	-	-	PUNCT
bracis-19080	61	5	tuning	tuning	NOUN
bracis-19080	61	6	,	,	PUNCT
bracis-19080	61	7	the	the	DET
bracis-19080	61	8	same	same	ADJ
bracis-19080	61	9	pre	pre	ADJ
bracis-19080	61	10	-	-	ADJ
bracis-19080	61	11	trained	train	VERB
bracis-19080	61	12	model	model	NOUN
bracis-19080	61	13	can	can	AUX
bracis-19080	61	14	become	become	VERB
bracis-19080	61	15	multiple	multiple	ADJ
bracis-19080	61	16	specialized	specialized	ADJ
bracis-19080	61	17	models	model	NOUN
bracis-19080	61	18	without	without	ADP
bracis-19080	61	19	training	training	NOUN
bracis-19080	61	20	from	from	ADP
bracis-19080	61	21	the	the	DET
bracis-19080	61	22	ground	ground	NOUN
bracis-19080	61	23	up	up	ADP
bracis-19080	61	24	,	,	PUNCT
bracis-19080	61	25	saving	save	VERB
bracis-19080	61	26	both	both	CCONJ
bracis-19080	61	27	computing	computing	NOUN
bracis-19080	61	28	and	and	CCONJ
bracis-19080	61	29	time	time	NOUN
bracis-19080	61	30	resources	resource	NOUN
bracis-19080	61	31	.	.	PUNCT
bracis-19080	62	1	in	in	ADP
bracis-19080	62	2	the	the	DET
bracis-19080	62	3	context	context	NOUN
bracis-19080	62	4	of	of	ADP
bracis-19080	62	5	nlp	nlp	NOUN
bracis-19080	62	6	problems	problem	NOUN
bracis-19080	62	7	,	,	PUNCT
bracis-19080	62	8	this	this	PRON
bracis-19080	62	9	has	have	AUX
bracis-19080	62	10	motivated	motivate	VERB
bracis-19080	62	11	researchers	researcher	NOUN
bracis-19080	62	12	to	to	PART
bracis-19080	62	13	propose	propose	VERB
bracis-19080	62	14	general	general	ADJ
bracis-19080	62	15	models	model	NOUN
bracis-19080	62	16	and	and	CCONJ
bracis-19080	62	17	make	make	VERB
bracis-19080	62	18	their	their	PRON
bracis-19080	62	19	pre	pre	ADJ
bracis-19080	62	20	-	-	ADJ
bracis-19080	62	21	trained	train	VERB
bracis-19080	62	22	versions	version	NOUN
bracis-19080	62	23	available	available	ADJ
bracis-19080	62	24	,	,	PUNCT
bracis-19080	62	25	the	the	DET
bracis-19080	62	26	most	most	ADV
bracis-19080	62	27	relevant	relevant	ADJ
bracis-19080	62	28	example	example	NOUN
bracis-19080	62	29	being	be	AUX
bracis-19080	62	30	the	the	DET
bracis-19080	62	31	contextual	contextual	ADJ
bracis-19080	62	32	and	and	CCONJ
bracis-19080	62	33	word	word	NOUN
bracis-19080	62	34	embeddings	embedding	NOUN
bracis-19080	62	35	we	we	PRON
bracis-19080	62	36	discuss	discuss	VERB
bracis-19080	62	37	next	next	ADV
bracis-19080	62	38	.	.	PUNCT
bracis-19080	63	1	2.2	2.2	NUM
bracis-19080	63	2	contextual	contextual	ADJ
bracis-19080	63	3	and	and	CCONJ
bracis-19080	63	4	 	 	SPACE
bracis-19080	63	5	word	word	NOUN
bracis-19080	63	6	embeddings	embedding	VERB
bracis-19080	63	7	a	a	DET
bracis-19080	63	8	statistical	statistical	ADJ
bracis-19080	63	9	language	language	NOUN
bracis-19080	63	10	model	model	NOUN
bracis-19080	63	11	is	be	AUX
bracis-19080	63	12	a	a	DET
bracis-19080	63	13	probability	probability	NOUN
bracis-19080	63	14	distribution	distribution	NOUN
bracis-19080	63	15	function	function	NOUN
bracis-19080	63	16	of	of	ADP
bracis-19080	63	17	word	word	NOUN
bracis-19080	63	18	sequences	sequence	NOUN
bracis-19080	63	19	in	in	ADP
bracis-19080	63	20	a	a	DET
bracis-19080	63	21	given	give	VERB
bracis-19080	63	22	language	language	NOUN
bracis-19080	63	23	.	.	PUNCT
bracis-19080	64	1	training	training	NOUN
bracis-19080	64	2	language	language	NOUN
bracis-19080	64	3	models	model	NOUN
bracis-19080	64	4	is	be	AUX
bracis-19080	64	5	a	a	DET
bracis-19080	64	6	very	very	ADV
bracis-19080	64	7	computationally	computationally	ADV
bracis-19080	64	8	intensive	intensive	ADJ
bracis-19080	64	9	task	task	NOUN
bracis-19080	64	10	,	,	PUNCT
bracis-19080	64	11	easily	easily	ADV
bracis-19080	64	12	plagued	plague	VERB
bracis-19080	64	13	by	by	ADP
bracis-19080	64	14	the	the	DET
bracis-19080	64	15	curse	curse	NOUN
bracis-19080	64	16	of	of	ADP
bracis-19080	64	17	dimensionality	dimensionality	NOUN
bracis-19080	64	18	,	,	PUNCT
bracis-19080	64	19	for	for	ADP
bracis-19080	64	20	two	two	NUM
bracis-19080	64	21	main	main	ADJ
bracis-19080	64	22	reasons	reason	NOUN
bracis-19080	64	23	.	.	PUNCT
bracis-19080	65	1	first	first	ADV
bracis-19080	65	2	,	,	PUNCT
bracis-19080	65	3	the	the	DET
bracis-19080	65	4	number	number	NOUN
bracis-19080	65	5	of	of	ADP
bracis-19080	65	6	word	word	NOUN
bracis-19080	65	7	co	co	NOUN
bracis-19080	65	8	-	-	NOUN
bracis-19080	65	9	occurrence	occurrence	NOUN
bracis-19080	65	10	is	be	AUX
bracis-19080	65	11	typically	typically	ADV
bracis-19080	65	12	large	large	ADJ
bracis-19080	65	13	.	.	PUNCT
bracis-19080	66	1	second	second	ADV
bracis-19080	66	2	,	,	PUNCT
bracis-19080	66	3	the	the	DET
bracis-19080	66	4	word	word	NOUN
bracis-19080	66	5	sequences	sequence	NOUN
bracis-19080	66	6	vary	vary	VERB
bracis-19080	66	7	considerably	considerably	ADV
bracis-19080	66	8	across	across	ADP
bracis-19080	66	9	multiple	multiple	ADJ
bracis-19080	66	10	text	text	NOUN
bracis-19080	66	11	datasets	dataset	NOUN
bracis-19080	66	12	,	,	PUNCT
bracis-19080	66	13	and	and	CCONJ
bracis-19080	66	14	even	even	ADV
bracis-19080	66	15	from	from	ADP
bracis-19080	66	16	training	training	NOUN
bracis-19080	66	17	to	to	ADP
bracis-19080	66	18	test	test	NOUN
bracis-19080	66	19	sets	set	NOUN
bracis-19080	66	20	.	.	PUNCT
bracis-19080	67	1	these	these	DET
bracis-19080	67	2	challenges	challenge	NOUN
bracis-19080	67	3	motivated	motivate	VERB
bracis-19080	67	4	the	the	DET
bracis-19080	67	5	distributed	distribute	VERB
bracis-19080	67	6	word	word	NOUN
bracis-19080	67	7	representation	representation	NOUN
bracis-19080	67	8	 	 	SPACE
bracis-19080	68	1	[	[	X
bracis-19080	68	2	3	3	NUM
bracis-19080	68	3	]	]	PUNCT
bracis-19080	68	4	,	,	PUNCT
bracis-19080	68	5	models	model	NOUN
bracis-19080	68	6	in	in	ADP
bracis-19080	68	7	which	which	PRON
bracis-19080	68	8	both	both	CCONJ
bracis-19080	68	9	the	the	DET
bracis-19080	68	10	distributed	distribute	VERB
bracis-19080	68	11	representation	representation	NOUN
bracis-19080	68	12	of	of	ADP
bracis-19080	68	13	each	each	DET
bracis-19080	68	14	word	word	NOUN
bracis-19080	68	15	and	and	CCONJ
bracis-19080	68	16	the	the	DET
bracis-19080	68	17	probability	probability	NOUN
bracis-19080	68	18	function	function	NOUN
bracis-19080	68	19	for	for	ADP
bracis-19080	68	20	the	the	DET
bracis-19080	68	21	word	word	NOUN
bracis-19080	68	22	sequence	sequence	NOUN
bracis-19080	68	23	are	be	AUX
bracis-19080	68	24	learned	learn	VERB
bracis-19080	68	25	simultaneously	simultaneously	ADV
bracis-19080	68	26	.	.	PUNCT
bracis-19080	69	1	distributed	distribute	VERB
bracis-19080	69	2	representations	representation	NOUN
bracis-19080	69	3	can	can	AUX
bracis-19080	69	4	achieve	achieve	VERB
bracis-19080	69	5	better	well	ADJ
bracis-19080	69	6	generalization	generalization	NOUN
bracis-19080	69	7	since	since	SCONJ
bracis-19080	69	8	these	these	DET
bracis-19080	69	9	models	model	NOUN
bracis-19080	69	10	are	be	AUX
bracis-19080	69	11	able	able	ADJ
bracis-19080	69	12	to	to	PART
bracis-19080	69	13	attribute	attribute	VERB
bracis-19080	69	14	a	a	DET
bracis-19080	69	15	high	high	ADJ
bracis-19080	69	16	probability	probability	NOUN
bracis-19080	69	17	of	of	ADP
bracis-19080	69	18	occurrence	occurrence	NOUN
bracis-19080	69	19	to	to	PART
bracis-19080	69	20	word	word	NOUN
bracis-19080	69	21	sequences	sequence	NOUN
bracis-19080	69	22	it	it	PRON
bracis-19080	69	23	had	have	AUX
bracis-19080	69	24	never	never	ADV
bracis-19080	69	25	seen	see	VERB
bracis-19080	69	26	before	before	ADV
bracis-19080	69	27	if	if	SCONJ
bracis-19080	69	28	similar	similar	ADJ
bracis-19080	69	29	sequences	sequence	NOUN
bracis-19080	69	30	have	have	AUX
bracis-19080	69	31	been	be	AUX
bracis-19080	69	32	provided	provide	VERB
bracis-19080	69	33	during	during	ADP
bracis-19080	69	34	training	training	NOUN
bracis-19080	69	35	.	.	PUNCT
bracis-19080	70	1	popular	popular	ADJ
bracis-19080	70	2	word	word	NOUN
bracis-19080	70	3	embeddings	embedding	NOUN
bracis-19080	70	4	like	like	ADP
bracis-19080	70	5	word2vec	word2vec	NOUN
bracis-19080	70	6	 	 	SPACE
bracis-19080	70	7	[	[	PUNCT
bracis-19080	70	8	13	13	NUM
bracis-19080	70	9	]	]	PUNCT
bracis-19080	70	10	,	,	PUNCT
bracis-19080	70	11	glove	glove	NOUN
bracis-19080	70	12	 	 	SPACE
bracis-19080	71	1	[	[	X
bracis-19080	71	2	14	14	NUM
bracis-19080	71	3	]	]	PUNCT
bracis-19080	71	4	,	,	PUNCT
bracis-19080	71	5	and	and	CCONJ
bracis-19080	71	6	fasttext	fasttext	PROPN
bracis-19080	71	7	 	 	SPACE
bracis-19080	72	1	[	[	X
bracis-19080	72	2	4	4	X
bracis-19080	72	3	]	]	PUNCT
bracis-19080	72	4	consider	consider	VERB
bracis-19080	72	5	all	all	DET
bracis-19080	72	6	sentences	sentence	NOUN
bracis-19080	72	7	in	in	ADP
bracis-19080	72	8	which	which	PRON
bracis-19080	72	9	a	a	DET
bracis-19080	72	10	given	give	VERB
bracis-19080	72	11	word	word	NOUN
bracis-19080	72	12	appear	appear	VERB
bracis-19080	72	13	to	to	PART
bracis-19080	72	14	build	build	VERB
bracis-19080	72	15	a	a	DET
bracis-19080	72	16	global	global	ADJ
bracis-19080	72	17	vector	vector	NOUN
bracis-19080	72	18	representation	representation	NOUN
bracis-19080	72	19	for	for	ADP
bracis-19080	72	20	each	each	DET
bracis-19080	72	21	word	word	NOUN
bracis-19080	72	22	in	in	ADP
bracis-19080	72	23	the	the	DET
bracis-19080	72	24	corpus	corpus	NOUN
bracis-19080	72	25	.	.	PUNCT
bracis-19080	73	1	however	however	ADV
bracis-19080	73	2	,	,	PUNCT
bracis-19080	73	3	a	a	DET
bracis-19080	73	4	word	word	NOUN
bracis-19080	73	5	can	can	AUX
bracis-19080	73	6	have	have	VERB
bracis-19080	73	7	different	different	ADJ
bracis-19080	73	8	meanings	meaning	NOUN
bracis-19080	73	9	depending	depend	VERB
bracis-19080	73	10	on	on	ADP
bracis-19080	73	11	the	the	DET
bracis-19080	73	12	sentence	sentence	NOUN
bracis-19080	73	13	and	and	CCONJ
bracis-19080	73	14	context	context	NOUN
bracis-19080	73	15	it	it	PRON
bracis-19080	73	16	is	be	AUX
bracis-19080	73	17	in	in	ADP
bracis-19080	73	18	.	.	PUNCT
bracis-19080	74	1	to	to	PART
bracis-19080	74	2	solve	solve	VERB
bracis-19080	74	3	this	this	DET
bracis-19080	74	4	problem	problem	NOUN
bracis-19080	74	5	,	,	PUNCT
bracis-19080	74	6	different	different	ADJ
bracis-19080	74	7	contextual	contextual	ADJ
bracis-19080	74	8	embeddings	embedding	NOUN
bracis-19080	74	9	were	be	AUX
bracis-19080	74	10	proposed	propose	VERB
bracis-19080	74	11	.	.	PUNCT
bracis-19080	75	1	next	next	ADV
bracis-19080	75	2	,	,	PUNCT
bracis-19080	75	3	we	we	PRON
bracis-19080	75	4	discuss	discuss	VERB
bracis-19080	75	5	the	the	DET
bracis-19080	75	6	two	two	NUM
bracis-19080	75	7	contextual	contextual	ADJ
bracis-19080	75	8	approaches	approach	NOUN
bracis-19080	75	9	we	we	PRON
bracis-19080	75	10	use	use	VERB
bracis-19080	75	11	in	in	ADP
bracis-19080	75	12	this	this	DET
bracis-19080	75	13	work	work	NOUN
bracis-19080	75	14	,	,	PUNCT
bracis-19080	75	15	namely	namely	ADV
bracis-19080	75	16	bert	bert	NOUN
bracis-19080	75	17	and	and	CCONJ
bracis-19080	75	18	sentence	sentence	NOUN
bracis-19080	75	19	-	-	PUNCT
bracis-19080	75	20	bert	bert	NOUN
bracis-19080	75	21	.	.	PUNCT
bracis-19080	76	1	2.3	2.3	NUM
bracis-19080	76	2	bert	bert	NOUN
bracis-19080	76	3	the	the	DET
bracis-19080	76	4	bidirectional	bidirectional	ADJ
bracis-19080	76	5	encoder	encoder	NOUN
bracis-19080	76	6	representations	representation	VERB
bracis-19080	76	7	from	from	ADP
bracis-19080	76	8	transformers	transformer	NOUN
bracis-19080	76	9	 	 	SPACE
bracis-19080	76	10	(	(	PUNCT
bracis-19080	76	11	bert	bert	NOUN
bracis-19080	76	12	)	)	PUNCT
bracis-19080	76	13	architecture	architecture	NOUN
bracis-19080	76	14	brought	bring	VERB
bracis-19080	76	15	great	great	ADJ
bracis-19080	76	16	advances	advance	NOUN
bracis-19080	76	17	to	to	ADP
bracis-19080	76	18	the	the	DET
bracis-19080	76	19	nlp	nlp	NOUN
bracis-19080	76	20	field	field	NOUN
bracis-19080	76	21	,	,	PUNCT
bracis-19080	76	22	lately	lately	ADV
bracis-19080	76	23	becoming	become	VERB
bracis-19080	76	24	the	the	DET
bracis-19080	76	25	current	current	ADJ
bracis-19080	76	26	state	state	NOUN
bracis-19080	76	27	-	-	PUNCT
bracis-19080	76	28	of	of	ADP
bracis-19080	76	29	-	-	PUNCT
bracis-19080	76	30	the	the	DET
bracis-19080	76	31	-	-	PUNCT
bracis-19080	76	32	art	art	NOUN
bracis-19080	76	33	in	in	ADP
bracis-19080	76	34	different	different	ADJ
bracis-19080	76	35	tasks	task	NOUN
bracis-19080	77	1	[	[	X
bracis-19080	77	2	7	7	NUM
bracis-19080	77	3	]	]	PUNCT
bracis-19080	77	4	.	.	PUNCT
bracis-19080	78	1	bert	bert	PROPN
bracis-19080	78	2	is	be	AUX
bracis-19080	78	3	a	a	DET
bracis-19080	78	4	language	language	NOUN
bracis-19080	78	5	model	model	NOUN
bracis-19080	78	6	designed	design	VERB
bracis-19080	78	7	to	to	PART
bracis-19080	78	8	create	create	VERB
bracis-19080	78	9	deep	deep	ADJ
bracis-19080	78	10	bidirectional	bidirectional	ADJ
bracis-19080	78	11	vector	vector	NOUN
bracis-19080	78	12	representations	representation	NOUN
bracis-19080	78	13	from	from	ADP
bracis-19080	78	14	unsupervised	unsupervised	ADJ
bracis-19080	78	15	texts	text	NOUN
bracis-19080	78	16	.	.	PUNCT
bracis-19080	79	1	the	the	DET
bracis-19080	79	2	bidirectional	bidirectional	ADJ
bracis-19080	79	3	approach	approach	NOUN
bracis-19080	79	4	models	model	NOUN
bracis-19080	79	5	context	context	VERB
bracis-19080	79	6	both	both	PRON
bracis-19080	79	7	to	to	ADP
bracis-19080	79	8	the	the	DET
bracis-19080	79	9	right	right	NOUN
bracis-19080	79	10	and	and	CCONJ
bracis-19080	79	11	to	to	ADP
bracis-19080	79	12	the	the	DET
bracis-19080	79	13	left	left	NOUN
bracis-19080	79	14	of	of	ADP
bracis-19080	79	15	each	each	PRON
bracis-19080	79	16	token	token	VERB
bracis-19080	79	17	in	in	ADP
bracis-19080	79	18	the	the	DET
bracis-19080	79	19	input	input	NOUN
bracis-19080	79	20	sequence	sequence	NOUN
bracis-19080	79	21	.	.	PUNCT
bracis-19080	80	1	the	the	DET
bracis-19080	80	2	bert	bert	PROPN
bracis-19080	80	3	architecture	architecture	NOUN
bracis-19080	80	4	can	can	AUX
bracis-19080	80	5	be	be	AUX
bracis-19080	80	6	seen	see	VERB
bracis-19080	80	7	on	on	ADP
bracis-19080	80	8	fig	fig	NOUN
bracis-19080	80	9	.	.	PUNCT
bracis-19080	80	10	 	 	SPACE
bracis-19080	81	1	1	1	NUM
bracis-19080	81	2	,	,	PUNCT
bracis-19080	81	3	where	where	SCONJ
bracis-19080	81	4	the	the	DET
bracis-19080	81	5	pre	pre	ADJ
bracis-19080	81	6	-	-	ADJ
bracis-19080	81	7	trained	train	VERB
bracis-19080	81	8	scenario	scenario	NOUN
bracis-19080	81	9	is	be	AUX
bracis-19080	81	10	seen	see	VERB
bracis-19080	81	11	on	on	ADP
bracis-19080	81	12	the	the	DET
bracis-19080	81	13	left	left	NOUN
bracis-19080	81	14	and	and	CCONJ
bracis-19080	81	15	the	the	DET
bracis-19080	81	16	fine	fine	ADV
bracis-19080	81	17	-	-	PUNCT
bracis-19080	81	18	tuning	tune	VERB
bracis-19080	81	19	scenario	scenario	NOUN
bracis-19080	81	20	is	be	AUX
bracis-19080	81	21	illustrated	illustrate	VERB
bracis-19080	81	22	for	for	ADP
bracis-19080	81	23	different	different	ADJ
bracis-19080	81	24	tasks	task	NOUN
bracis-19080	81	25	on	on	ADP
bracis-19080	81	26	the	the	DET
bracis-19080	81	27	right	right	NOUN
bracis-19080	81	28	.	.	PUNCT
bracis-19080	82	1	fig	fig	NOUN
bracis-19080	82	2	.	.	PUNCT
bracis-19080	83	1	1	1	X
bracis-19080	83	2	.	.	X
bracis-19080	83	3	the	the	DET
bracis-19080	83	4	bert	bert	PROPN
bracis-19080	83	5	architecture	architecture	NOUN
bracis-19080	83	6	 	 	SPACE
bracis-19080	84	1	[	[	X
bracis-19080	84	2	7	7	X
bracis-19080	84	3	]	]	PUNCT
bracis-19080	84	4	for	for	ADP
bracis-19080	84	5	pre	pre	ADJ
bracis-19080	84	6	-	-	NOUN
bracis-19080	84	7	training	training	ADJ
bracis-19080	84	8	 	 	SPACE
bracis-19080	84	9	(	(	PUNCT
bracis-19080	84	10	left	leave	VERB
bracis-19080	84	11	)	)	PUNCT
bracis-19080	84	12	or	or	CCONJ
bracis-19080	84	13	fine	fine	ADV
bracis-19080	84	14	-	-	PUNCT
bracis-19080	84	15	tuning	tuning	NOUN
bracis-19080	84	16	 	 	SPACE
bracis-19080	84	17	(	(	PUNCT
bracis-19080	84	18	right	right	NOUN
bracis-19080	84	19	)	)	PUNCT
bracis-19080	84	20	.	.	PUNCT
bracis-19080	85	1	full	full	ADJ
bracis-19080	85	2	size	size	NOUN
bracis-19080	85	3	image	image	NOUN
bracis-19080	85	4	an	an	DET
bracis-19080	85	5	improvement	improvement	NOUN
bracis-19080	85	6	over	over	ADP
bracis-19080	85	7	other	other	ADJ
bracis-19080	85	8	architectures	architecture	NOUN
bracis-19080	85	9	is	be	AUX
bracis-19080	85	10	that	that	SCONJ
bracis-19080	85	11	bert	bert	PROPN
bracis-19080	85	12	can	can	AUX
bracis-19080	85	13	be	be	AUX
bracis-19080	85	14	applied	apply	VERB
bracis-19080	85	15	to	to	ADP
bracis-19080	85	16	tasks	task	NOUN
bracis-19080	85	17	that	that	PRON
bracis-19080	85	18	take	take	VERB
bracis-19080	85	19	as	as	ADP
bracis-19080	85	20	input	input	NOUN
bracis-19080	85	21	either	either	CCONJ
bracis-19080	85	22	individual	individual	ADJ
bracis-19080	85	23	sentences	sentence	NOUN
bracis-19080	85	24	or	or	CCONJ
bracis-19080	85	25	pairs	pair	NOUN
bracis-19080	85	26	thereof	thereof	ADV
bracis-19080	85	27	.	.	PUNCT
bracis-19080	86	1	in	in	ADP
bracis-19080	86	2	the	the	DET
bracis-19080	86	3	latter	latter	ADJ
bracis-19080	86	4	case	case	NOUN
bracis-19080	86	5	,	,	PUNCT
bracis-19080	86	6	both	both	DET
bracis-19080	86	7	sentences	sentence	NOUN
bracis-19080	86	8	are	be	AUX
bracis-19080	86	9	separated	separate	VERB
bracis-19080	86	10	by	by	ADP
bracis-19080	86	11	a	a	DET
bracis-19080	86	12	special	special	ADJ
bracis-19080	86	13	token	token	NOUN
bracis-19080	86	14	dubbed	dub	VERB
bracis-19080	86	15	[	[	X
bracis-19080	86	16	sep	sep	NOUN
bracis-19080	86	17	]	]	X
bracis-19080	86	18	.	.	PUNCT
bracis-19080	87	1	whether	whether	SCONJ
bracis-19080	87	2	for	for	ADP
bracis-19080	87	3	individual	individual	ADJ
bracis-19080	87	4	or	or	CCONJ
bracis-19080	87	5	sentence	sentence	NOUN
bracis-19080	87	6	pairs	pair	NOUN
bracis-19080	87	7	,	,	PUNCT
bracis-19080	87	8	the	the	DET
bracis-19080	87	9	first	first	ADJ
bracis-19080	87	10	token	token	VERB
bracis-19080	87	11	fed	fed	PROPN
bracis-19080	87	12	to	to	ADP
bracis-19080	87	13	the	the	DET
bracis-19080	87	14	network	network	NOUN
bracis-19080	87	15	is	be	AUX
bracis-19080	87	16	a	a	DET
bracis-19080	87	17	special	special	ADJ
bracis-19080	87	18	token	token	VERB
bracis-19080	87	19	dubbed	dub	VERB
bracis-19080	88	1	[	[	X
bracis-19080	88	2	cls	cls	NOUN
bracis-19080	88	3	]	]	PUNCT
bracis-19080	88	4	,	,	PUNCT
bracis-19080	88	5	which	which	PRON
bracis-19080	88	6	serves	serve	VERB
bracis-19080	88	7	as	as	ADP
bracis-19080	88	8	encoding	encode	VERB
bracis-19080	88	9	for	for	ADP
bracis-19080	88	10	supervised	supervised	ADJ
bracis-19080	88	11	tasks	task	NOUN
bracis-19080	88	12	after	after	ADP
bracis-19080	88	13	training	training	NOUN
bracis-19080	88	14	.	.	PUNCT
bracis-19080	89	1	alternatively	alternatively	ADV
bracis-19080	89	2	,	,	PUNCT
bracis-19080	89	3	a	a	DET
bracis-19080	89	4	bert	bert	NOUN
bracis-19080	89	5	embedding	embed	VERB
bracis-19080	89	6	can	can	AUX
bracis-19080	89	7	be	be	AUX
bracis-19080	89	8	obtained	obtain	VERB
bracis-19080	89	9	by	by	ADP
bracis-19080	89	10	aggregating	aggregate	VERB
bracis-19080	89	11	the	the	DET
bracis-19080	89	12	three	three	NUM
bracis-19080	89	13	types	type	NOUN
bracis-19080	89	14	of	of	ADP
bracis-19080	89	15	internal	internal	ADJ
bracis-19080	89	16	bert	bert	PROPN
bracis-19080	89	17	embeddings	embedding	NOUN
bracis-19080	89	18	,	,	PUNCT
bracis-19080	89	19	namely	namely	ADV
bracis-19080	89	20	token	token	ADJ
bracis-19080	89	21	,	,	PUNCT
bracis-19080	89	22	segment	segment	NOUN
bracis-19080	89	23	,	,	PUNCT
bracis-19080	89	24	and	and	CCONJ
bracis-19080	89	25	position	position	NOUN
bracis-19080	89	26	.	.	PUNCT
bracis-19080	90	1	2.4	2.4	NUM
bracis-19080	90	2	sentence	sentence	NOUN
bracis-19080	90	3	-	-	PUNCT
bracis-19080	90	4	bert	bert	NOUN
bracis-19080	90	5	(	(	PUNCT
bracis-19080	90	6	sbert	sbert	NOUN
bracis-19080	90	7	)	)	PUNCT
bracis-19080	90	8	although	although	SCONJ
bracis-19080	90	9	bert	bert	PROPN
bracis-19080	90	10	has	have	AUX
bracis-19080	90	11	become	become	VERB
bracis-19080	90	12	the	the	DET
bracis-19080	90	13	state	state	NOUN
bracis-19080	90	14	-	-	PUNCT
bracis-19080	90	15	of	of	ADP
bracis-19080	90	16	-	-	PUNCT
bracis-19080	90	17	the	the	DET
bracis-19080	90	18	-	-	PUNCT
bracis-19080	90	19	art	art	NOUN
bracis-19080	90	20	for	for	ADP
bracis-19080	90	21	several	several	ADJ
bracis-19080	90	22	nlp	nlp	NOUN
bracis-19080	90	23	tasks	task	NOUN
bracis-19080	90	24	,	,	PUNCT
bracis-19080	90	25	the	the	DET
bracis-19080	90	26	algorithm	algorithm	NOUN
bracis-19080	90	27	faces	face	VERB
bracis-19080	90	28	scalability	scalability	NOUN
bracis-19080	90	29	issues	issue	NOUN
bracis-19080	90	30	in	in	ADP
bracis-19080	90	31	tasks	task	NOUN
bracis-19080	90	32	such	such	ADJ
bracis-19080	90	33	as	as	ADP
bracis-19080	90	34	sts	st	NOUN
bracis-19080	90	35	where	where	SCONJ
bracis-19080	90	36	there	there	PRON
bracis-19080	90	37	is	be	VERB
bracis-19080	90	38	the	the	DET
bracis-19080	90	39	need	need	NOUN
bracis-19080	90	40	to	to	PART
bracis-19080	90	41	feed	feed	VERB
bracis-19080	90	42	the	the	DET
bracis-19080	90	43	network	network	NOUN
bracis-19080	90	44	with	with	ADP
bracis-19080	90	45	sentence	sentence	NOUN
bracis-19080	90	46	pairs	pair	NOUN
bracis-19080	90	47	.	.	PUNCT
bracis-19080	91	1	according	accord	VERB
bracis-19080	91	2	to	to	ADP
bracis-19080	91	3	reimers	reimer	NOUN
bracis-19080	91	4	and	and	CCONJ
bracis-19080	91	5	gurevych	gurevych	NOUN
bracis-19080	91	6	 	 	SPACE
bracis-19080	91	7	[	[	X
bracis-19080	91	8	17	17	NUM
bracis-19080	91	9	]	]	PUNCT
bracis-19080	91	10	,	,	PUNCT
bracis-19080	91	11	even	even	ADV
bracis-19080	91	12	a	a	DET
bracis-19080	91	13	moderate	moderate	ADJ
bracis-19080	91	14	sized	sized	ADJ
bracis-19080	91	15	training	training	NOUN
bracis-19080	91	16	corpus	corpus	NOUN
bracis-19080	91	17	of	of	ADP
bracis-19080	91	18	10000	10000	NUM
bracis-19080	91	19	sentences	sentence	NOUN
bracis-19080	91	20	requires	require	VERB
bracis-19080	91	21	50	50	NUM
bracis-19080	91	22	million	million	NUM
bracis-19080	91	23	computational	computational	ADJ
bracis-19080	91	24	inferences	inference	NOUN
bracis-19080	91	25	,	,	PUNCT
bracis-19080	91	26	rendering	render	VERB
bracis-19080	91	27	bert	bert	NOUN
bracis-19080	91	28	impractical	impractical	ADJ
bracis-19080	91	29	for	for	ADP
bracis-19080	91	30	sts	st	NOUN
bracis-19080	91	31	.	.	PUNCT
bracis-19080	92	1	as	as	ADP
bracis-19080	92	2	an	an	DET
bracis-19080	92	3	alternative	alternative	NOUN
bracis-19080	92	4	,	,	PUNCT
bracis-19080	92	5	sbert	sbert	NOUN
bracis-19080	92	6	was	be	AUX
bracis-19080	92	7	proposed	propose	VERB
bracis-19080	92	8	as	as	ADP
bracis-19080	92	9	a	a	DET
bracis-19080	92	10	more	more	ADV
bracis-19080	92	11	efficient	efficient	ADJ
bracis-19080	92	12	approach	approach	NOUN
bracis-19080	92	13	to	to	ADP
bracis-19080	92	14	this	this	DET
bracis-19080	92	15	type	type	NOUN
bracis-19080	92	16	of	of	ADP
bracis-19080	92	17	tasks	task	NOUN
bracis-19080	92	18	.	.	PUNCT
bracis-19080	93	1	sbert	sbert	NOUN
bracis-19080	93	2	 	 	SPACE
bracis-19080	94	1	[	[	X
bracis-19080	94	2	17	17	NUM
bracis-19080	94	3	]	]	PUNCT
bracis-19080	94	4	is	be	AUX
bracis-19080	94	5	a	a	DET
bracis-19080	94	6	siamese	siamese	ADJ
bracis-19080	94	7	architecture	architecture	NOUN
bracis-19080	94	8	in	in	ADP
bracis-19080	94	9	which	which	PRON
bracis-19080	94	10	the	the	DET
bracis-19080	94	11	sentence	sentence	NOUN
bracis-19080	94	12	pairs	pair	NOUN
bracis-19080	94	13	are	be	AUX
bracis-19080	94	14	fed	feed	VERB
bracis-19080	94	15	to	to	ADP
bracis-19080	94	16	two	two	NUM
bracis-19080	94	17	parallel	parallel	ADJ
bracis-19080	94	18	but	but	CCONJ
bracis-19080	94	19	connected	connected	ADJ
bracis-19080	94	20	network	network	NOUN
bracis-19080	94	21	pathways	pathway	NOUN
bracis-19080	94	22	to	to	PART
bracis-19080	94	23	produce	produce	VERB
bracis-19080	94	24	an	an	DET
bracis-19080	94	25	estimate	estimate	NOUN
bracis-19080	94	26	of	of	ADP
bracis-19080	94	27	the	the	DET
bracis-19080	94	28	similarity	similarity	NOUN
bracis-19080	94	29	between	between	ADP
bracis-19080	94	30	the	the	DET
bracis-19080	94	31	sentences	sentence	NOUN
bracis-19080	94	32	(	(	PUNCT
bracis-19080	94	33	fig	fig	NOUN
bracis-19080	94	34	.	.	PUNCT
bracis-19080	94	35	 	 	SPACE
bracis-19080	94	36	2a	2a	NUM
bracis-19080	94	37	)	)	PUNCT
bracis-19080	94	38	.	.	PUNCT
bracis-19080	95	1	the	the	DET
bracis-19080	95	2	first	first	ADJ
bracis-19080	95	3	layers	layer	NOUN
bracis-19080	95	4	consist	consist	VERB
bracis-19080	95	5	of	of	ADP
bracis-19080	95	6	regular	regular	ADJ
bracis-19080	95	7	bert	bert	NOUN
bracis-19080	95	8	networks	network	NOUN
bracis-19080	95	9	followed	follow	VERB
bracis-19080	95	10	by	by	ADP
bracis-19080	95	11	a	a	DET
bracis-19080	95	12	pooling	pool	VERB
bracis-19080	95	13	operation	operation	NOUN
bracis-19080	95	14	to	to	PART
bracis-19080	95	15	ensure	ensure	VERB
bracis-19080	95	16	embeddings	embedding	NOUN
bracis-19080	95	17	produced	produce	VERB
bracis-19080	95	18	for	for	ADP
bracis-19080	95	19	each	each	DET
bracis-19080	95	20	sentence	sentence	NOUN
bracis-19080	95	21	have	have	VERB
bracis-19080	95	22	the	the	DET
bracis-19080	95	23	same	same	ADJ
bracis-19080	95	24	vector	vector	NOUN
bracis-19080	95	25	size	size	NOUN
bracis-19080	95	26	.	.	PUNCT
bracis-19080	96	1	the	the	DET
bracis-19080	96	2	network	network	NOUN
bracis-19080	96	3	outputs	output	VERB
bracis-19080	96	4	the	the	DET
bracis-19080	96	5	cosine	cosine	NOUN
bracis-19080	96	6	similarity	similarity	NOUN
bracis-19080	96	7	between	between	ADP
bracis-19080	96	8	these	these	DET
bracis-19080	96	9	embeddings	embedding	NOUN
bracis-19080	96	10	.	.	PUNCT
bracis-19080	97	1	since	since	SCONJ
bracis-19080	97	2	both	both	DET
bracis-19080	97	3	bert	bert	PROPN
bracis-19080	97	4	networks	network	NOUN
bracis-19080	97	5	take	take	VERB
bracis-19080	97	6	a	a	DET
bracis-19080	97	7	single	single	ADJ
bracis-19080	97	8	sentence	sentence	NOUN
bracis-19080	97	9	and	and	CCONJ
bracis-19080	97	10	their	their	PRON
bracis-19080	97	11	weights	weight	NOUN
bracis-19080	97	12	are	be	AUX
bracis-19080	97	13	shared	share	VERB
bracis-19080	97	14	,	,	PUNCT
bracis-19080	97	15	the	the	DET
bracis-19080	97	16	number	number	NOUN
bracis-19080	97	17	of	of	ADP
bracis-19080	97	18	computational	computational	ADJ
bracis-19080	97	19	inferences	inference	NOUN
bracis-19080	97	20	is	be	AUX
bracis-19080	97	21	much	much	ADV
bracis-19080	97	22	reduced	reduce	VERB
bracis-19080	97	23	 	 	SPACE
bracis-19080	98	1	[	[	X
bracis-19080	98	2	17	17	NUM
bracis-19080	98	3	]	]	PUNCT
bracis-19080	98	4	.	.	PUNCT
bracis-19080	99	1	fig	fig	NOUN
bracis-19080	99	2	.	.	PUNCT
bracis-19080	100	1	2	2	X
bracis-19080	100	2	.	.	PUNCT
bracis-19080	100	3	sbert	sbert	NOUN
bracis-19080	100	4	architecture	architecture	NOUN
bracis-19080	100	5	 	 	SPACE
bracis-19080	100	6	(	(	PUNCT
bracis-19080	100	7	left	left	ADJ
bracis-19080	100	8	)	)	PUNCT
bracis-19080	100	9	and	and	CCONJ
bracis-19080	100	10	multilingual	multilingual	ADJ
bracis-19080	100	11	knowledge	knowledge	NOUN
bracis-19080	100	12	distillation	distillation	NOUN
bracis-19080	100	13	 	 	SPACE
bracis-19080	100	14	(	(	PUNCT
bracis-19080	100	15	right	right	NOUN
bracis-19080	100	16	)	)	PUNCT
bracis-19080	100	17	.	.	PUNCT
bracis-19080	101	1	full	full	ADJ
bracis-19080	101	2	size	size	NOUN
bracis-19080	101	3	image	image	NOUN
bracis-19080	101	4	2.5	2.5	NUM
bracis-19080	101	5	knowledge	knowledge	NOUN
bracis-19080	101	6	distillation	distillation	NOUN
bracis-19080	101	7	another	another	DET
bracis-19080	101	8	concept	concept	NOUN
bracis-19080	101	9	relevant	relevant	ADJ
bracis-19080	101	10	to	to	ADP
bracis-19080	101	11	this	this	DET
bracis-19080	101	12	study	study	NOUN
bracis-19080	101	13	is	be	AUX
bracis-19080	101	14	knowledge	knowledge	NOUN
bracis-19080	101	15	distillation	distillation	NOUN
bracis-19080	101	16	,	,	PUNCT
bracis-19080	101	17	one	one	NUM
bracis-19080	101	18	of	of	ADP
bracis-19080	101	19	the	the	DET
bracis-19080	101	20	deep	deep	ADJ
bracis-19080	101	21	learning	learning	NOUN
bracis-19080	101	22	training	training	NOUN
bracis-19080	101	23	approaches	approach	NOUN
bracis-19080	101	24	where	where	SCONJ
bracis-19080	101	25	one	one	NUM
bracis-19080	101	26	model	model	NOUN
bracis-19080	101	27	is	be	AUX
bracis-19080	101	28	trained	train	VERB
bracis-19080	101	29	to	to	PART
bracis-19080	101	30	learn	learn	VERB
bracis-19080	101	31	the	the	DET
bracis-19080	101	32	knowledge	knowledge	NOUN
bracis-19080	101	33	encoded	encode	VERB
bracis-19080	101	34	by	by	ADP
bracis-19080	101	35	another	another	PRON
bracis-19080	101	36	 	 	SPACE
bracis-19080	102	1	[	[	X
bracis-19080	102	2	10	10	NUM
bracis-19080	102	3	]	]	PUNCT
bracis-19080	102	4	.	.	PUNCT
bracis-19080	103	1	in	in	ADP
bracis-19080	103	2	this	this	DET
bracis-19080	103	3	training	training	NOUN
bracis-19080	103	4	approach	approach	NOUN
bracis-19080	103	5	,	,	PUNCT
bracis-19080	103	6	a	a	DET
bracis-19080	103	7	student	student	NOUN
bracis-19080	103	8	model	model	NOUN
bracis-19080	103	9	attempts	attempt	VERB
bracis-19080	103	10	to	to	PART
bracis-19080	103	11	emulate	emulate	VERB
bracis-19080	103	12	a	a	DET
bracis-19080	103	13	teacher	teacher	NOUN
bracis-19080	103	14	model	model	NOUN
bracis-19080	103	15	with	with	ADP
bracis-19080	103	16	similar	similar	ADJ
bracis-19080	103	17	(	(	PUNCT
bracis-19080	103	18	or	or	CCONJ
bracis-19080	103	19	even	even	ADV
bracis-19080	103	20	higher	high	ADJ
bracis-19080	103	21	)	)	PUNCT
bracis-19080	103	22	performance	performance	NOUN
bracis-19080	103	23	.	.	PUNCT
bracis-19080	104	1	a	a	DET
bracis-19080	104	2	knowledge	knowledge	NOUN
bracis-19080	104	3	distillation	distillation	NOUN
bracis-19080	104	4	-	-	PUNCT
bracis-19080	104	5	based	base	VERB
bracis-19080	104	6	approach	approach	NOUN
bracis-19080	104	7	has	have	AUX
bracis-19080	104	8	been	be	AUX
bracis-19080	104	9	proposed	propose	VERB
bracis-19080	104	10	by	by	ADP
bracis-19080	104	11	reimers	reimer	NOUN
bracis-19080	104	12	and	and	CCONJ
bracis-19080	104	13	gurevych	gurevych	NOUN
bracis-19080	104	14	 	 	SPACE
bracis-19080	105	1	[	[	X
bracis-19080	105	2	17	17	NUM
bracis-19080	105	3	]	]	PUNCT
bracis-19080	105	4	to	to	PART
bracis-19080	105	5	produce	produce	VERB
bracis-19080	105	6	multilingual	multilingual	ADJ
bracis-19080	105	7	sbert	sbert	NOUN
bracis-19080	105	8	models	model	NOUN
bracis-19080	105	9	and	and	CCONJ
bracis-19080	105	10	is	be	AUX
bracis-19080	105	11	illustrated	illustrate	VERB
bracis-19080	105	12	in	in	ADP
bracis-19080	105	13	fig	fig	NOUN
bracis-19080	105	14	.	.	PUNCT
bracis-19080	105	15	 	 	SPACE
bracis-19080	106	1	2b	2b	NOUN
bracis-19080	106	2	.	.	PUNCT
bracis-19080	107	1	the	the	DET
bracis-19080	107	2	teacher	teacher	NOUN
bracis-19080	107	3	model	model	NOUN
bracis-19080	107	4	is	be	AUX
bracis-19080	107	5	an	an	DET
bracis-19080	107	6	english	english	ADJ
bracis-19080	107	7	pre	pre	VERB
bracis-19080	107	8	-	-	ADJ
bracis-19080	107	9	trained	train	VERB
bracis-19080	107	10	sbert	sbert	NOUN
bracis-19080	107	11	model	model	NOUN
bracis-19080	107	12	and	and	CCONJ
bracis-19080	107	13	the	the	DET
bracis-19080	107	14	student	student	NOUN
bracis-19080	107	15	model	model	NOUN
bracis-19080	107	16	is	be	AUX
bracis-19080	107	17	trained	train	VERB
bracis-19080	107	18	to	to	PART
bracis-19080	107	19	produce	produce	VERB
bracis-19080	107	20	multilingual	multilingual	ADJ
bracis-19080	107	21	embeddings	embedding	NOUN
bracis-19080	107	22	.	.	PUNCT
bracis-19080	108	1	training	training	NOUN
bracis-19080	108	2	is	be	AUX
bracis-19080	108	3	structured	structure	VERB
bracis-19080	108	4	as	as	SCONJ
bracis-19080	108	5	follows	follow	VERB
bracis-19080	108	6	.	.	PUNCT
bracis-19080	109	1	first	first	ADV
bracis-19080	109	2	,	,	PUNCT
bracis-19080	109	3	the	the	DET
bracis-19080	109	4	input	input	NOUN
bracis-19080	109	5	for	for	ADP
bracis-19080	109	6	teacher	teacher	NOUN
bracis-19080	109	7	and	and	CCONJ
bracis-19080	109	8	student	student	NOUN
bracis-19080	109	9	models	model	NOUN
bracis-19080	109	10	are	be	AUX
bracis-19080	109	11	different	different	ADJ
bracis-19080	109	12	,	,	PUNCT
bracis-19080	109	13	since	since	SCONJ
bracis-19080	109	14	the	the	DET
bracis-19080	109	15	original	original	ADJ
bracis-19080	109	16	sentence	sentence	NOUN
bracis-19080	109	17	fed	feed	VERB
bracis-19080	109	18	to	to	ADP
bracis-19080	109	19	the	the	DET
bracis-19080	109	20	teacher	teacher	NOUN
bracis-19080	109	21	is	be	AUX
bracis-19080	109	22	enriched	enrich	VERB
bracis-19080	109	23	with	with	ADP
bracis-19080	109	24	parallel	parallel	ADJ
bracis-19080	109	25	,	,	PUNCT
bracis-19080	109	26	translated	translate	VERB
bracis-19080	109	27	versions	version	NOUN
bracis-19080	109	28	for	for	ADP
bracis-19080	109	29	the	the	DET
bracis-19080	109	30	student	student	NOUN
bracis-19080	109	31	.	.	PUNCT
bracis-19080	110	1	second	second	ADJ
bracis-19080	110	2	,	,	PUNCT
bracis-19080	110	3	the	the	DET
bracis-19080	110	4	objective	objective	ADJ
bracis-19080	110	5	function	function	NOUN
bracis-19080	110	6	is	be	AUX
bracis-19080	110	7	to	to	PART
bracis-19080	110	8	reduce	reduce	VERB
bracis-19080	110	9	the	the	DET
bracis-19080	110	10	mean	mean	ADJ
bracis-19080	110	11	squared	square	VERB
bracis-19080	110	12	error	error	NOUN
bracis-19080	110	13	losses	loss	NOUN
bracis-19080	110	14	comparing	compare	VERB
bracis-19080	110	15	every	every	DET
bracis-19080	110	16	student	student	NOUN
bracis-19080	110	17	embedding	embed	VERB
bracis-19080	110	18	produced	produce	VERB
bracis-19080	110	19	with	with	ADP
bracis-19080	110	20	the	the	DET
bracis-19080	110	21	original	original	ADJ
bracis-19080	110	22	embedding	embedding	NOUN
bracis-19080	110	23	produced	produce	VERB
bracis-19080	110	24	by	by	ADP
bracis-19080	110	25	the	the	DET
bracis-19080	110	26	teacher	teacher	NOUN
bracis-19080	110	27	.	.	PUNCT
bracis-19080	111	1	effectively	effectively	ADV
bracis-19080	111	2	,	,	PUNCT
bracis-19080	111	3	translated	translate	VERB
bracis-19080	111	4	sentences	sentence	NOUN
bracis-19080	111	5	are	be	AUX
bracis-19080	111	6	mapped	map	VERB
bracis-19080	111	7	to	to	ADP
bracis-19080	111	8	the	the	DET
bracis-19080	111	9	same	same	ADJ
bracis-19080	111	10	location	location	NOUN
bracis-19080	111	11	in	in	ADP
bracis-19080	111	12	the	the	DET
bracis-19080	111	13	euclidean	euclidean	ADJ
bracis-19080	111	14	space	space	NOUN
bracis-19080	111	15	of	of	ADP
bracis-19080	111	16	the	the	DET
bracis-19080	111	17	original	original	ADJ
bracis-19080	111	18	sentence	sentence	NOUN
bracis-19080	111	19	.	.	PUNCT
bracis-19080	112	1	two	two	NUM
bracis-19080	112	2	other	other	ADJ
bracis-19080	112	3	aspects	aspect	NOUN
bracis-19080	112	4	of	of	ADP
bracis-19080	112	5	that	that	DET
bracis-19080	112	6	work	work	NOUN
bracis-19080	112	7	are	be	AUX
bracis-19080	112	8	further	far	ADV
bracis-19080	112	9	worth	worth	ADJ
bracis-19080	112	10	mentioning	mention	VERB
bracis-19080	112	11	.	.	PUNCT
bracis-19080	113	1	first	first	ADV
bracis-19080	113	2	,	,	PUNCT
bracis-19080	113	3	the	the	DET
bracis-19080	113	4	student	student	NOUN
bracis-19080	113	5	models	model	NOUN
bracis-19080	113	6	are	be	AUX
bracis-19080	113	7	trained	train	VERB
bracis-19080	113	8	with	with	ADP
bracis-19080	113	9	parallel	parallel	ADJ
bracis-19080	113	10	data	datum	NOUN
bracis-19080	113	11	for	for	ADP
bracis-19080	113	12	50	50	NUM
bracis-19080	113	13	languages	language	NOUN
bracis-19080	113	14	including	include	VERB
bracis-19080	113	15	portuguese	portuguese	PROPN
bracis-19080	113	16	.	.	PUNCT
bracis-19080	114	1	second	second	ADV
bracis-19080	114	2	,	,	PUNCT
bracis-19080	114	3	the	the	DET
bracis-19080	114	4	student	student	NOUN
bracis-19080	114	5	model	model	NOUN
bracis-19080	114	6	is	be	AUX
bracis-19080	114	7	learned	learn	VERB
bracis-19080	114	8	by	by	ADP
bracis-19080	114	9	a	a	DET
bracis-19080	114	10	xlm	xlm	PROPN
bracis-19080	114	11	-	-	PUNCT
bracis-19080	114	12	r	r	NOUN
bracis-19080	114	13	algorithm	algorithm	NOUN
bracis-19080	114	14	 	 	SPACE
bracis-19080	115	1	[	[	X
bracis-19080	115	2	6	6	NUM
bracis-19080	115	3	]	]	PUNCT
bracis-19080	115	4	,	,	PUNCT
bracis-19080	115	5	a	a	DET
bracis-19080	115	6	cross	cross	ADJ
bracis-19080	115	7	-	-	ADJ
bracis-19080	115	8	lingual	lingual	ADJ
bracis-19080	115	9	version	version	NOUN
bracis-19080	115	10	of	of	ADP
bracis-19080	115	11	bert	bert	PROPN
bracis-19080	115	12	.	.	PUNCT
bracis-19080	116	1	in	in	ADP
bracis-19080	116	2	detail	detail	NOUN
bracis-19080	116	3	,	,	PUNCT
bracis-19080	116	4	authors	author	NOUN
bracis-19080	116	5	claim	claim	VERB
bracis-19080	116	6	xlm	xlm	NOUN
bracis-19080	116	7	-	-	PUNCT
bracis-19080	116	8	r	r	NOUN
bracis-19080	116	9	is	be	AUX
bracis-19080	116	10	better	well	ADV
bracis-19080	116	11	suited	suit	VERB
bracis-19080	116	12	for	for	ADP
bracis-19080	116	13	multilingual	multilingual	ADJ
bracis-19080	116	14	learning	learning	NOUN
bracis-19080	116	15	than	than	ADP
bracis-19080	116	16	bert	bert	PROPN
bracis-19080	116	17	given	give	VERB
bracis-19080	116	18	its	its	PRON
bracis-19080	116	19	ability	ability	NOUN
bracis-19080	116	20	to	to	PART
bracis-19080	116	21	handle	handle	VERB
bracis-19080	116	22	non	non	ADJ
bracis-19080	116	23	-	-	ADJ
bracis-19080	116	24	latin	latin	ADJ
bracis-19080	116	25	alphabets	alphabet	NOUN
bracis-19080	116	26	.	.	PUNCT
bracis-19080	117	1	as	as	SCONJ
bracis-19080	117	2	reviewed	review	VERB
bracis-19080	117	3	in	in	ADP
bracis-19080	117	4	this	this	DET
bracis-19080	117	5	section	section	NOUN
bracis-19080	117	6	,	,	PUNCT
bracis-19080	117	7	the	the	DET
bracis-19080	117	8	literature	literature	NOUN
bracis-19080	117	9	on	on	ADP
bracis-19080	117	10	contextual	contextual	ADJ
bracis-19080	117	11	embeddings	embedding	NOUN
bracis-19080	117	12	has	have	AUX
bracis-19080	117	13	matured	mature	VERB
bracis-19080	117	14	over	over	ADP
bracis-19080	117	15	recent	recent	ADJ
bracis-19080	117	16	years	year	NOUN
bracis-19080	117	17	.	.	PUNCT
bracis-19080	118	1	still	still	ADV
bracis-19080	118	2	,	,	PUNCT
bracis-19080	118	3	models	model	NOUN
bracis-19080	118	4	are	be	AUX
bracis-19080	118	5	usually	usually	ADV
bracis-19080	118	6	developed	develop	VERB
bracis-19080	118	7	and	and	CCONJ
bracis-19080	118	8	made	make	VERB
bracis-19080	118	9	available	available	ADJ
bracis-19080	118	10	for	for	ADP
bracis-19080	118	11	nlp	nlp	ADJ
bracis-19080	118	12	applications	application	NOUN
bracis-19080	118	13	in	in	ADP
bracis-19080	118	14	english	english	PROPN
bracis-19080	118	15	.	.	PUNCT
bracis-19080	119	1	in	in	ADP
bracis-19080	119	2	the	the	DET
bracis-19080	119	3	next	next	ADJ
bracis-19080	119	4	section	section	NOUN
bracis-19080	119	5	,	,	PUNCT
bracis-19080	119	6	we	we	PRON
bracis-19080	119	7	discuss	discuss	VERB
bracis-19080	119	8	the	the	DET
bracis-19080	119	9	sts	sts	ADJ
bracis-19080	119	10	problem	problem	NOUN
bracis-19080	119	11	in	in	ADP
bracis-19080	119	12	the	the	DET
bracis-19080	119	13	context	context	NOUN
bracis-19080	119	14	of	of	ADP
bracis-19080	119	15	the	the	DET
bracis-19080	119	16	portuguese	portuguese	ADJ
bracis-19080	119	17	language	language	NOUN
bracis-19080	119	18	.	.	PUNCT
bracis-19080	120	1	3	3	NUM
bracis-19080	120	2	semantic	semantic	ADJ
bracis-19080	120	3	textual	textual	ADJ
bracis-19080	120	4	similarity	similarity	NOUN
bracis-19080	120	5	in	in	ADP
bracis-19080	120	6	 	 	SPACE
bracis-19080	120	7	portuguese	portuguese	NOUN
bracis-19080	120	8	as	as	SCONJ
bracis-19080	120	9	previously	previously	ADV
bracis-19080	120	10	discussed	discuss	VERB
bracis-19080	120	11	,	,	PUNCT
bracis-19080	120	12	semantic	semantic	ADJ
bracis-19080	120	13	textual	textual	ADJ
bracis-19080	120	14	similarity	similarity	NOUN
bracis-19080	120	15	 	 	SPACE
bracis-19080	120	16	(	(	PUNCT
bracis-19080	120	17	sts	st	NOUN
bracis-19080	120	18	)	)	PUNCT
bracis-19080	120	19	is	be	AUX
bracis-19080	120	20	a	a	DET
bracis-19080	120	21	relevant	relevant	ADJ
bracis-19080	120	22	nlp	nlp	NOUN
bracis-19080	120	23	application	application	NOUN
bracis-19080	120	24	with	with	ADP
bracis-19080	120	25	international	international	ADJ
bracis-19080	120	26	efforts	effort	NOUN
bracis-19080	120	27	to	to	PART
bracis-19080	120	28	foster	foster	VERB
bracis-19080	120	29	its	its	PRON
bracis-19080	120	30	research	research	NOUN
bracis-19080	120	31	.	.	PUNCT
bracis-19080	121	1	sts	st	NOUN
bracis-19080	121	2	can	can	AUX
bracis-19080	121	3	be	be	AUX
bracis-19080	121	4	defined	define	VERB
bracis-19080	121	5	as	as	ADP
bracis-19080	121	6	a	a	DET
bracis-19080	121	7	regression	regression	NOUN
bracis-19080	121	8	task	task	NOUN
bracis-19080	121	9	where	where	SCONJ
bracis-19080	121	10	one	one	PRON
bracis-19080	121	11	wants	want	VERB
bracis-19080	121	12	to	to	PART
bracis-19080	121	13	numerically	numerically	ADV
bracis-19080	121	14	measure	measure	VERB
bracis-19080	121	15	the	the	DET
bracis-19080	121	16	similarity	similarity	NOUN
bracis-19080	121	17	between	between	ADP
bracis-19080	121	18	two	two	NUM
bracis-19080	121	19	sentences	sentence	NOUN
bracis-19080	121	20	.	.	PUNCT
bracis-19080	122	1	higher	high	ADJ
bracis-19080	122	2	scores	score	NOUN
bracis-19080	122	3	indicate	indicate	VERB
bracis-19080	122	4	a	a	DET
bracis-19080	122	5	stronger	strong	ADJ
bracis-19080	122	6	similarity	similarity	NOUN
bracis-19080	122	7	between	between	ADP
bracis-19080	122	8	a	a	DET
bracis-19080	122	9	sentence	sentence	NOUN
bracis-19080	122	10	pair	pair	NOUN
bracis-19080	122	11	.	.	PUNCT
bracis-19080	123	1	among	among	ADP
bracis-19080	123	2	the	the	DET
bracis-19080	123	3	main	main	ADJ
bracis-19080	123	4	international	international	ADJ
bracis-19080	123	5	efforts	effort	NOUN
bracis-19080	123	6	concerning	concern	VERB
bracis-19080	123	7	sts	st	NOUN
bracis-19080	123	8	,	,	PUNCT
bracis-19080	123	9	the	the	DET
bracis-19080	123	10	international	international	ADJ
bracis-19080	123	11	workshop	workshop	NOUN
bracis-19080	123	12	on	on	ADP
bracis-19080	123	13	semantic	semantic	ADJ
bracis-19080	123	14	evaluation	evaluation	NOUN
bracis-19080	123	15	 	 	SPACE
bracis-19080	123	16	(	(	PUNCT
bracis-19080	123	17	semeval	semeval	NOUN
bracis-19080	123	18	 	 	SPACE
bracis-19080	124	1	[	[	X
bracis-19080	124	2	1	1	NUM
bracis-19080	124	3	]	]	PUNCT
bracis-19080	124	4	)	)	PUNCT
bracis-19080	124	5	is	be	AUX
bracis-19080	124	6	a	a	DET
bracis-19080	124	7	series	series	NOUN
bracis-19080	124	8	of	of	ADP
bracis-19080	124	9	workshops	workshop	NOUN
bracis-19080	124	10	focused	focus	VERB
bracis-19080	124	11	on	on	ADP
bracis-19080	124	12	semantic	semantic	ADJ
bracis-19080	124	13	nlp	nlp	NOUN
bracis-19080	124	14	problems	problem	NOUN
bracis-19080	124	15	.	.	PUNCT
bracis-19080	125	1	each	each	DET
bracis-19080	125	2	year	year	NOUN
bracis-19080	125	3	,	,	PUNCT
bracis-19080	125	4	several	several	ADJ
bracis-19080	125	5	shared	share	VERB
bracis-19080	125	6	tasks	task	NOUN
bracis-19080	125	7	are	be	AUX
bracis-19080	125	8	hosted	host	VERB
bracis-19080	125	9	,	,	PUNCT
bracis-19080	125	10	in	in	ADP
bracis-19080	125	11	which	which	PRON
bracis-19080	125	12	high	high	ADJ
bracis-19080	125	13	-	-	PUNCT
bracis-19080	125	14	quality	quality	NOUN
bracis-19080	125	15	datasets	dataset	NOUN
bracis-19080	125	16	are	be	AUX
bracis-19080	125	17	used	use	VERB
bracis-19080	125	18	to	to	ADP
bracis-19080	125	19	benchmark	benchmark	NOUN
bracis-19080	125	20	submissions	submission	NOUN
bracis-19080	125	21	from	from	ADP
bracis-19080	125	22	participating	participate	VERB
bracis-19080	125	23	research	research	NOUN
bracis-19080	125	24	teams	team	NOUN
bracis-19080	125	25	.	.	PUNCT
bracis-19080	126	1	several	several	ADJ
bracis-19080	126	2	similarity	similarity	NOUN
bracis-19080	126	3	tasks	task	NOUN
bracis-19080	126	4	have	have	AUX
bracis-19080	126	5	been	be	AUX
bracis-19080	126	6	defined	define	VERB
bracis-19080	126	7	at	at	ADP
bracis-19080	126	8	semeval	semeval	NOUN
bracis-19080	126	9	,	,	PUNCT
bracis-19080	126	10	including	include	VERB
bracis-19080	126	11	cross	cross	NOUN
bracis-19080	126	12	-	-	ADJ
bracis-19080	126	13	lingual	lingual	ADJ
bracis-19080	126	14	and	and	CCONJ
bracis-19080	126	15	multilingual	multilingual	ADJ
bracis-19080	126	16	sts	st	NOUN
bracis-19080	126	17	 	 	SPACE
bracis-19080	126	18	[	[	X
bracis-19080	126	19	5	5	NUM
bracis-19080	126	20	]	]	PUNCT
bracis-19080	126	21	.	.	PUNCT
bracis-19080	127	1	the	the	DET
bracis-19080	127	2	first	first	ADJ
bracis-19080	127	3	dataset	dataset	NOUN
bracis-19080	127	4	that	that	PRON
bracis-19080	127	5	included	include	VERB
bracis-19080	127	6	semantic	semantic	ADJ
bracis-19080	127	7	textual	textual	ADJ
bracis-19080	127	8	similarity	similarity	NOUN
bracis-19080	127	9	between	between	ADP
bracis-19080	127	10	pairs	pair	NOUN
bracis-19080	127	11	of	of	ADP
bracis-19080	127	12	sentences	sentence	NOUN
bracis-19080	127	13	in	in	ADP
bracis-19080	127	14	portuguese	portuguese	PROPN
bracis-19080	127	15	was	be	AUX
bracis-19080	127	16	assin	assin	PROPN
bracis-19080	127	17	 	 	SPACE
bracis-19080	128	1	[	[	X
bracis-19080	128	2	9	9	NUM
bracis-19080	128	3	]	]	PUNCT
bracis-19080	128	4	.	.	PUNCT
bracis-19080	129	1	later	later	ADV
bracis-19080	129	2	,	,	PUNCT
bracis-19080	129	3	assin	assin	PROPN
bracis-19080	129	4	 	 	SPACE
bracis-19080	129	5	2	2	NUM
bracis-19080	129	6	 	 	SPACE
bracis-19080	129	7	[	[	X
bracis-19080	129	8	15	15	NUM
bracis-19080	129	9	]	]	PUNCT
bracis-19080	129	10	proposed	propose	VERB
bracis-19080	129	11	a	a	DET
bracis-19080	129	12	new	new	ADJ
bracis-19080	129	13	dataset	dataset	NOUN
bracis-19080	129	14	,	,	PUNCT
bracis-19080	129	15	based	base	VERB
bracis-19080	129	16	on	on	ADP
bracis-19080	129	17	sentences	sentence	NOUN
bracis-19080	129	18	from	from	ADP
bracis-19080	129	19	the	the	DET
bracis-19080	129	20	sick	sick	ADJ
bracis-19080	129	21	-	-	PUNCT
bracis-19080	129	22	br	br	NOUN
bracis-19080	129	23	dataset	dataset	NOUN
bracis-19080	129	24	 	 	SPACE
bracis-19080	130	1	[	[	X
bracis-19080	130	2	16	16	NUM
bracis-19080	130	3	]	]	PUNCT
bracis-19080	130	4	.	.	PUNCT
bracis-19080	131	1	next	next	ADV
bracis-19080	131	2	,	,	PUNCT
bracis-19080	131	3	we	we	PRON
bracis-19080	131	4	describe	describe	VERB
bracis-19080	131	5	each	each	DET
bracis-19080	131	6	shared	share	VERB
bracis-19080	131	7	task	task	NOUN
bracis-19080	131	8	,	,	PUNCT
bracis-19080	131	9	its	its	PRON
bracis-19080	131	10	associated	associated	ADJ
bracis-19080	131	11	datasets	dataset	NOUN
bracis-19080	131	12	,	,	PUNCT
bracis-19080	131	13	and	and	CCONJ
bracis-19080	131	14	the	the	DET
bracis-19080	131	15	best	good	ADJ
bracis-19080	131	16	results	result	NOUN
bracis-19080	131	17	observed	observe	VERB
bracis-19080	131	18	for	for	ADP
bracis-19080	131	19	each	each	PRON
bracis-19080	131	20	.	.	PUNCT
bracis-19080	132	1	assin	assin	PROPN
bracis-19080	132	2	stands	stand	VERB
bracis-19080	132	3	for	for	ADP
bracis-19080	132	4	evaluation	evaluation	NOUN
bracis-19080	132	5	of	of	ADP
bracis-19080	132	6	semantic	semantic	ADJ
bracis-19080	132	7	textual	textual	ADJ
bracis-19080	132	8	similarity	similarity	NOUN
bracis-19080	132	9	and	and	CCONJ
bracis-19080	132	10	natural	natural	ADJ
bracis-19080	132	11	language	language	NOUN
bracis-19080	132	12	inference	inference	NOUN
bracis-19080	132	13	 	 	SPACE
bracis-19080	133	1	[	[	X
bracis-19080	133	2	9	9	NUM
bracis-19080	133	3	]	]	PUNCT
bracis-19080	133	4	,	,	PUNCT
bracis-19080	133	5	a	a	DET
bracis-19080	133	6	shared	share	VERB
bracis-19080	133	7	task	task	NOUN
bracis-19080	133	8	organized	organize	VERB
bracis-19080	133	9	at	at	ADP
bracis-19080	133	10	the	the	DET
bracis-19080	133	11	propor	propor	ADJ
bracis-19080	133	12	2016	2016	NUM
bracis-19080	133	13	conference	conference	NOUN
bracis-19080	133	14	.	.	PUNCT
bracis-19080	134	1	the	the	DET
bracis-19080	134	2	assin	assin	PROPN
bracis-19080	134	3	datasets	dataset	NOUN
bracis-19080	134	4	comprise	comprise	VERB
bracis-19080	134	5	(	(	PUNCT
bracis-19080	134	6	i	i	NOUN
bracis-19080	134	7	)	)	PUNCT
bracis-19080	134	8	 	 	SPACE
bracis-19080	134	9	pt	pt	PROPN
bracis-19080	134	10	-	-	PUNCT
bracis-19080	134	11	pt	pt	NOUN
bracis-19080	134	12	,	,	PUNCT
bracis-19080	134	13	containing	contain	VERB
bracis-19080	134	14	sentences	sentence	NOUN
bracis-19080	134	15	in	in	ADP
bracis-19080	134	16	european	european	ADJ
bracis-19080	134	17	portuguese	portuguese	PROPN
bracis-19080	134	18	,	,	PUNCT
bracis-19080	134	19	and	and	CCONJ
bracis-19080	134	20	(	(	PUNCT
bracis-19080	134	21	ii	ii	NOUN
bracis-19080	134	22	)	)	PUNCT
bracis-19080	134	23	 	 	SPACE
bracis-19080	134	24	pt	pt	PROPN
bracis-19080	134	25	-	-	PUNCT
bracis-19080	134	26	br	br	NOUN
bracis-19080	134	27	,	,	PUNCT
bracis-19080	134	28	containing	contain	VERB
bracis-19080	134	29	sentences	sentence	NOUN
bracis-19080	134	30	in	in	ADP
bracis-19080	134	31	brazilian	brazilian	ADJ
bracis-19080	134	32	portuguese	portuguese	PROPN
bracis-19080	134	33	.	.	PUNCT
bracis-19080	135	1	each	each	DET
bracis-19080	135	2	dataset	dataset	NOUN
bracis-19080	135	3	(	(	PUNCT
bracis-19080	135	4	pt	pt	NOUN
bracis-19080	135	5	-	-	PUNCT
bracis-19080	135	6	pt	pt	NOUN
bracis-19080	135	7	and	and	CCONJ
bracis-19080	135	8	pt	pt	NOUN
bracis-19080	135	9	-	-	PUNCT
bracis-19080	135	10	br	br	NOUN
bracis-19080	135	11	)	)	PUNCT
bracis-19080	135	12	has	have	VERB
bracis-19080	135	13	5,000	5,000	NUM
bracis-19080	135	14	sentences	sentence	NOUN
bracis-19080	135	15	:	:	PUNCT
bracis-19080	135	16	2,500	2,500	NUM
bracis-19080	135	17	for	for	ADP
bracis-19080	135	18	training	training	NOUN
bracis-19080	135	19	,	,	PUNCT
bracis-19080	135	20	500	500	NUM
bracis-19080	135	21	for	for	ADP
bracis-19080	135	22	validation	validation	NOUN
bracis-19080	135	23	,	,	PUNCT
bracis-19080	135	24	and	and	CCONJ
bracis-19080	135	25	2,000	2,000	NUM
bracis-19080	135	26	for	for	ADP
bracis-19080	135	27	testing	testing	NOUN
bracis-19080	135	28	.	.	PUNCT
bracis-19080	136	1	sentences	sentence	NOUN
bracis-19080	136	2	in	in	ADP
bracis-19080	136	3	each	each	DET
bracis-19080	136	4	dataset	dataset	NOUN
bracis-19080	136	5	were	be	AUX
bracis-19080	136	6	obtained	obtain	VERB
bracis-19080	136	7	from	from	ADP
bracis-19080	136	8	google	google	PROPN
bracis-19080	136	9	news	news	PROPN
bracis-19080	136	10	,	,	PUNCT
bracis-19080	136	11	processed	process	VERB
bracis-19080	136	12	in	in	ADP
bracis-19080	136	13	three	three	NUM
bracis-19080	136	14	steps	step	NOUN
bracis-19080	136	15	.	.	PUNCT
bracis-19080	137	1	first	first	ADJ
bracis-19080	137	2	,	,	PUNCT
bracis-19080	137	3	vectorial	vectorial	ADJ
bracis-19080	137	4	space	space	NOUN
bracis-19080	137	5	models	model	NOUN
bracis-19080	137	6	 	 	SPACE
bracis-19080	138	1	[	[	X
bracis-19080	138	2	21	21	NUM
bracis-19080	138	3	]	]	PUNCT
bracis-19080	138	4	were	be	AUX
bracis-19080	138	5	used	use	VERB
bracis-19080	138	6	to	to	PART
bracis-19080	138	7	select	select	VERB
bracis-19080	138	8	similar	similar	ADJ
bracis-19080	138	9	sentences	sentence	NOUN
bracis-19080	138	10	from	from	ADP
bracis-19080	138	11	different	different	ADJ
bracis-19080	138	12	documents	document	NOUN
bracis-19080	138	13	.	.	PUNCT
bracis-19080	139	1	second	second	ADJ
bracis-19080	139	2	,	,	PUNCT
bracis-19080	139	3	these	these	DET
bracis-19080	139	4	sentences	sentence	NOUN
bracis-19080	139	5	were	be	AUX
bracis-19080	139	6	filtered	filter	VERB
bracis-19080	139	7	manually	manually	ADV
bracis-19080	139	8	,	,	PUNCT
bracis-19080	139	9	and	and	CCONJ
bracis-19080	139	10	noisy	noisy	ADJ
bracis-19080	139	11	pairs	pair	NOUN
bracis-19080	139	12	were	be	AUX
bracis-19080	139	13	removed	remove	VERB
bracis-19080	139	14	.	.	PUNCT
bracis-19080	140	1	finally	finally	ADV
bracis-19080	140	2	,	,	PUNCT
bracis-19080	140	3	filtered	filter	VERB
bracis-19080	140	4	sentence	sentence	NOUN
bracis-19080	140	5	pairs	pair	NOUN
bracis-19080	140	6	were	be	AUX
bracis-19080	140	7	annotated	annotate	VERB
bracis-19080	140	8	by	by	ADP
bracis-19080	140	9	four	four	NUM
bracis-19080	140	10	human	human	ADJ
bracis-19080	140	11	judges	judge	NOUN
bracis-19080	140	12	.	.	PUNCT
bracis-19080	141	1	scoring	scoring	NOUN
bracis-19080	141	2	rules	rule	NOUN
bracis-19080	141	3	are	be	AUX
bracis-19080	141	4	describe	describe	NOUN
bracis-19080	141	5	below	below	ADP
bracis-19080	141	6	:	:	PUNCT
bracis-19080	141	7	1	1	X
bracis-19080	141	8	.	.	X
bracis-19080	141	9	sentences	sentence	NOUN
bracis-19080	141	10	are	be	AUX
bracis-19080	141	11	completely	completely	ADV
bracis-19080	141	12	different	different	ADJ
bracis-19080	141	13	.	.	PUNCT
bracis-19080	142	1	they	they	PRON
bracis-19080	142	2	may	may	AUX
bracis-19080	142	3	talk	talk	VERB
bracis-19080	142	4	about	about	ADP
bracis-19080	142	5	the	the	DET
bracis-19080	142	6	same	same	ADJ
bracis-19080	142	7	fact	fact	NOUN
bracis-19080	142	8	,	,	PUNCT
bracis-19080	142	9	but	but	CCONJ
bracis-19080	142	10	this	this	PRON
bracis-19080	142	11	can	can	AUX
bracis-19080	142	12	not	not	PART
bracis-19080	142	13	be	be	AUX
bracis-19080	142	14	determined	determine	VERB
bracis-19080	142	15	without	without	ADP
bracis-19080	142	16	context	context	NOUN
bracis-19080	142	17	.	.	PUNCT
bracis-19080	143	1	2	2	X
bracis-19080	143	2	.	.	X
bracis-19080	143	3	sentences	sentence	NOUN
bracis-19080	143	4	refer	refer	VERB
bracis-19080	143	5	to	to	ADP
bracis-19080	143	6	different	different	ADJ
bracis-19080	143	7	facts	fact	NOUN
bracis-19080	143	8	and	and	CCONJ
bracis-19080	143	9	are	be	AUX
bracis-19080	143	10	not	not	PART
bracis-19080	143	11	similar	similar	ADJ
bracis-19080	143	12	,	,	PUNCT
bracis-19080	143	13	but	but	CCONJ
bracis-19080	143	14	are	be	AUX
bracis-19080	143	15	about	about	ADV
bracis-19080	143	16	the	the	DET
bracis-19080	143	17	same	same	ADJ
bracis-19080	143	18	subject	subject	NOUN
bracis-19080	143	19	 	 	SPACE
bracis-19080	143	20	(	(	PUNCT
bracis-19080	143	21	e.g.	e.g.	ADV
bracis-19080	143	22	a	a	DET
bracis-19080	143	23	soccer	soccer	NOUN
bracis-19080	143	24	match	match	NOUN
bracis-19080	143	25	or	or	CCONJ
bracis-19080	143	26	elections	election	NOUN
bracis-19080	143	27	)	)	PUNCT
bracis-19080	143	28	.	.	PUNCT
bracis-19080	144	1	3	3	X
bracis-19080	144	2	.	.	X
bracis-19080	145	1	the	the	DET
bracis-19080	145	2	sentences	sentence	NOUN
bracis-19080	145	3	present	present	VERB
bracis-19080	145	4	some	some	DET
bracis-19080	145	5	similarity	similarity	NOUN
bracis-19080	145	6	;	;	PUNCT
bracis-19080	145	7	can	can	AUX
bracis-19080	145	8	refer	refer	VERB
bracis-19080	145	9	to	to	ADP
bracis-19080	145	10	the	the	DET
bracis-19080	145	11	same	same	ADJ
bracis-19080	145	12	fact	fact	NOUN
bracis-19080	145	13	or	or	CCONJ
bracis-19080	145	14	not	not	PART
bracis-19080	145	15	.	.	PUNCT
bracis-19080	146	1	4	4	X
bracis-19080	146	2	.	.	X
bracis-19080	146	3	sentence	sentence	NOUN
bracis-19080	146	4	content	content	NOUN
bracis-19080	146	5	is	be	AUX
bracis-19080	146	6	very	very	ADV
bracis-19080	146	7	similar	similar	ADJ
bracis-19080	146	8	,	,	PUNCT
bracis-19080	146	9	but	but	CCONJ
bracis-19080	146	10	one	one	NUM
bracis-19080	146	11	 	 	SPACE
bracis-19080	146	12	(	(	PUNCT
bracis-19080	146	13	or	or	CCONJ
bracis-19080	146	14	both	both	PRON
bracis-19080	146	15	)	)	PUNCT
bracis-19080	146	16	has	have	VERB
bracis-19080	146	17	some	some	DET
bracis-19080	146	18	exclusive	exclusive	ADJ
bracis-19080	146	19	information	information	NOUN
bracis-19080	146	20	 	 	SPACE
bracis-19080	146	21	(	(	PUNCT
bracis-19080	146	22	e.g.	e.g.	ADV
bracis-19080	146	23	a	a	DET
bracis-19080	146	24	date	date	NOUN
bracis-19080	146	25	or	or	CCONJ
bracis-19080	146	26	locality	locality	NOUN
bracis-19080	146	27	)	)	PUNCT
bracis-19080	146	28	.	.	PUNCT
bracis-19080	147	1	5	5	X
bracis-19080	147	2	.	.	X
bracis-19080	147	3	sentences	sentence	NOUN
bracis-19080	147	4	have	have	VERB
bracis-19080	147	5	practically	practically	ADV
bracis-19080	147	6	the	the	DET
bracis-19080	147	7	same	same	ADJ
bracis-19080	147	8	meaning	meaning	NOUN
bracis-19080	147	9	,	,	PUNCT
bracis-19080	147	10	possibly	possibly	ADV
bracis-19080	147	11	with	with	ADP
bracis-19080	147	12	a	a	DET
bracis-19080	147	13	minimal	minimal	ADJ
bracis-19080	147	14	difference	difference	NOUN
bracis-19080	147	15	 	 	SPACE
bracis-19080	147	16	(	(	PUNCT
bracis-19080	147	17	e.g.	e.g.	ADV
bracis-19080	147	18	an	an	DET
bracis-19080	147	19	adjective	adjective	NOUN
bracis-19080	147	20	that	that	PRON
bracis-19080	147	21	does	do	AUX
bracis-19080	147	22	not	not	PART
bracis-19080	147	23	change	change	VERB
bracis-19080	147	24	interpretation	interpretation	NOUN
bracis-19080	147	25	)	)	PUNCT
bracis-19080	147	26	.	.	PUNCT
bracis-19080	148	1	assin	assin	PROPN
bracis-19080	148	2	 	 	SPACE
bracis-19080	148	3	2	2	NUM
bracis-19080	148	4	was	be	AUX
bracis-19080	148	5	the	the	DET
bracis-19080	148	6	second	second	ADJ
bracis-19080	148	7	edition	edition	NOUN
bracis-19080	148	8	of	of	ADP
bracis-19080	148	9	assin	assin	PROPN
bracis-19080	148	10	 	 	SPACE
bracis-19080	149	1	[	[	X
bracis-19080	149	2	15	15	NUM
bracis-19080	149	3	]	]	PUNCT
bracis-19080	149	4	,	,	PUNCT
bracis-19080	149	5	held	hold	VERB
bracis-19080	149	6	at	at	ADP
bracis-19080	149	7	the	the	DET
bracis-19080	149	8	stil	stil	ADJ
bracis-19080	149	9	2019	2019	NUM
bracis-19080	149	10	symposium	symposium	NOUN
bracis-19080	149	11	in	in	ADP
bracis-19080	149	12	brazil	brazil	PROPN
bracis-19080	149	13	.	.	PUNCT
bracis-19080	150	1	the	the	DET
bracis-19080	150	2	dataset	dataset	NOUN
bracis-19080	150	3	is	be	AUX
bracis-19080	150	4	a	a	DET
bracis-19080	150	5	translation	translation	NOUN
bracis-19080	150	6	and	and	CCONJ
bracis-19080	150	7	manual	manual	ADJ
bracis-19080	150	8	adaptation	adaptation	NOUN
bracis-19080	150	9	of	of	ADP
bracis-19080	150	10	the	the	DET
bracis-19080	150	11	sick	sick	ADJ
bracis-19080	150	12	dataset	dataset	NOUN
bracis-19080	150	13	 	 	SPACE
bracis-19080	151	1	[	[	X
bracis-19080	151	2	12	12	NUM
bracis-19080	151	3	]	]	PUNCT
bracis-19080	151	4	,	,	PUNCT
bracis-19080	151	5	the	the	DET
bracis-19080	151	6	sick	sick	ADJ
bracis-19080	151	7	-	-	PUNCT
bracis-19080	151	8	br	br	NOUN
bracis-19080	151	9	dataset	dataset	NOUN
bracis-19080	151	10	 	 	SPACE
bracis-19080	152	1	[	[	X
bracis-19080	152	2	16	16	NUM
bracis-19080	152	3	]	]	PUNCT
bracis-19080	152	4	.	.	PUNCT
bracis-19080	153	1	in	in	ADP
bracis-19080	153	2	addition	addition	NOUN
bracis-19080	153	3	,	,	PUNCT
bracis-19080	153	4	assin	assin	PROPN
bracis-19080	153	5	 	 	SPACE
bracis-19080	153	6	2	2	NUM
bracis-19080	153	7	also	also	ADV
bracis-19080	153	8	includes	include	VERB
bracis-19080	153	9	manually	manually	ADV
bracis-19080	153	10	-	-	PUNCT
bracis-19080	153	11	generated	generate	VERB
bracis-19080	153	12	sentence	sentence	NOUN
bracis-19080	153	13	pairs	pair	NOUN
bracis-19080	153	14	,	,	PUNCT
bracis-19080	153	15	which	which	PRON
bracis-19080	153	16	were	be	AUX
bracis-19080	153	17	reviewed	review	VERB
bracis-19080	153	18	by	by	ADP
bracis-19080	153	19	human	human	ADJ
bracis-19080	153	20	judges	judge	NOUN
bracis-19080	153	21	.	.	PUNCT
bracis-19080	154	1	each	each	DET
bracis-19080	154	2	pair	pair	NOUN
bracis-19080	154	3	of	of	ADP
bracis-19080	154	4	sentences	sentence	NOUN
bracis-19080	154	5	in	in	ADP
bracis-19080	154	6	assin	assin	PROPN
bracis-19080	154	7	 	 	SPACE
bracis-19080	154	8	2	2	NUM
bracis-19080	154	9	was	be	AUX
bracis-19080	154	10	annotated	annotate	VERB
bracis-19080	154	11	by	by	ADP
bracis-19080	154	12	at	at	ADV
bracis-19080	154	13	least	least	ADV
bracis-19080	154	14	four	four	NUM
bracis-19080	154	15	native	native	ADJ
bracis-19080	154	16	brazilian	brazilian	ADJ
bracis-19080	154	17	portuguese	portuguese	ADJ
bracis-19080	154	18	speakers	speaker	NOUN
bracis-19080	154	19	with	with	ADP
bracis-19080	154	20	linguistic	linguistic	ADJ
bracis-19080	154	21	training	training	NOUN
bracis-19080	154	22	background	background	NOUN
bracis-19080	154	23	.	.	PUNCT
bracis-19080	155	1	for	for	ADP
bracis-19080	155	2	sts	st	NOUN
bracis-19080	155	3	,	,	PUNCT
bracis-19080	155	4	the	the	DET
bracis-19080	155	5	final	final	ADJ
bracis-19080	155	6	score	score	NOUN
bracis-19080	155	7	was	be	AUX
bracis-19080	155	8	the	the	DET
bracis-19080	155	9	average	average	NOUN
bracis-19080	155	10	of	of	ADP
bracis-19080	155	11	the	the	DET
bracis-19080	155	12	scores	score	NOUN
bracis-19080	155	13	of	of	ADP
bracis-19080	155	14	each	each	DET
bracis-19080	155	15	annotator	annotator	NOUN
bracis-19080	155	16	.	.	PUNCT
bracis-19080	156	1	the	the	DET
bracis-19080	156	2	final	final	ADJ
bracis-19080	156	3	dataset	dataset	NOUN
bracis-19080	156	4	has	have	VERB
bracis-19080	156	5	10000	10000	NUM
bracis-19080	156	6	pairs	pair	NOUN
bracis-19080	156	7	of	of	ADP
bracis-19080	156	8	sentences	sentence	NOUN
bracis-19080	156	9	:	:	PUNCT
bracis-19080	156	10	6500	6500	NUM
bracis-19080	156	11	used	use	VERB
bracis-19080	156	12	for	for	ADP
bracis-19080	156	13	training	training	NOUN
bracis-19080	156	14	,	,	PUNCT
bracis-19080	156	15	500	500	NUM
bracis-19080	156	16	for	for	ADP
bracis-19080	156	17	validation	validation	NOUN
bracis-19080	156	18	,	,	PUNCT
bracis-19080	156	19	and	and	CCONJ
bracis-19080	156	20	3000	3000	NUM
bracis-19080	156	21	for	for	ADP
bracis-19080	156	22	testing	testing	NOUN
bracis-19080	156	23	.	.	PUNCT
bracis-19080	157	1	baseline	baseline	PROPN
bracis-19080	157	2	results	result	NOUN
bracis-19080	157	3	for	for	ADP
bracis-19080	157	4	the	the	DET
bracis-19080	157	5	assin	assin	PROPN
bracis-19080	157	6	and	and	CCONJ
bracis-19080	157	7	assin	assin	PROPN
bracis-19080	157	8	 	 	SPACE
bracis-19080	157	9	2	2	NUM
bracis-19080	157	10	datasets	dataset	NOUN
bracis-19080	157	11	are	be	AUX
bracis-19080	157	12	listed	list	VERB
bracis-19080	157	13	at	at	ADP
bracis-19080	157	14	the	the	DET
bracis-19080	157	15	top	top	NOUN
bracis-19080	157	16	of	of	ADP
bracis-19080	157	17	table	table	NOUN
bracis-19080	157	18	 	 	SPACE
bracis-19080	157	19	1	1	NUM
bracis-19080	157	20	.	.	PUNCT
bracis-19080	157	21	pearson	pearson	PROPN
bracis-19080	157	22	correlation	correlation	PROPN
bracis-19080	157	23	coefficient	coefficient	NOUN
bracis-19080	157	24	 	 	SPACE
bracis-19080	157	25	(	(	PUNCT
bracis-19080	157	26	to	to	PART
bracis-19080	157	27	be	be	AUX
bracis-19080	157	28	maximized	maximize	VERB
bracis-19080	157	29	)	)	PUNCT
bracis-19080	157	30	is	be	AUX
bracis-19080	157	31	considered	consider	VERB
bracis-19080	157	32	as	as	ADP
bracis-19080	157	33	the	the	DET
bracis-19080	157	34	primary	primary	ADJ
bracis-19080	157	35	metric	metric	NOUN
bracis-19080	157	36	for	for	ADP
bracis-19080	157	37	the	the	DET
bracis-19080	157	38	comparison	comparison	NOUN
bracis-19080	157	39	,	,	PUNCT
bracis-19080	157	40	complemented	complement	VERB
bracis-19080	157	41	by	by	ADP
bracis-19080	157	42	the	the	DET
bracis-19080	157	43	mean	mean	ADJ
bracis-19080	157	44	squared	square	VERB
bracis-19080	157	45	error	error	NOUN
bracis-19080	157	46	 	 	SPACE
bracis-19080	157	47	(	(	PUNCT
bracis-19080	157	48	mse	mse	PROPN
bracis-19080	157	49	,	,	PUNCT
bracis-19080	157	50	to	to	PART
bracis-19080	157	51	be	be	AUX
bracis-19080	157	52	minimized	minimize	VERB
bracis-19080	157	53	)	)	PUNCT
bracis-19080	157	54	.	.	PUNCT
bracis-19080	158	1	the	the	DET
bracis-19080	158	2	three	three	NUM
bracis-19080	158	3	topmost	topmost	ADJ
bracis-19080	158	4	works	work	NOUN
bracis-19080	158	5	are	be	AUX
bracis-19080	158	6	the	the	DET
bracis-19080	158	7	best	good	ADJ
bracis-19080	158	8	results	result	NOUN
bracis-19080	158	9	obtained	obtain	VERB
bracis-19080	158	10	during	during	ADP
bracis-19080	158	11	the	the	DET
bracis-19080	158	12	shared	share	VERB
bracis-19080	158	13	tasks	task	NOUN
bracis-19080	158	14	,	,	PUNCT
bracis-19080	158	15	whereas	whereas	SCONJ
bracis-19080	158	16	the	the	DET
bracis-19080	158	17	following	follow	VERB
bracis-19080	158	18	two	two	NUM
bracis-19080	158	19	models	model	NOUN
bracis-19080	158	20	were	be	AUX
bracis-19080	158	21	proposed	propose	VERB
bracis-19080	158	22	in	in	ADP
bracis-19080	158	23	a	a	DET
bracis-19080	158	24	work	work	NOUN
bracis-19080	158	25	published	publish	VERB
bracis-19080	158	26	after	after	ADP
bracis-19080	158	27	assin	assin	PROPN
bracis-19080	158	28	 	 	SPACE
bracis-19080	158	29	2	2	NUM
bracis-19080	158	30	 	 	SPACE
bracis-19080	159	1	[	[	X
bracis-19080	159	2	8].footnote	8].footnote	NUM
bracis-19080	159	3	1	1	NUM
bracis-19080	159	4	for	for	ADP
bracis-19080	159	5	brevity	brevity	NOUN
bracis-19080	159	6	,	,	PUNCT
bracis-19080	159	7	the	the	DET
bracis-19080	159	8	models	model	NOUN
bracis-19080	159	9	we	we	PRON
bracis-19080	159	10	assess	assess	VERB
bracis-19080	159	11	in	in	ADP
bracis-19080	159	12	this	this	DET
bracis-19080	159	13	work	work	NOUN
bracis-19080	159	14	are	be	AUX
bracis-19080	159	15	also	also	ADV
bracis-19080	159	16	given	give	VERB
bracis-19080	159	17	in	in	ADP
bracis-19080	159	18	table	table	NOUN
bracis-19080	159	19	 	 	SPACE
bracis-19080	159	20	1	1	NUM
bracis-19080	159	21	,	,	PUNCT
bracis-19080	159	22	and	and	CCONJ
bracis-19080	159	23	are	be	AUX
bracis-19080	159	24	described	describe	VERB
bracis-19080	159	25	in	in	ADP
bracis-19080	159	26	the	the	DET
bracis-19080	159	27	next	next	ADJ
bracis-19080	159	28	section	section	NOUN
bracis-19080	159	29	.	.	PUNCT
bracis-19080	160	1	baseline	baseline	PROPN
bracis-19080	160	2	results	result	NOUN
bracis-19080	160	3	.	.	PUNCT
bracis-19080	161	1	the	the	DET
bracis-19080	161	2	work	work	NOUN
bracis-19080	161	3	of	of	ADP
bracis-19080	161	4	fialho	fialho	NOUN
bracis-19080	161	5	,	,	PUNCT
bracis-19080	161	6	coheur	coheur	ADV
bracis-19080	161	7	,	,	PUNCT
bracis-19080	161	8	and	and	CCONJ
bracis-19080	161	9	quaresma	quaresma	VERB
bracis-19080	161	10	 	 	SPACE
bracis-19080	162	1	[	[	X
bracis-19080	162	2	8	8	NUM
bracis-19080	162	3	]	]	PUNCT
bracis-19080	162	4	considerably	considerably	ADV
bracis-19080	162	5	advanced	advance	VERB
bracis-19080	162	6	the	the	DET
bracis-19080	162	7	state	state	NOUN
bracis-19080	162	8	-	-	PUNCT
bracis-19080	162	9	of	of	ADP
bracis-19080	162	10	-	-	PUNCT
bracis-19080	162	11	the	the	DET
bracis-19080	162	12	-	-	PUNCT
bracis-19080	162	13	art	art	NOUN
bracis-19080	162	14	for	for	ADP
bracis-19080	162	15	both	both	DET
bracis-19080	162	16	assin	assin	PROPN
bracis-19080	162	17	pt	pt	PROPN
bracis-19080	162	18	-	-	PUNCT
bracis-19080	162	19	pt	pt	PROPN
bracis-19080	162	20	and	and	CCONJ
bracis-19080	162	21	assin	assin	PROPN
bracis-19080	162	22	pt	pt	PROPN
bracis-19080	162	23	-	-	PUNCT
bracis-19080	162	24	br	br	NOUN
bracis-19080	162	25	,	,	PUNCT
bracis-19080	162	26	and	and	CCONJ
bracis-19080	162	27	to	to	ADP
bracis-19080	162	28	a	a	DET
bracis-19080	162	29	lesser	less	ADJ
bracis-19080	162	30	extent	extent	NOUN
bracis-19080	162	31	also	also	ADV
bracis-19080	162	32	for	for	ADP
bracis-19080	162	33	assin	assin	PROPN
bracis-19080	162	34	 	 	SPACE
bracis-19080	162	35	2	2	NUM
bracis-19080	162	36	.	.	PUNCT
bracis-19080	162	37	specifically	specifically	ADV
bracis-19080	162	38	,	,	PUNCT
bracis-19080	162	39	the	the	DET
bracis-19080	162	40	authors	author	NOUN
bracis-19080	162	41	experimentally	experimentally	ADV
bracis-19080	162	42	assessed	assess	VERB
bracis-19080	162	43	bert	bert	NOUN
bracis-19080	162	44	models	model	NOUN
bracis-19080	162	45	for	for	ADP
bracis-19080	162	46	sts	st	NOUN
bracis-19080	162	47	and	and	CCONJ
bracis-19080	162	48	nli	nli	NOUN
bracis-19080	162	49	in	in	ADP
bracis-19080	162	50	portuguese	portuguese	PROPN
bracis-19080	162	51	using	use	VERB
bracis-19080	162	52	the	the	DET
bracis-19080	162	53	datasets	dataset	NOUN
bracis-19080	162	54	assin	assin	PROPN
bracis-19080	162	55	,	,	PUNCT
bracis-19080	162	56	assin	assin	PROPN
bracis-19080	162	57	 	 	SPACE
bracis-19080	162	58	2	2	NUM
bracis-19080	162	59	,	,	PUNCT
bracis-19080	162	60	and	and	CCONJ
bracis-19080	162	61	sick	sick	ADJ
bracis-19080	162	62	-	-	PUNCT
bracis-19080	162	63	br	br	NOUN
bracis-19080	162	64	.	.	PUNCT
bracis-19080	163	1	for	for	ADP
bracis-19080	163	2	sts	st	NOUN
bracis-19080	163	3	,	,	PUNCT
bracis-19080	163	4	two	two	NUM
bracis-19080	163	5	approaches	approach	NOUN
bracis-19080	163	6	were	be	AUX
bracis-19080	163	7	considered	consider	VERB
bracis-19080	163	8	:	:	PUNCT
bracis-19080	163	9	bert	bert	PROPN
bracis-19080	163	10	as	as	ADP
bracis-19080	163	11	a	a	DET
bracis-19080	163	12	regressor	regressor	NOUN
bracis-19080	163	13	or	or	CCONJ
bracis-19080	163	14	bert	bert	NOUN
bracis-19080	163	15	embeddings	embedding	NOUN
bracis-19080	163	16	as	as	ADP
bracis-19080	163	17	input	input	NOUN
bracis-19080	163	18	for	for	ADP
bracis-19080	163	19	regression	regression	NOUN
bracis-19080	163	20	algorithms	algorithm	NOUN
bracis-19080	163	21	.	.	PUNCT
bracis-19080	164	1	three	three	NUM
bracis-19080	164	2	models	model	NOUN
bracis-19080	164	3	fine	fine	ADV
bracis-19080	164	4	-	-	PUNCT
bracis-19080	164	5	tuned	tune	VERB
bracis-19080	164	6	for	for	ADP
bracis-19080	164	7	sts	st	NOUN
bracis-19080	164	8	were	be	AUX
bracis-19080	164	9	evaluated	evaluate	VERB
bracis-19080	164	10	,	,	PUNCT
bracis-19080	164	11	namely	namely	ADV
bracis-19080	164	12	(	(	PUNCT
bracis-19080	164	13	i	i	NOUN
bracis-19080	164	14	)	)	PUNCT
bracis-19080	164	15	 	 	SPACE
bracis-19080	164	16	one	one	NUM
bracis-19080	164	17	multilingual	multilingual	ADJ
bracis-19080	164	18	and	and	CCONJ
bracis-19080	164	19	(	(	PUNCT
bracis-19080	164	20	ii	ii	NOUN
bracis-19080	164	21	)	)	PUNCT
bracis-19080	164	22	 	 	SPACE
bracis-19080	164	23	two	two	NUM
bracis-19080	164	24	monolingual	monolingual	ADJ
bracis-19080	164	25	versions	version	NOUN
bracis-19080	164	26	pre	pre	ADJ
bracis-19080	164	27	-	-	VERB
bracis-19080	164	28	trained	train	VERB
bracis-19080	164	29	in	in	ADP
bracis-19080	164	30	brazilian	brazilian	ADJ
bracis-19080	164	31	portuguese	portuguese	PROPN
bracis-19080	164	32	differing	differ	VERB
bracis-19080	164	33	in	in	ADP
bracis-19080	164	34	size	size	NOUN
bracis-19080	164	35	 	 	SPACE
bracis-19080	164	36	(	(	PUNCT
bracis-19080	164	37	ptbert	ptbert	NOUN
bracis-19080	164	38	-	-	PUNCT
bracis-19080	164	39	base\(_\textit{ft}\	base\(_\textit{ft}\	NOUN
bracis-19080	164	40	)	)	PUNCT
bracis-19080	164	41	and	and	CCONJ
bracis-19080	164	42	ptbert	ptbert	NOUN
bracis-19080	164	43	-	-	PUNCT
bracis-19080	164	44	large\(_\textit{ft}\	large\(_\textit{ft}\	NOUN
bracis-19080	164	45	)	)	PUNCT
bracis-19080	164	46	)	)	PUNCT
bracis-19080	164	47	.	.	PUNCT
bracis-19080	165	1	ptbert	ptbert	NOUN
bracis-19080	165	2	-	-	PUNCT
bracis-19080	165	3	large\(_\textit{ft}\	large\(_\textit{ft}\	NOUN
bracis-19080	165	4	)	)	PUNCT
bracis-19080	165	5	used	use	VERB
bracis-19080	165	6	as	as	ADP
bracis-19080	165	7	a	a	DET
bracis-19080	165	8	regressor	regressor	NOUN
bracis-19080	165	9	achieved	achieve	VERB
bracis-19080	165	10	the	the	DET
bracis-19080	165	11	best	good	ADJ
bracis-19080	165	12	results	result	NOUN
bracis-19080	165	13	.	.	PUNCT
bracis-19080	166	1	table	table	NOUN
bracis-19080	166	2	1	1	NUM
bracis-19080	166	3	.	.	PUNCT
bracis-19080	167	1	results	result	NOUN
bracis-19080	167	2	using	use	VERB
bracis-19080	167	3	baseline	baseline	NOUN
bracis-19080	167	4	,	,	PUNCT
bracis-19080	167	5	pre	pre	ADJ
bracis-19080	167	6	-	-	VERB
bracis-19080	167	7	trained	trained	ADJ
bracis-19080	167	8	,	,	PUNCT
bracis-19080	167	9	and	and	CCONJ
bracis-19080	167	10	fine	fine	ADV
bracis-19080	167	11	-	-	PUNCT
bracis-19080	167	12	tuned	tune	VERB
bracis-19080	167	13	models.full	models.full	ADJ
bracis-19080	167	14	size	size	NOUN
bracis-19080	167	15	table	table	NOUN
bracis-19080	167	16	as	as	SCONJ
bracis-19080	167	17	discussed	discuss	VERB
bracis-19080	167	18	in	in	ADP
bracis-19080	167	19	this	this	DET
bracis-19080	167	20	section	section	NOUN
bracis-19080	167	21	,	,	PUNCT
bracis-19080	167	22	assin	assin	PROPN
bracis-19080	167	23	and	and	CCONJ
bracis-19080	167	24	assin	assin	PROPN
bracis-19080	167	25	 	 	SPACE
bracis-19080	167	26	2	2	NUM
bracis-19080	167	27	are	be	AUX
bracis-19080	167	28	relevant	relevant	ADJ
bracis-19080	167	29	portuguese	portuguese	NOUN
bracis-19080	167	30	sts	st	VERB
bracis-19080	167	31	datasets	dataset	NOUN
bracis-19080	167	32	for	for	ADP
bracis-19080	167	33	which	which	PRON
bracis-19080	167	34	bert	bert	NOUN
bracis-19080	167	35	models	model	NOUN
bracis-19080	167	36	are	be	AUX
bracis-19080	167	37	the	the	DET
bracis-19080	167	38	state	state	NOUN
bracis-19080	167	39	-	-	PUNCT
bracis-19080	167	40	of	of	ADP
bracis-19080	167	41	-	-	PUNCT
bracis-19080	167	42	the	the	DET
bracis-19080	167	43	-	-	PUNCT
bracis-19080	167	44	art	art	NOUN
bracis-19080	167	45	.	.	PUNCT
bracis-19080	168	1	yet	yet	ADV
bracis-19080	168	2	,	,	PUNCT
bracis-19080	168	3	the	the	DET
bracis-19080	168	4	computational	computational	ADJ
bracis-19080	168	5	overhead	overhead	NOUN
bracis-19080	168	6	incurred	incur	VERB
bracis-19080	168	7	for	for	ADP
bracis-19080	168	8	bert	bert	NOUN
bracis-19080	168	9	training	training	NOUN
bracis-19080	168	10	limits	limit	NOUN
bracis-19080	168	11	its	its	PRON
bracis-19080	168	12	applicability	applicability	NOUN
bracis-19080	168	13	to	to	ADP
bracis-19080	168	14	problems	problem	NOUN
bracis-19080	168	15	with	with	ADP
bracis-19080	168	16	a	a	DET
bracis-19080	168	17	moderate	moderate	ADJ
bracis-19080	168	18	size	size	NOUN
bracis-19080	168	19	sentence	sentence	NOUN
bracis-19080	168	20	collection	collection	NOUN
bracis-19080	168	21	.	.	PUNCT
bracis-19080	169	1	in	in	ADP
bracis-19080	169	2	the	the	DET
bracis-19080	169	3	next	next	ADJ
bracis-19080	169	4	section	section	NOUN
bracis-19080	169	5	,	,	PUNCT
bracis-19080	169	6	we	we	PRON
bracis-19080	169	7	discuss	discuss	VERB
bracis-19080	169	8	how	how	SCONJ
bracis-19080	169	9	we	we	PRON
bracis-19080	169	10	employ	employ	VERB
bracis-19080	169	11	and	and	CCONJ
bracis-19080	169	12	assess	assess	VERB
bracis-19080	169	13	sbert	sbert	NOUN
bracis-19080	169	14	models	model	NOUN
bracis-19080	169	15	on	on	ADP
bracis-19080	169	16	the	the	DET
bracis-19080	169	17	portuguese	portuguese	PROPN
bracis-19080	169	18	sts	st	VERB
bracis-19080	169	19	datasets	dataset	NOUN
bracis-19080	169	20	reviewed	review	VERB
bracis-19080	169	21	here	here	ADV
bracis-19080	169	22	.	.	PUNCT
bracis-19080	170	1	4	4	NUM
bracis-19080	170	2	experimental	experimental	ADJ
bracis-19080	170	3	assessment	assessment	NOUN
bracis-19080	170	4	sbert	sbert	NOUN
bracis-19080	170	5	models	model	NOUN
bracis-19080	170	6	have	have	AUX
bracis-19080	170	7	been	be	AUX
bracis-19080	170	8	successfully	successfully	ADV
bracis-19080	170	9	employed	employ	VERB
bracis-19080	170	10	on	on	ADP
bracis-19080	170	11	nlp	nlp	NOUN
bracis-19080	170	12	applications	application	NOUN
bracis-19080	170	13	where	where	SCONJ
bracis-19080	170	14	bert	bert	NOUN
bracis-19080	170	15	models	model	NOUN
bracis-19080	170	16	had	have	VERB
bracis-19080	170	17	state	state	NOUN
bracis-19080	170	18	-	-	PUNCT
bracis-19080	170	19	of	of	ADP
bracis-19080	170	20	-	-	PUNCT
bracis-19080	170	21	the	the	DET
bracis-19080	170	22	-	-	PUNCT
bracis-19080	170	23	art	art	NOUN
bracis-19080	170	24	performance	performance	NOUN
bracis-19080	170	25	 	 	SPACE
bracis-19080	170	26	[	[	X
bracis-19080	170	27	17	17	NUM
bracis-19080	170	28	]	]	PUNCT
bracis-19080	170	29	.	.	PUNCT
bracis-19080	171	1	furthermore	furthermore	ADV
bracis-19080	171	2	,	,	PUNCT
bracis-19080	171	3	multilingual	multilingual	ADJ
bracis-19080	171	4	pre	pre	ADJ
bracis-19080	171	5	-	-	ADJ
bracis-19080	171	6	trained	train	VERB
bracis-19080	171	7	sbert	sbert	NOUN
bracis-19080	171	8	models	model	NOUN
bracis-19080	171	9	have	have	AUX
bracis-19080	171	10	been	be	AUX
bracis-19080	171	11	made	make	VERB
bracis-19080	171	12	available	available	ADJ
bracis-19080	171	13	in	in	ADP
bracis-19080	171	14	an	an	DET
bracis-19080	171	15	open	open	ADJ
bracis-19080	171	16	source	source	NOUN
bracis-19080	171	17	sbert	sbert	NOUN
bracis-19080	171	18	repository	repository	NOUN
bracis-19080	171	19	 	 	SPACE
bracis-19080	171	20	(	(	PUNCT
bracis-19080	171	21	https://www.sbert.net	https://www.sbert.net	PROPN
bracis-19080	171	22	)	)	PUNCT
bracis-19080	171	23	.	.	PUNCT
bracis-19080	172	1	in	in	ADP
bracis-19080	172	2	this	this	DET
bracis-19080	172	3	section	section	NOUN
bracis-19080	172	4	,	,	PUNCT
bracis-19080	172	5	we	we	PRON
bracis-19080	172	6	describe	describe	VERB
bracis-19080	172	7	our	our	PRON
bracis-19080	172	8	experimental	experimental	ADJ
bracis-19080	172	9	analysis	analysis	NOUN
bracis-19080	172	10	in	in	ADP
bracis-19080	172	11	which	which	PRON
bracis-19080	172	12	we	we	PRON
bracis-19080	172	13	assess	assess	VERB
bracis-19080	172	14	the	the	DET
bracis-19080	172	15	performance	performance	NOUN
bracis-19080	172	16	of	of	ADP
bracis-19080	172	17	sbert	sbert	NOUN
bracis-19080	172	18	models	model	NOUN
bracis-19080	172	19	with	with	ADP
bracis-19080	172	20	and	and	CCONJ
bracis-19080	172	21	without	without	ADP
bracis-19080	172	22	fine	fine	ADV
bracis-19080	172	23	-	-	PUNCT
bracis-19080	172	24	tuning	tuning	NOUN
bracis-19080	172	25	for	for	ADP
bracis-19080	172	26	sts	st	NOUN
bracis-19080	172	27	in	in	ADP
bracis-19080	172	28	portuguese	portuguese	PROPN
bracis-19080	172	29	.	.	PUNCT
bracis-19080	173	1	initially	initially	ADV
bracis-19080	173	2	,	,	PUNCT
bracis-19080	173	3	we	we	PRON
bracis-19080	173	4	detail	detail	VERB
bracis-19080	173	5	the	the	DET
bracis-19080	173	6	models	model	NOUN
bracis-19080	173	7	we	we	PRON
bracis-19080	173	8	consider	consider	VERB
bracis-19080	173	9	,	,	PUNCT
bracis-19080	173	10	following	follow	VERB
bracis-19080	173	11	the	the	DET
bracis-19080	173	12	order	order	NOUN
bracis-19080	173	13	given	give	VERB
bracis-19080	173	14	in	in	ADP
bracis-19080	173	15	table	table	NOUN
bracis-19080	173	16	 	 	SPACE
bracis-19080	173	17	1.footnote	1.footnote	NUM
bracis-19080	173	18	2	2	NUM
bracis-19080	173	19	later	later	ADV
bracis-19080	173	20	,	,	PUNCT
bracis-19080	173	21	we	we	PRON
bracis-19080	173	22	discuss	discuss	VERB
bracis-19080	173	23	the	the	DET
bracis-19080	173	24	two	two	NUM
bracis-19080	173	25	stages	stage	NOUN
bracis-19080	173	26	of	of	ADP
bracis-19080	173	27	our	our	PRON
bracis-19080	173	28	experiments	experiment	NOUN
bracis-19080	173	29	,	,	PUNCT
bracis-19080	173	30	differing	differ	VERB
bracis-19080	173	31	as	as	ADP
bracis-19080	173	32	to	to	ADP
bracis-19080	173	33	whether	whether	SCONJ
bracis-19080	173	34	fine	fine	ADV
bracis-19080	173	35	-	-	PUNCT
bracis-19080	173	36	tuning	tuning	NOUN
bracis-19080	173	37	is	be	AUX
bracis-19080	173	38	employed	employ	VERB
bracis-19080	173	39	.	.	PUNCT
bracis-19080	174	1	4.1	4.1	NUM
bracis-19080	174	2	models	model	NOUN
bracis-19080	174	3	assessed	assess	VERB
bracis-19080	174	4	as	as	SCONJ
bracis-19080	174	5	listed	list	VERB
bracis-19080	174	6	in	in	ADP
bracis-19080	174	7	table	table	NOUN
bracis-19080	174	8	 	 	SPACE
bracis-19080	174	9	1	1	NUM
bracis-19080	174	10	,	,	PUNCT
bracis-19080	174	11	the	the	DET
bracis-19080	174	12	top	top	ADV
bracis-19080	174	13	-	-	PUNCT
bracis-19080	174	14	most	most	ADJ
bracis-19080	174	15	five	five	NUM
bracis-19080	174	16	models	model	NOUN
bracis-19080	174	17	we	we	PRON
bracis-19080	174	18	consider	consider	VERB
bracis-19080	174	19	as	as	ADP
bracis-19080	174	20	baseline	baseline	NOUN
bracis-19080	174	21	are	be	AUX
bracis-19080	174	22	(	(	PUNCT
bracis-19080	174	23	i	i	NOUN
bracis-19080	174	24	)	)	PUNCT
bracis-19080	174	25	 	 	SPACE
bracis-19080	174	26	the	the	DET
bracis-19080	174	27	best	well	ADV
bracis-19080	174	28	-	-	PUNCT
bracis-19080	174	29	performing	perform	VERB
bracis-19080	174	30	algorithms	algorithm	NOUN
bracis-19080	174	31	of	of	ADP
bracis-19080	174	32	the	the	DET
bracis-19080	174	33	assin	assin	PROPN
bracis-19080	174	34	and	and	CCONJ
bracis-19080	174	35	assin	assin	PROPN
bracis-19080	174	36	 	 	SPACE
bracis-19080	174	37	2	2	NUM
bracis-19080	174	38	shared	share	VERB
bracis-19080	174	39	tasks	task	NOUN
bracis-19080	174	40	,	,	PUNCT
bracis-19080	174	41	and	and	CCONJ
bracis-19080	174	42	(	(	PUNCT
bracis-19080	174	43	ii	ii	NOUN
bracis-19080	174	44	)	)	PUNCT
bracis-19080	174	45	 	 	SPACE
bracis-19080	174	46	the	the	DET
bracis-19080	174	47	state	state	NOUN
bracis-19080	174	48	-	-	PUNCT
bracis-19080	174	49	of	of	ADP
bracis-19080	174	50	-	-	PUNCT
bracis-19080	174	51	the	the	DET
bracis-19080	174	52	-	-	PUNCT
bracis-19080	174	53	art	art	NOUN
bracis-19080	174	54	bert	bert	NOUN
bracis-19080	174	55	models	model	NOUN
bracis-19080	174	56	 	 	SPACE
bracis-19080	175	1	[	[	X
bracis-19080	175	2	8	8	NUM
bracis-19080	175	3	]	]	PUNCT
bracis-19080	175	4	,	,	PUNCT
bracis-19080	175	5	already	already	ADV
bracis-19080	175	6	described	describe	VERB
bracis-19080	175	7	in	in	ADP
bracis-19080	175	8	the	the	DET
bracis-19080	175	9	previous	previous	ADJ
bracis-19080	175	10	section	section	NOUN
bracis-19080	175	11	.	.	PUNCT
bracis-19080	176	1	in	in	ADP
bracis-19080	176	2	addition	addition	NOUN
bracis-19080	176	3	,	,	PUNCT
bracis-19080	176	4	we	we	PRON
bracis-19080	176	5	also	also	ADV
bracis-19080	176	6	include	include	VERB
bracis-19080	176	7	as	as	SCONJ
bracis-19080	176	8	baseline	baseline	VERB
bracis-19080	176	9	two	two	NUM
bracis-19080	176	10	portuguese	portuguese	ADJ
bracis-19080	176	11	pre	pre	ADJ
bracis-19080	176	12	-	-	ADJ
bracis-19080	176	13	trained	train	VERB
bracis-19080	176	14	bert	bert	NOUN
bracis-19080	176	15	models	model	NOUN
bracis-19080	176	16	called	call	VERB
bracis-19080	176	17	 	 	SPACE
bracis-19080	176	18	bertimbau	bertimbau	NOUN
bracis-19080	176	19	 	 	SPACE
bracis-19080	176	20	[	[	X
bracis-19080	176	21	20	20	NUM
bracis-19080	176	22	]	]	PUNCT
bracis-19080	176	23	.	.	PUNCT
bracis-19080	177	1	remaining	remain	VERB
bracis-19080	177	2	models	model	NOUN
bracis-19080	177	3	given	give	VERB
bracis-19080	177	4	in	in	ADP
bracis-19080	177	5	table	table	NOUN
bracis-19080	177	6	 	 	SPACE
bracis-19080	177	7	1	1	NUM
bracis-19080	177	8	are	be	AUX
bracis-19080	177	9	used	use	VERB
bracis-19080	177	10	within	within	ADP
bracis-19080	177	11	the	the	DET
bracis-19080	177	12	sbert	sbert	NOUN
bracis-19080	177	13	architecture	architecture	NOUN
bracis-19080	177	14	.	.	PUNCT
bracis-19080	178	1	the	the	DET
bracis-19080	178	2	first	first	ADJ
bracis-19080	178	3	two	two	NUM
bracis-19080	178	4	reuse	reuse	NOUN
bracis-19080	178	5	bertimbau	bertimbau	NOUN
bracis-19080	178	6	weights	weight	NOUN
bracis-19080	178	7	,	,	PUNCT
bracis-19080	178	8	given	give	VERB
bracis-19080	178	9	the	the	DET
bracis-19080	178	10	compatibility	compatibility	NOUN
bracis-19080	178	11	between	between	ADP
bracis-19080	178	12	bert	bert	PROPN
bracis-19080	178	13	and	and	CCONJ
bracis-19080	178	14	sbert	sbert	NOUN
bracis-19080	178	15	.	.	PUNCT
bracis-19080	179	1	the	the	DET
bracis-19080	179	2	remaining	remain	VERB
bracis-19080	179	3	three	three	NUM
bracis-19080	179	4	models	model	NOUN
bracis-19080	179	5	are	be	AUX
bracis-19080	179	6	multilingual	multilingual	ADJ
bracis-19080	179	7	pre	pre	ADJ
bracis-19080	179	8	-	-	ADJ
bracis-19080	179	9	trained	train	VERB
bracis-19080	179	10	models	model	NOUN
bracis-19080	179	11	open	open	ADJ
bracis-19080	179	12	sourced	source	VERB
bracis-19080	179	13	by	by	ADP
bracis-19080	179	14	sbert	sbert	NOUN
bracis-19080	179	15	proponents	proponent	NOUN
bracis-19080	179	16	.	.	PUNCT
bracis-19080	180	1	we	we	PRON
bracis-19080	180	2	choose	choose	VERB
bracis-19080	180	3	to	to	PART
bracis-19080	180	4	employ	employ	VERB
bracis-19080	180	5	pre	pre	ADJ
bracis-19080	180	6	-	-	ADJ
bracis-19080	180	7	trained	train	VERB
bracis-19080	180	8	models	model	NOUN
bracis-19080	180	9	instead	instead	ADV
bracis-19080	180	10	of	of	ADP
bracis-19080	180	11	training	training	NOUN
bracis-19080	180	12	models	model	NOUN
bracis-19080	180	13	from	from	ADP
bracis-19080	180	14	scratch	scratch	NOUN
bracis-19080	180	15	for	for	ADP
bracis-19080	180	16	two	two	NUM
bracis-19080	180	17	major	major	ADJ
bracis-19080	180	18	reasons	reason	NOUN
bracis-19080	180	19	.	.	PUNCT
bracis-19080	181	1	first	first	ADV
bracis-19080	181	2	,	,	PUNCT
bracis-19080	181	3	the	the	DET
bracis-19080	181	4	computational	computational	ADJ
bracis-19080	181	5	cost	cost	NOUN
bracis-19080	181	6	required	require	VERB
bracis-19080	181	7	for	for	ADP
bracis-19080	181	8	training	training	NOUN
bracis-19080	181	9	deep	deep	ADJ
bracis-19080	181	10	learning	learn	VERB
bracis-19080	181	11	nlp	nlp	NOUN
bracis-19080	181	12	models	model	NOUN
bracis-19080	181	13	from	from	ADP
bracis-19080	181	14	scratch	scratch	NOUN
bracis-19080	181	15	is	be	AUX
bracis-19080	181	16	considerable	considerable	ADJ
bracis-19080	181	17	.	.	PUNCT
bracis-19080	182	1	second	second	ADV
bracis-19080	182	2	,	,	PUNCT
bracis-19080	182	3	the	the	DET
bracis-19080	182	4	pre	pre	ADJ
bracis-19080	182	5	-	-	ADJ
bracis-19080	182	6	trained	train	VERB
bracis-19080	182	7	models	model	NOUN
bracis-19080	182	8	have	have	AUX
bracis-19080	182	9	often	often	ADV
bracis-19080	182	10	been	be	AUX
bracis-19080	182	11	trained	train	VERB
bracis-19080	182	12	on	on	ADP
bracis-19080	182	13	very	very	ADV
bracis-19080	182	14	rich	rich	ADJ
bracis-19080	182	15	corpora	corpora	NOUN
bracis-19080	182	16	,	,	PUNCT
bracis-19080	182	17	whether	whether	SCONJ
bracis-19080	182	18	in	in	ADP
bracis-19080	182	19	portuguese	portuguese	NOUN
bracis-19080	182	20	or	or	CCONJ
bracis-19080	182	21	multilingual	multilingual	ADJ
bracis-19080	182	22	.	.	PUNCT
bracis-19080	183	1	below	below	ADV
bracis-19080	183	2	,	,	PUNCT
bracis-19080	183	3	we	we	PRON
bracis-19080	183	4	describe	describe	VERB
bracis-19080	183	5	the	the	DET
bracis-19080	183	6	bertimbau	bertimbau	NOUN
bracis-19080	183	7	and	and	CCONJ
bracis-19080	183	8	multilingual	multilingual	ADJ
bracis-19080	183	9	models	model	NOUN
bracis-19080	183	10	.	.	PUNCT
bracis-19080	184	1	bertimbau	bertimbau	NOUN
bracis-19080	184	2	models	model	NOUN
bracis-19080	184	3	were	be	AUX
bracis-19080	184	4	trained	train	VERB
bracis-19080	184	5	by	by	ADP
bracis-19080	184	6	neuralmind	neuralmind	NOUN
bracis-19080	184	7	and	and	CCONJ
bracis-19080	184	8	are	be	AUX
bracis-19080	184	9	available	available	ADJ
bracis-19080	184	10	in	in	ADP
bracis-19080	184	11	two	two	NUM
bracis-19080	184	12	sizes	size	NOUN
bracis-19080	184	13	,	,	PUNCT
bracis-19080	184	14	namely	namely	ADV
bracis-19080	184	15	bertimbau	bertimbau	NOUN
bracis-19080	184	16	-	-	PUNCT
bracis-19080	184	17	base	base	NOUN
bracis-19080	184	18	and	and	CCONJ
bracis-19080	184	19	bertimbau	bertimbau	NOUN
bracis-19080	184	20	-	-	PUNCT
bracis-19080	184	21	large	large	ADJ
bracis-19080	184	22	.	.	PUNCT
bracis-19080	185	1	pre	pre	ADJ
bracis-19080	185	2	-	-	ADJ
bracis-19080	185	3	training	training	ADJ
bracis-19080	185	4	bert	bert	NOUN
bracis-19080	185	5	requires	require	VERB
bracis-19080	185	6	a	a	DET
bracis-19080	185	7	vocabulary	vocabulary	NOUN
bracis-19080	185	8	and	and	CCONJ
bracis-19080	185	9	a	a	DET
bracis-19080	185	10	training	training	NOUN
bracis-19080	185	11	corpus	corpus	NOUN
bracis-19080	185	12	.	.	PUNCT
bracis-19080	186	1	neuralmind	neuralmind	PROPN
bracis-19080	186	2	produced	produce	VERB
bracis-19080	186	3	a	a	DET
bracis-19080	186	4	vocabulary	vocabulary	NOUN
bracis-19080	186	5	from	from	ADP
bracis-19080	186	6	the	the	DET
bracis-19080	186	7	brazilian	brazilian	ADJ
bracis-19080	186	8	portuguese	portuguese	ADJ
bracis-19080	186	9	version	version	NOUN
bracis-19080	186	10	of	of	ADP
bracis-19080	186	11	the	the	DET
bracis-19080	186	12	wikipedia	wikipedia	PROPN
bracis-19080	186	13	,	,	PUNCT
bracis-19080	186	14	and	and	CCONJ
bracis-19080	186	15	used	use	VERB
bracis-19080	186	16	the	the	DET
bracis-19080	186	17	brwac	brwac	PROPN
bracis-19080	186	18	corpus	corpus	PROPN
bracis-19080	186	19	 	 	SPACE
bracis-19080	187	1	[	[	X
bracis-19080	187	2	22	22	NUM
bracis-19080	187	3	]	]	PUNCT
bracis-19080	187	4	.	.	PUNCT
bracis-19080	188	1	pre	pre	VERB
bracis-19080	188	2	-	-	ADJ
bracis-19080	188	3	training	training	NOUN
bracis-19080	188	4	followed	follow	VERB
bracis-19080	188	5	the	the	DET
bracis-19080	188	6	original	original	ADJ
bracis-19080	188	7	bert	bert	PROPN
bracis-19080	188	8	paper	paper	NOUN
bracis-19080	188	9	 	 	SPACE
bracis-19080	189	1	[	[	X
bracis-19080	189	2	7	7	NUM
bracis-19080	189	3	]	]	PUNCT
bracis-19080	189	4	.	.	PUNCT
bracis-19080	190	1	bertimbau	bertimbau	NOUN
bracis-19080	190	2	-	-	PUNCT
bracis-19080	190	3	base	base	NOUN
bracis-19080	190	4	and	and	CCONJ
bracis-19080	190	5	bertimbau	bertimbau	NOUN
bracis-19080	190	6	-	-	PUNCT
bracis-19080	190	7	large	large	ADJ
bracis-19080	190	8	were	be	AUX
bracis-19080	190	9	respectively	respectively	ADV
bracis-19080	190	10	warm	warm	ADJ
bracis-19080	190	11	-	-	PUNCT
bracis-19080	190	12	started	start	VERB
bracis-19080	190	13	with	with	ADP
bracis-19080	190	14	the	the	DET
bracis-19080	190	15	multilingual	multilingual	ADJ
bracis-19080	190	16	bert	bert	NOUN
bracis-19080	190	17	-	-	PUNCT
bracis-19080	190	18	base	base	NOUN
bracis-19080	190	19	and	and	CCONJ
bracis-19080	190	20	english	english	ADJ
bracis-19080	190	21	bert	bert	PROPN
bracis-19080	190	22	-	-	PUNCT
bracis-19080	190	23	large	large	ADJ
bracis-19080	190	24	checkpoints	checkpoint	NOUN
bracis-19080	190	25	.	.	PUNCT
bracis-19080	191	1	sbert	sbert	NOUN
bracis-19080	191	2	models	model	NOUN
bracis-19080	191	3	pre	pre	VERB
bracis-19080	191	4	-	-	VERB
bracis-19080	191	5	trained	train	VERB
bracis-19080	191	6	in	in	ADP
bracis-19080	191	7	portuguese	portuguese	PROPN
bracis-19080	191	8	are	be	AUX
bracis-19080	191	9	not	not	PART
bracis-19080	191	10	openly	openly	ADV
bracis-19080	191	11	available	available	ADJ
bracis-19080	191	12	.	.	PUNCT
bracis-19080	192	1	instead	instead	ADV
bracis-19080	192	2	,	,	PUNCT
bracis-19080	192	3	we	we	PRON
bracis-19080	192	4	adopt	adopt	VERB
bracis-19080	192	5	two	two	NUM
bracis-19080	192	6	sets	set	NOUN
bracis-19080	192	7	of	of	ADP
bracis-19080	192	8	models	model	NOUN
bracis-19080	192	9	,	,	PUNCT
bracis-19080	192	10	given	give	VERB
bracis-19080	192	11	in	in	ADP
bracis-19080	192	12	table	table	NOUN
bracis-19080	192	13	 	 	SPACE
bracis-19080	192	14	1	1	NUM
bracis-19080	192	15	on	on	ADP
bracis-19080	192	16	the	the	DET
bracis-19080	192	17	five	five	NUM
bracis-19080	192	18	rows	row	NOUN
bracis-19080	192	19	that	that	PRON
bracis-19080	192	20	follow	follow	VERB
bracis-19080	192	21	baseline	baseline	NOUN
bracis-19080	192	22	models	model	NOUN
bracis-19080	192	23	.	.	PUNCT
bracis-19080	193	1	the	the	DET
bracis-19080	193	2	first	first	ADJ
bracis-19080	193	3	two	two	NUM
bracis-19080	193	4	models	model	NOUN
bracis-19080	193	5	,	,	PUNCT
bracis-19080	193	6	labeled	label	VERB
bracis-19080	193	7	bertimbau	bertimbau	NOUN
bracis-19080	193	8	-	-	PUNCT
bracis-19080	193	9	base	base	NOUN
bracis-19080	193	10	 	 	SPACE
bracis-19080	193	11	(	(	PUNCT
bracis-19080	193	12	sbert	sbert	NOUN
bracis-19080	193	13	)	)	PUNCT
bracis-19080	193	14	and	and	CCONJ
bracis-19080	193	15	bertimbau	bertimbau	NOUN
bracis-19080	193	16	-	-	PUNCT
bracis-19080	193	17	large	large	ADJ
bracis-19080	193	18	 	 	SPACE
bracis-19080	193	19	(	(	PUNCT
bracis-19080	193	20	sbert	sbert	NOUN
bracis-19080	193	21	)	)	PUNCT
bracis-19080	193	22	,	,	PUNCT
bracis-19080	193	23	reuse	reuse	VERB
bracis-19080	193	24	the	the	DET
bracis-19080	193	25	trained	train	VERB
bracis-19080	193	26	weights	weight	NOUN
bracis-19080	193	27	provided	provide	VERB
bracis-19080	193	28	by	by	ADP
bracis-19080	193	29	bertimbau	bertimbau	NOUN
bracis-19080	193	30	models	model	NOUN
bracis-19080	193	31	.	.	PUNCT
bracis-19080	194	1	the	the	DET
bracis-19080	194	2	remaining	remain	VERB
bracis-19080	194	3	three	three	NUM
bracis-19080	194	4	models	model	NOUN
bracis-19080	194	5	are	be	AUX
bracis-19080	194	6	multilingual	multilingual	ADJ
bracis-19080	194	7	models	model	NOUN
bracis-19080	194	8	open	open	ADV
bracis-19080	194	9	-	-	PUNCT
bracis-19080	194	10	sourced	source	VERB
bracis-19080	194	11	by	by	ADP
bracis-19080	194	12	sbert	sbert	NOUN
bracis-19080	194	13	proponents	proponent	NOUN
bracis-19080	194	14	,	,	PUNCT
bracis-19080	194	15	having	having	AUX
bracis-19080	194	16	been	be	AUX
bracis-19080	194	17	trained	train	VERB
bracis-19080	194	18	using	use	VERB
bracis-19080	194	19	the	the	DET
bracis-19080	194	20	knowledge	knowledge	NOUN
bracis-19080	194	21	distillation	distillation	NOUN
bracis-19080	194	22	process	process	NOUN
bracis-19080	194	23	.	.	PUNCT
bracis-19080	195	1	among	among	ADP
bracis-19080	195	2	the	the	DET
bracis-19080	195	3	models	model	NOUN
bracis-19080	195	4	available	available	ADJ
bracis-19080	195	5	,	,	PUNCT
bracis-19080	195	6	we	we	PRON
bracis-19080	195	7	used	use	VERB
bracis-19080	195	8	the	the	DET
bracis-19080	195	9	ones	one	NOUN
bracis-19080	195	10	for	for	ADP
bracis-19080	195	11	which	which	PRON
bracis-19080	195	12	the	the	DET
bracis-19080	195	13	student	student	NOUN
bracis-19080	195	14	-	-	PUNCT
bracis-19080	195	15	model	model	NOUN
bracis-19080	195	16	was	be	AUX
bracis-19080	195	17	learned	learn	VERB
bracis-19080	195	18	by	by	ADP
bracis-19080	195	19	the	the	DET
bracis-19080	195	20	xlm	xlm	PROPN
bracis-19080	195	21	-	-	PUNCT
bracis-19080	195	22	r	r	NOUN
bracis-19080	195	23	algorithm	algorithm	NOUN
bracis-19080	195	24	 	 	SPACE
bracis-19080	196	1	[	[	X
bracis-19080	196	2	6	6	NUM
bracis-19080	196	3	]	]	PUNCT
bracis-19080	196	4	.	.	PUNCT
bracis-19080	197	1	different	different	ADJ
bracis-19080	197	2	teacher	teacher	NOUN
bracis-19080	197	3	-	-	PUNCT
bracis-19080	197	4	models	model	NOUN
bracis-19080	197	5	were	be	AUX
bracis-19080	197	6	used	use	VERB
bracis-19080	197	7	,	,	PUNCT
bracis-19080	197	8	described	describe	VERB
bracis-19080	197	9	below	below	ADV
bracis-19080	197	10	under	under	ADP
bracis-19080	197	11	the	the	DET
bracis-19080	197	12	notation	notation	NOUN
bracis-19080	198	1	\(\langle	\(\langle	PROPN
bracis-19080	198	2	\textsf	\textsf	PROPN
bracis-19080	198	3	{	{	PUNCT
bracis-19080	198	4	student	student	NOUN
bracis-19080	198	5	,	,	PUNCT
bracis-19080	198	6	teacher	teacher	NOUN
bracis-19080	198	7	,	,	PUNCT
bracis-19080	198	8	task}\rangle	task}\rangle	PROPN
bracis-19080	198	9	\	\	PROPN
bracis-19080	198	10	)	)	PUNCT
bracis-19080	198	11	,	,	PUNCT
bracis-19080	198	12	indicating	indicate	VERB
bracis-19080	198	13	the	the	DET
bracis-19080	198	14	student	student	NOUN
bracis-19080	198	15	-	-	PUNCT
bracis-19080	198	16	model	model	NOUN
bracis-19080	198	17	and	and	CCONJ
bracis-19080	198	18	the	the	DET
bracis-19080	198	19	task	task	NOUN
bracis-19080	198	20	for	for	ADP
bracis-19080	198	21	which	which	PRON
bracis-19080	198	22	the	the	DET
bracis-19080	198	23	teacher	teacher	NOUN
bracis-19080	198	24	-	-	PUNCT
bracis-19080	198	25	model	model	NOUN
bracis-19080	198	26	was	be	AUX
bracis-19080	198	27	originally	originally	ADV
bracis-19080	198	28	trained	train	VERB
bracis-19080	198	29	.	.	PUNCT
bracis-19080	199	1	\(\langle	\(\langle	PROPN
bracis-19080	199	2	\textsf	\textsf	PROPN
bracis-19080	199	3	{	{	PUNCT
bracis-19080	199	4	xlm	xlm	PROPN
bracis-19080	199	5	-	-	PUNCT
bracis-19080	199	6	r	r	NOUN
bracis-19080	199	7	,	,	PUNCT
bracis-19080	199	8	sbert	sbert	NOUN
bracis-19080	199	9	,	,	PUNCT
bracis-19080	199	10	nli+sts}\rangle	nli+sts}\rangle	PROPN
bracis-19080	199	11	\	\	PROPN
bracis-19080	199	12	):	):	PUNCT
bracis-19080	199	13	xlm	xlm	PROPN
bracis-19080	199	14	-	-	PUNCT
bracis-19080	199	15	r	r	NOUN
bracis-19080	199	16	learned	learn	VERB
bracis-19080	199	17	a	a	DET
bracis-19080	199	18	multilingual	multilingual	ADJ
bracis-19080	199	19	model	model	NOUN
bracis-19080	199	20	from	from	ADP
bracis-19080	199	21	an	an	DET
bracis-19080	199	22	sbert	sbert	NOUN
bracis-19080	199	23	model	model	NOUN
bracis-19080	199	24	fine	fine	ADV
bracis-19080	199	25	-	-	PUNCT
bracis-19080	199	26	tuned	tune	VERB
bracis-19080	199	27	for	for	ADP
bracis-19080	199	28	nli	nli	NOUN
bracis-19080	199	29	and	and	CCONJ
bracis-19080	199	30	sts	st	VERB
bracis-19080	199	31	in	in	ADP
bracis-19080	199	32	english	english	PROPN
bracis-19080	199	33	.	.	PUNCT
bracis-19080	200	1	\(\langle	\(\langle	PROPN
bracis-19080	200	2	\textsf	\textsf	PROPN
bracis-19080	200	3	{	{	PUNCT
bracis-19080	200	4	xlm	xlm	PROPN
bracis-19080	200	5	-	-	PUNCT
bracis-19080	200	6	r	r	NOUN
bracis-19080	200	7	,	,	PUNCT
bracis-19080	200	8	sbert	sbert	NOUN
bracis-19080	200	9	,	,	PUNCT
bracis-19080	200	10	paraphrases}\rangle	paraphrases}\rangle	NUM
bracis-19080	200	11	\	\	PROPN
bracis-19080	200	12	):	):	PUNCT
bracis-19080	200	13	xlm	xlm	PROPN
bracis-19080	200	14	-	-	PUNCT
bracis-19080	200	15	r	r	NOUN
bracis-19080	200	16	learned	learn	VERB
bracis-19080	200	17	a	a	DET
bracis-19080	200	18	multilingual	multilingual	ADJ
bracis-19080	200	19	model	model	NOUN
bracis-19080	200	20	from	from	ADP
bracis-19080	200	21	an	an	DET
bracis-19080	200	22	sbert	sbert	NOUN
bracis-19080	200	23	model	model	NOUN
bracis-19080	200	24	trained	train	VERB
bracis-19080	200	25	on	on	ADP
bracis-19080	200	26	an	an	DET
bracis-19080	200	27	english	english	ADJ
bracis-19080	200	28	paraphases	paraphase	NOUN
bracis-19080	200	29	dataset	dataset	VERB
bracis-19080	200	30	.	.	PUNCT
bracis-19080	201	1	\(\langle	\(\langle	VERB
bracis-19080	201	2	\textsf	\textsf	PROPN
bracis-19080	201	3	{	{	PUNCT
bracis-19080	201	4	xlm	xlm	PROPN
bracis-19080	201	5	-	-	PUNCT
bracis-19080	201	6	r	r	NOUN
bracis-19080	201	7	,	,	PUNCT
bracis-19080	201	8	sbert	sbert	NOUN
bracis-19080	201	9	,	,	PUNCT
bracis-19080	201	10	cross	cross	ADJ
bracis-19080	201	11	-	-	ADJ
bracis-19080	201	12	en	en	ADJ
bracis-19080	201	13	-	-	PUNCT
bracis-19080	201	14	de}\rangle	de}\rangle	ADJ
bracis-19080	201	15	\	\	NOUN
bracis-19080	201	16	):	):	PUNCT
bracis-19080	201	17	fine	fine	ADV
bracis-19080	201	18	-	-	PUNCT
bracis-19080	201	19	tuned	tune	VERB
bracis-19080	201	20	\(\langle	\(\langle	NOUN
bracis-19080	201	21	\textsf	\textsf	PROPN
bracis-19080	201	22	{	{	PUNCT
bracis-19080	201	23	xlm	xlm	PROPN
bracis-19080	201	24	-	-	PUNCT
bracis-19080	201	25	r	r	NOUN
bracis-19080	201	26	,	,	PUNCT
bracis-19080	201	27	sbert	sbert	NOUN
bracis-19080	201	28	,	,	PUNCT
bracis-19080	201	29	paraphrases}\rangle	paraphrases}\rangle	NUM
bracis-19080	201	30	\	\	NOUN
bracis-19080	201	31	)	)	PUNCT
bracis-19080	201	32	model	model	NOUN
bracis-19080	201	33	for	for	ADP
bracis-19080	201	34	the	the	DET
bracis-19080	201	35	stsbenchmark	stsbenchmark	NOUN
bracis-19080	201	36	 	 	SPACE
bracis-19080	202	1	[	[	X
bracis-19080	202	2	5	5	NUM
bracis-19080	202	3	]	]	PUNCT
bracis-19080	202	4	,	,	PUNCT
bracis-19080	202	5	with	with	ADP
bracis-19080	202	6	english	english	ADJ
bracis-19080	202	7	and/or	and/or	CCONJ
bracis-19080	202	8	german	german	ADJ
bracis-19080	202	9	sentences	sentence	NOUN
bracis-19080	202	10	.	.	PUNCT
bracis-19080	203	1	training	train	VERB
bracis-19080	203	2	multilingual	multilingual	ADJ
bracis-19080	203	3	versions	version	NOUN
bracis-19080	203	4	used	use	VERB
bracis-19080	203	5	a	a	DET
bracis-19080	203	6	range	range	NOUN
bracis-19080	203	7	of	of	ADP
bracis-19080	203	8	datasets	dataset	NOUN
bracis-19080	203	9	,	,	PUNCT
bracis-19080	203	10	and	and	CCONJ
bracis-19080	203	11	so	so	ADV
bracis-19080	203	12	contextual	contextual	ADJ
bracis-19080	203	13	embeddings	embedding	NOUN
bracis-19080	203	14	produced	produce	VERB
bracis-19080	203	15	by	by	ADP
bracis-19080	203	16	sbert	sbert	NOUN
bracis-19080	203	17	were	be	AUX
bracis-19080	203	18	generalized	generalize	VERB
bracis-19080	203	19	by	by	ADP
bracis-19080	203	20	xlm	xlm	PROPN
bracis-19080	203	21	-	-	PUNCT
bracis-19080	203	22	r	r	NOUN
bracis-19080	203	23	for	for	ADP
bracis-19080	203	24	50	50	NUM
bracis-19080	203	25	different	different	ADJ
bracis-19080	203	26	languages	language	NOUN
bracis-19080	203	27	.	.	PUNCT
bracis-19080	204	1	further	further	ADJ
bracis-19080	204	2	information	information	NOUN
bracis-19080	204	3	about	about	ADP
bracis-19080	204	4	their	their	PRON
bracis-19080	204	5	training	training	NOUN
bracis-19080	204	6	is	be	AUX
bracis-19080	204	7	available	available	ADJ
bracis-19080	204	8	in	in	ADP
bracis-19080	204	9	the	the	DET
bracis-19080	204	10	original	original	ADJ
bracis-19080	204	11	paper	paper	NOUN
bracis-19080	204	12	 	 	SPACE
bracis-19080	205	1	[	[	X
bracis-19080	205	2	18	18	NUM
bracis-19080	205	3	]	]	PUNCT
bracis-19080	205	4	.	.	PUNCT
bracis-19080	206	1	4.2	4.2	NUM
bracis-19080	206	2	experimental	experimental	ADJ
bracis-19080	206	3	setup	setup	NOUN
bracis-19080	206	4	as	as	SCONJ
bracis-19080	206	5	previously	previously	ADV
bracis-19080	206	6	discussed	discuss	VERB
bracis-19080	206	7	,	,	PUNCT
bracis-19080	206	8	we	we	PRON
bracis-19080	206	9	split	split	VERB
bracis-19080	206	10	our	our	PRON
bracis-19080	206	11	experiments	experiment	NOUN
bracis-19080	206	12	in	in	ADP
bracis-19080	206	13	two	two	NUM
bracis-19080	206	14	parts	part	NOUN
bracis-19080	206	15	to	to	PART
bracis-19080	206	16	isolate	isolate	VERB
bracis-19080	206	17	the	the	DET
bracis-19080	206	18	effect	effect	NOUN
bracis-19080	206	19	of	of	ADP
bracis-19080	206	20	fine	fine	ADV
bracis-19080	206	21	-	-	PUNCT
bracis-19080	206	22	tuning	tuning	NOUN
bracis-19080	206	23	.	.	PUNCT
bracis-19080	207	1	this	this	PRON
bracis-19080	207	2	is	be	AUX
bracis-19080	207	3	reflected	reflect	VERB
bracis-19080	207	4	in	in	ADP
bracis-19080	207	5	table	table	NOUN
bracis-19080	207	6	 	 	SPACE
bracis-19080	207	7	1	1	NUM
bracis-19080	207	8	,	,	PUNCT
bracis-19080	207	9	where	where	SCONJ
bracis-19080	207	10	we	we	PRON
bracis-19080	207	11	label	label	VERB
bracis-19080	207	12	results	result	NOUN
bracis-19080	207	13	obtained	obtain	VERB
bracis-19080	207	14	from	from	ADP
bracis-19080	207	15	a	a	DET
bracis-19080	207	16	given	give	VERB
bracis-19080	207	17	model	model	NOUN
bracis-19080	207	18	after	after	ADP
bracis-19080	207	19	fine	fine	ADV
bracis-19080	207	20	-	-	PUNCT
bracis-19080	207	21	tuning	tuning	NOUN
bracis-19080	207	22	with	with	ADP
bracis-19080	207	23	the	the	DET
bracis-19080	207	24	subscript	subscript	NOUN
bracis-19080	207	25	\(\textit{ft}\	\(\textit{ft}\	PROPN
bracis-19080	207	26	)	)	PUNCT
bracis-19080	207	27	.	.	PUNCT
bracis-19080	208	1	each	each	DET
bracis-19080	208	2	part	part	NOUN
bracis-19080	208	3	of	of	ADP
bracis-19080	208	4	the	the	DET
bracis-19080	208	5	investigation	investigation	NOUN
bracis-19080	208	6	is	be	AUX
bracis-19080	208	7	further	far	ADV
bracis-19080	208	8	detailed	detail	VERB
bracis-19080	208	9	next	next	ADV
bracis-19080	208	10	.	.	PUNCT
bracis-19080	209	1	experiments	experiment	NOUN
bracis-19080	209	2	without	without	ADP
bracis-19080	209	3	fine	fine	ADV
bracis-19080	209	4	-	-	PUNCT
bracis-19080	209	5	tuning	tuning	NOUN
bracis-19080	209	6	.	.	PUNCT
bracis-19080	210	1	in	in	ADP
bracis-19080	210	2	the	the	DET
bracis-19080	210	3	first	first	ADJ
bracis-19080	210	4	part	part	NOUN
bracis-19080	210	5	of	of	ADP
bracis-19080	210	6	our	our	PRON
bracis-19080	210	7	experiments	experiment	NOUN
bracis-19080	210	8	,	,	PUNCT
bracis-19080	210	9	we	we	PRON
bracis-19080	210	10	employed	employ	VERB
bracis-19080	210	11	the	the	DET
bracis-19080	210	12	pre	pre	ADJ
bracis-19080	210	13	-	-	ADJ
bracis-19080	210	14	trained	train	VERB
bracis-19080	210	15	models	model	NOUN
bracis-19080	210	16	on	on	ADP
bracis-19080	210	17	assin	assin	PROPN
bracis-19080	210	18	and	and	CCONJ
bracis-19080	210	19	assin	assin	PROPN
bracis-19080	210	20	 	 	SPACE
bracis-19080	210	21	2	2	NUM
bracis-19080	210	22	datasets	dataset	NOUN
bracis-19080	210	23	without	without	ADP
bracis-19080	210	24	fine	fine	ADV
bracis-19080	210	25	-	-	PUNCT
bracis-19080	210	26	tuning	tuning	NOUN
bracis-19080	210	27	.	.	PUNCT
bracis-19080	211	1	concretely	concretely	ADV
bracis-19080	211	2	,	,	PUNCT
bracis-19080	211	3	labels	label	NOUN
bracis-19080	211	4	provided	provide	VERB
bracis-19080	211	5	by	by	ADP
bracis-19080	211	6	the	the	DET
bracis-19080	211	7	datasets	dataset	NOUN
bracis-19080	211	8	were	be	AUX
bracis-19080	211	9	used	use	VERB
bracis-19080	211	10	only	only	ADV
bracis-19080	211	11	to	to	PART
bracis-19080	211	12	validate	validate	VERB
bracis-19080	211	13	the	the	DET
bracis-19080	211	14	results	result	NOUN
bracis-19080	211	15	.	.	PUNCT
bracis-19080	212	1	for	for	ADP
bracis-19080	212	2	the	the	DET
bracis-19080	212	3	baseline	baseline	ADJ
bracis-19080	212	4	bertimbau	bertimbau	NOUN
bracis-19080	212	5	models	model	NOUN
bracis-19080	212	6	 	 	SPACE
bracis-19080	212	7	(	(	PUNCT
bracis-19080	212	8	bertimbau	bertimbau	NOUN
bracis-19080	212	9	-	-	PUNCT
bracis-19080	212	10	base	base	NOUN
bracis-19080	212	11	and	and	CCONJ
bracis-19080	212	12	bertimbau	bertimbau	NOUN
bracis-19080	212	13	-	-	PUNCT
bracis-19080	212	14	large	large	ADJ
bracis-19080	212	15	)	)	PUNCT
bracis-19080	212	16	,	,	PUNCT
bracis-19080	212	17	each	each	DET
bracis-19080	212	18	input	input	NOUN
bracis-19080	212	19	sentence	sentence	NOUN
bracis-19080	212	20	was	be	AUX
bracis-19080	212	21	fed	feed	VERB
bracis-19080	212	22	individually	individually	ADV
bracis-19080	212	23	to	to	ADP
bracis-19080	212	24	the	the	DET
bracis-19080	212	25	model	model	NOUN
bracis-19080	212	26	and	and	CCONJ
bracis-19080	212	27	the	the	DET
bracis-19080	212	28	corresponding	corresponding	ADJ
bracis-19080	212	29	embedding	embedding	NOUN
bracis-19080	212	30	was	be	AUX
bracis-19080	212	31	obtained	obtain	VERB
bracis-19080	212	32	from	from	ADP
bracis-19080	212	33	the	the	DET
bracis-19080	212	34	[	[	X
bracis-19080	212	35	cls	cls	X
bracis-19080	212	36	]	]	X
bracis-19080	212	37	token	token	PROPN
bracis-19080	212	38	.	.	PUNCT
bracis-19080	213	1	the	the	DET
bracis-19080	213	2	similarity	similarity	NOUN
bracis-19080	213	3	score	score	NOUN
bracis-19080	213	4	for	for	ADP
bracis-19080	213	5	a	a	DET
bracis-19080	213	6	pair	pair	NOUN
bracis-19080	213	7	of	of	ADP
bracis-19080	213	8	sentences	sentence	NOUN
bracis-19080	213	9	was	be	AUX
bracis-19080	213	10	calculated	calculate	VERB
bracis-19080	213	11	as	as	ADP
bracis-19080	213	12	the	the	DET
bracis-19080	213	13	cosine	cosine	NOUN
bracis-19080	213	14	distance	distance	NOUN
bracis-19080	213	15	between	between	ADP
bracis-19080	213	16	the	the	DET
bracis-19080	213	17	sentence	sentence	NOUN
bracis-19080	213	18	embeddings	embedding	NOUN
bracis-19080	213	19	produced	produce	VERB
bracis-19080	213	20	.	.	PUNCT
bracis-19080	214	1	then	then	ADV
bracis-19080	214	2	,	,	PUNCT
bracis-19080	214	3	we	we	PRON
bracis-19080	214	4	compare	compare	VERB
bracis-19080	214	5	the	the	DET
bracis-19080	214	6	cosine	cosine	NOUN
bracis-19080	214	7	similarity	similarity	NOUN
bracis-19080	214	8	directly	directly	ADV
bracis-19080	214	9	with	with	ADP
bracis-19080	214	10	the	the	DET
bracis-19080	214	11	labels	label	NOUN
bracis-19080	214	12	of	of	ADP
bracis-19080	214	13	assin	assin	PROPN
bracis-19080	214	14	and	and	CCONJ
bracis-19080	214	15	assin	assin	PROPN
bracis-19080	214	16	2	2	NUM
bracis-19080	214	17	using	use	VERB
bracis-19080	214	18	pearson	pearson	NOUN
bracis-19080	214	19	correlation	correlation	NOUN
bracis-19080	214	20	.	.	PUNCT
bracis-19080	215	1	though	though	SCONJ
bracis-19080	215	2	cosine	cosine	NOUN
bracis-19080	215	3	distance	distance	NOUN
bracis-19080	215	4	ranges	range	VERB
bracis-19080	215	5	differs	differ	VERB
bracis-19080	215	6	from	from	ADP
bracis-19080	215	7	the	the	DET
bracis-19080	215	8	sts	st	NOUN
bracis-19080	215	9	score	score	NOUN
bracis-19080	215	10	range	range	NOUN
bracis-19080	215	11	,	,	PUNCT
bracis-19080	215	12	normalization	normalization	NOUN
bracis-19080	215	13	and	and	CCONJ
bracis-19080	215	14	scaling	scaling	NOUN
bracis-19080	215	15	is	be	AUX
bracis-19080	215	16	not	not	PART
bracis-19080	215	17	necessary	necessary	ADJ
bracis-19080	215	18	since	since	SCONJ
bracis-19080	215	19	pearson	pearson	NOUN
bracis-19080	215	20	correlation	correlation	NOUN
bracis-19080	215	21	only	only	ADV
bracis-19080	215	22	considers	consider	VERB
bracis-19080	215	23	the	the	DET
bracis-19080	215	24	linear	linear	ADJ
bracis-19080	215	25	relationship	relationship	NOUN
bracis-19080	215	26	between	between	ADP
bracis-19080	215	27	the	the	DET
bracis-19080	215	28	variables	variable	NOUN
bracis-19080	215	29	.	.	PUNCT
bracis-19080	216	1	experiments	experiment	NOUN
bracis-19080	216	2	with	with	ADP
bracis-19080	216	3	fine	fine	ADV
bracis-19080	216	4	-	-	PUNCT
bracis-19080	216	5	tuning	tuning	NOUN
bracis-19080	216	6	.	.	PUNCT
bracis-19080	217	1	in	in	ADP
bracis-19080	217	2	the	the	DET
bracis-19080	217	3	second	second	ADJ
bracis-19080	217	4	part	part	NOUN
bracis-19080	217	5	of	of	ADP
bracis-19080	217	6	our	our	PRON
bracis-19080	217	7	experimental	experimental	ADJ
bracis-19080	217	8	analysis	analysis	NOUN
bracis-19080	217	9	,	,	PUNCT
bracis-19080	217	10	we	we	PRON
bracis-19080	217	11	assessed	assess	VERB
bracis-19080	217	12	the	the	DET
bracis-19080	217	13	impact	impact	NOUN
bracis-19080	217	14	in	in	ADP
bracis-19080	217	15	performance	performance	NOUN
bracis-19080	217	16	when	when	SCONJ
bracis-19080	217	17	pre	pre	ADJ
bracis-19080	217	18	-	-	ADJ
bracis-19080	217	19	trained	train	VERB
bracis-19080	217	20	models	model	NOUN
bracis-19080	217	21	are	be	AUX
bracis-19080	217	22	fine	fine	ADV
bracis-19080	217	23	-	-	PUNCT
bracis-19080	217	24	tuned	tune	VERB
bracis-19080	217	25	for	for	ADP
bracis-19080	217	26	the	the	DET
bracis-19080	217	27	assin	assin	PROPN
bracis-19080	217	28	and	and	CCONJ
bracis-19080	217	29	assin	assin	PROPN
bracis-19080	217	30	 	 	SPACE
bracis-19080	217	31	2	2	NUM
bracis-19080	217	32	datasets	dataset	NOUN
bracis-19080	217	33	.	.	PUNCT
bracis-19080	218	1	in	in	ADP
bracis-19080	218	2	detail	detail	NOUN
bracis-19080	218	3	,	,	PUNCT
bracis-19080	218	4	we	we	PRON
bracis-19080	218	5	use	use	VERB
bracis-19080	218	6	the	the	DET
bracis-19080	218	7	available	available	ADJ
bracis-19080	218	8	validation	validation	NOUN
bracis-19080	218	9	labels	label	NOUN
bracis-19080	218	10	to	to	ADP
bracis-19080	218	11	fine	fine	ADJ
bracis-19080	218	12	-	-	PUNCT
bracis-19080	218	13	tune	tune	NOUN
bracis-19080	218	14	the	the	DET
bracis-19080	218	15	pre	pre	ADJ
bracis-19080	218	16	-	-	ADJ
bracis-19080	218	17	trained	train	VERB
bracis-19080	218	18	models	model	NOUN
bracis-19080	218	19	.	.	PUNCT
bracis-19080	219	1	the	the	DET
bracis-19080	219	2	developers	developer	NOUN
bracis-19080	219	3	of	of	ADP
bracis-19080	219	4	sbert	sbert	NOUN
bracis-19080	219	5	made	make	VERB
bracis-19080	219	6	available	available	ADJ
bracis-19080	219	7	a	a	DET
bracis-19080	219	8	standard	standard	ADJ
bracis-19080	219	9	fine	fine	ADV
bracis-19080	219	10	-	-	PUNCT
bracis-19080	219	11	tuning	tune	VERB
bracis-19080	219	12	script	script	NOUN
bracis-19080	219	13	,	,	PUNCT
bracis-19080	219	14	with	with	ADP
bracis-19080	219	15	recommended	recommend	VERB
bracis-19080	219	16	hyperparameters	hyperparameter	NOUN
bracis-19080	219	17	we	we	PRON
bracis-19080	219	18	adopted	adopt	VERB
bracis-19080	219	19	.	.	PUNCT
bracis-19080	220	1	for	for	ADP
bracis-19080	220	2	each	each	DET
bracis-19080	220	3	dataset	dataset	NOUN
bracis-19080	220	4	and	and	CCONJ
bracis-19080	220	5	model	model	NOUN
bracis-19080	220	6	,	,	PUNCT
bracis-19080	220	7	we	we	PRON
bracis-19080	220	8	ran	run	VERB
bracis-19080	220	9	fine	fine	ADV
bracis-19080	220	10	-	-	PUNCT
bracis-19080	220	11	tuning	tune	VERB
bracis-19080	220	12	ten	ten	NUM
bracis-19080	220	13	times	time	NOUN
bracis-19080	220	14	and	and	CCONJ
bracis-19080	220	15	report	report	NOUN
bracis-19080	220	16	mean	mean	VERB
bracis-19080	220	17	and	and	CCONJ
bracis-19080	220	18	standard	standard	ADJ
bracis-19080	220	19	deviation	deviation	NOUN
bracis-19080	220	20	results	result	NOUN
bracis-19080	220	21	.	.	PUNCT
bracis-19080	221	1	regarding	regard	VERB
bracis-19080	221	2	baseline	baseline	NOUN
bracis-19080	221	3	bertimbau	bertimbau	NOUN
bracis-19080	221	4	results	result	NOUN
bracis-19080	221	5	,	,	PUNCT
bracis-19080	221	6	fine	fine	ADV
bracis-19080	221	7	-	-	PUNCT
bracis-19080	221	8	tuning	tuning	NOUN
bracis-19080	221	9	would	would	AUX
bracis-19080	221	10	be	be	AUX
bracis-19080	221	11	computationally	computationally	ADV
bracis-19080	221	12	unfeasible	unfeasible	ADJ
bracis-19080	221	13	in	in	ADP
bracis-19080	221	14	the	the	DET
bracis-19080	221	15	context	context	NOUN
bracis-19080	221	16	of	of	ADP
bracis-19080	221	17	this	this	DET
bracis-19080	221	18	work	work	NOUN
bracis-19080	221	19	.	.	PUNCT
bracis-19080	222	1	yet	yet	ADV
bracis-19080	222	2	,	,	PUNCT
bracis-19080	222	3	we	we	PRON
bracis-19080	222	4	remark	remark	VERB
bracis-19080	222	5	that	that	SCONJ
bracis-19080	222	6	the	the	DET
bracis-19080	222	7	state	state	NOUN
bracis-19080	222	8	-	-	PUNCT
bracis-19080	222	9	of	of	ADP
bracis-19080	222	10	-	-	PUNCT
bracis-19080	222	11	the	the	DET
bracis-19080	222	12	-	-	PUNCT
bracis-19080	222	13	art	art	NOUN
bracis-19080	222	14	models	model	NOUN
bracis-19080	222	15	 	 	SPACE
bracis-19080	223	1	[	[	X
bracis-19080	223	2	8	8	NUM
bracis-19080	223	3	]	]	PUNCT
bracis-19080	223	4	are	be	AUX
bracis-19080	223	5	ptbert	ptbert	NOUN
bracis-19080	223	6	-	-	PUNCT
bracis-19080	223	7	base	base	NOUN
bracis-19080	223	8	and	and	CCONJ
bracis-19080	223	9	ptbert	ptbert	NOUN
bracis-19080	223	10	-	-	PUNCT
bracis-19080	223	11	large	large	ADJ
bracis-19080	223	12	networks	network	NOUN
bracis-19080	223	13	that	that	PRON
bracis-19080	223	14	have	have	AUX
bracis-19080	223	15	been	be	AUX
bracis-19080	223	16	fine	fine	ADV
bracis-19080	223	17	-	-	PUNCT
bracis-19080	223	18	tuned	tune	VERB
bracis-19080	223	19	for	for	ADP
bracis-19080	223	20	the	the	DET
bracis-19080	223	21	assin	assin	PROPN
bracis-19080	223	22	and	and	CCONJ
bracis-19080	223	23	assin	assin	PROPN
bracis-19080	223	24	 	 	SPACE
bracis-19080	223	25	2	2	NUM
bracis-19080	223	26	datasets	dataset	NOUN
bracis-19080	223	27	.	.	PUNCT
bracis-19080	224	1	as	as	ADP
bracis-19080	224	2	such	such	ADJ
bracis-19080	224	3	,	,	PUNCT
bracis-19080	224	4	these	these	DET
bracis-19080	224	5	models	model	NOUN
bracis-19080	224	6	serve	serve	VERB
bracis-19080	224	7	as	as	ADP
bracis-19080	224	8	reference	reference	NOUN
bracis-19080	224	9	for	for	ADP
bracis-19080	224	10	bert	bert	PROPN
bracis-19080	224	11	results	result	NOUN
bracis-19080	224	12	,	,	PUNCT
bracis-19080	224	13	though	though	SCONJ
bracis-19080	224	14	mean	mean	ADJ
bracis-19080	224	15	and	and	CCONJ
bracis-19080	224	16	standard	standard	ADJ
bracis-19080	224	17	deviation	deviation	NOUN
bracis-19080	224	18	have	have	AUX
bracis-19080	224	19	been	be	AUX
bracis-19080	224	20	reported	report	VERB
bracis-19080	224	21	only	only	ADV
bracis-19080	224	22	for	for	ADP
bracis-19080	224	23	five	five	NUM
bracis-19080	224	24	repetitions	repetition	NOUN
bracis-19080	224	25	.	.	PUNCT
bracis-19080	225	1	given	give	VERB
bracis-19080	225	2	the	the	DET
bracis-19080	225	3	large	large	ADJ
bracis-19080	225	4	number	number	NOUN
bracis-19080	225	5	of	of	ADP
bracis-19080	225	6	models	model	NOUN
bracis-19080	225	7	we	we	PRON
bracis-19080	225	8	consider	consider	VERB
bracis-19080	225	9	,	,	PUNCT
bracis-19080	225	10	our	our	PRON
bracis-19080	225	11	complete	complete	ADJ
bracis-19080	225	12	set	set	NOUN
bracis-19080	225	13	of	of	ADP
bracis-19080	225	14	results	result	NOUN
bracis-19080	225	15	is	be	AUX
bracis-19080	225	16	provided	provide	VERB
bracis-19080	225	17	as	as	ADP
bracis-19080	225	18	supplementary	supplementary	ADJ
bracis-19080	225	19	material	material	NOUN
bracis-19080	225	20	 	 	SPACE
bracis-19080	226	1	[	[	X
bracis-19080	226	2	2	2	NUM
bracis-19080	226	3	]	]	PUNCT
bracis-19080	226	4	.	.	PUNCT
bracis-19080	227	1	in	in	ADP
bracis-19080	227	2	the	the	DET
bracis-19080	227	3	next	next	ADJ
bracis-19080	227	4	sections	section	NOUN
bracis-19080	227	5	,	,	PUNCT
bracis-19080	227	6	we	we	PRON
bracis-19080	227	7	discuss	discuss	VERB
bracis-19080	227	8	the	the	DET
bracis-19080	227	9	most	most	ADV
bracis-19080	227	10	relevant	relevant	ADJ
bracis-19080	227	11	insights	insight	NOUN
bracis-19080	227	12	we	we	PRON
bracis-19080	227	13	observe	observe	VERB
bracis-19080	227	14	in	in	ADP
bracis-19080	227	15	our	our	PRON
bracis-19080	227	16	assessment	assessment	NOUN
bracis-19080	227	17	.	.	PUNCT
bracis-19080	228	1	5	5	NUM
bracis-19080	228	2	results	result	NOUN
bracis-19080	228	3	as	as	SCONJ
bracis-19080	228	4	previously	previously	ADV
bracis-19080	228	5	discussed	discuss	VERB
bracis-19080	228	6	,	,	PUNCT
bracis-19080	228	7	contextual	contextual	ADJ
bracis-19080	228	8	embeddings	embedding	NOUN
bracis-19080	228	9	are	be	AUX
bracis-19080	228	10	usually	usually	ADV
bracis-19080	228	11	produced	produce	VERB
bracis-19080	228	12	from	from	ADP
bracis-19080	228	13	pre	pre	ADJ
bracis-19080	228	14	-	-	ADJ
bracis-19080	228	15	trained	trained	ADJ
bracis-19080	228	16	or	or	CCONJ
bracis-19080	228	17	fine	fine	ADV
bracis-19080	228	18	-	-	PUNCT
bracis-19080	228	19	tuned	tune	VERB
bracis-19080	228	20	language	language	NOUN
bracis-19080	228	21	models	model	NOUN
bracis-19080	228	22	.	.	PUNCT
bracis-19080	229	1	in	in	ADP
bracis-19080	229	2	our	our	PRON
bracis-19080	229	3	setup	setup	NOUN
bracis-19080	229	4	,	,	PUNCT
bracis-19080	229	5	we	we	PRON
bracis-19080	229	6	have	have	AUX
bracis-19080	229	7	isolated	isolate	VERB
bracis-19080	229	8	experiments	experiment	NOUN
bracis-19080	229	9	without	without	ADP
bracis-19080	229	10	and	and	CCONJ
bracis-19080	229	11	with	with	ADP
bracis-19080	229	12	fine	fine	ADV
bracis-19080	229	13	-	-	PUNCT
bracis-19080	229	14	tuning	tuning	NOUN
bracis-19080	229	15	for	for	ADP
bracis-19080	229	16	sts	st	NOUN
bracis-19080	229	17	in	in	ADP
bracis-19080	229	18	portuguese	portuguese	PROPN
bracis-19080	229	19	.	.	PUNCT
bracis-19080	230	1	below	below	ADV
bracis-19080	230	2	,	,	PUNCT
bracis-19080	230	3	we	we	PRON
bracis-19080	230	4	discuss	discuss	VERB
bracis-19080	230	5	the	the	DET
bracis-19080	230	6	most	most	ADV
bracis-19080	230	7	relevant	relevant	ADJ
bracis-19080	230	8	insights	insight	NOUN
bracis-19080	230	9	observed	observe	VERB
bracis-19080	230	10	for	for	ADP
bracis-19080	230	11	each	each	DET
bracis-19080	230	12	part	part	NOUN
bracis-19080	230	13	of	of	ADP
bracis-19080	230	14	our	our	PRON
bracis-19080	230	15	investigation	investigation	NOUN
bracis-19080	230	16	.	.	PUNCT
bracis-19080	231	1	5.1	5.1	NUM
bracis-19080	231	2	pre	pre	ADJ
bracis-19080	231	3	-	-	ADJ
bracis-19080	231	4	trained	train	VERB
bracis-19080	231	5	models	model	NOUN
bracis-19080	231	6	without	without	ADP
bracis-19080	231	7	fine	fine	ADV
bracis-19080	231	8	-	-	PUNCT
bracis-19080	231	9	tuning	tuning	NOUN
bracis-19080	231	10	for	for	ADP
bracis-19080	231	11	 	 	SPACE
bracis-19080	231	12	sts	st	NOUN
bracis-19080	231	13	in	in	ADP
bracis-19080	231	14	 	 	SPACE
bracis-19080	231	15	portuguese	portuguese	NOUN
bracis-19080	231	16	in	in	ADP
bracis-19080	231	17	order	order	NOUN
bracis-19080	231	18	of	of	ADP
bracis-19080	231	19	appearance	appearance	NOUN
bracis-19080	231	20	in	in	ADP
bracis-19080	231	21	table	table	NOUN
bracis-19080	231	22	 	 	SPACE
bracis-19080	231	23	1	1	NUM
bracis-19080	231	24	,	,	PUNCT
bracis-19080	231	25	literature	literature	NOUN
bracis-19080	231	26	results	result	NOUN
bracis-19080	231	27	are	be	AUX
bracis-19080	231	28	given	give	VERB
bracis-19080	231	29	at	at	ADP
bracis-19080	231	30	the	the	DET
bracis-19080	231	31	top	top	NOUN
bracis-19080	231	32	,	,	PUNCT
bracis-19080	231	33	as	as	SCONJ
bracis-19080	231	34	previously	previously	ADV
bracis-19080	231	35	detailed	detail	VERB
bracis-19080	231	36	.	.	PUNCT
bracis-19080	232	1	the	the	DET
bracis-19080	232	2	following	follow	VERB
bracis-19080	232	3	two	two	NUM
bracis-19080	232	4	models	model	NOUN
bracis-19080	232	5	are	be	AUX
bracis-19080	232	6	also	also	ADV
bracis-19080	232	7	taken	take	VERB
bracis-19080	232	8	as	as	ADP
bracis-19080	232	9	baseline	baseline	NOUN
bracis-19080	232	10	.	.	PUNCT
bracis-19080	233	1	in	in	ADP
bracis-19080	233	2	detail	detail	NOUN
bracis-19080	233	3	,	,	PUNCT
bracis-19080	233	4	bertimbau	bertimbau	NOUN
bracis-19080	233	5	-	-	PUNCT
bracis-19080	233	6	base	base	NOUN
bracis-19080	233	7	and	and	CCONJ
bracis-19080	233	8	bertimbau	bertimbau	NOUN
bracis-19080	233	9	-	-	PUNCT
bracis-19080	233	10	large	large	ADJ
bracis-19080	233	11	are	be	AUX
bracis-19080	233	12	the	the	DET
bracis-19080	233	13	bertimbau	bertimbau	NOUN
bracis-19080	233	14	models	model	NOUN
bracis-19080	233	15	that	that	PRON
bracis-19080	233	16	have	have	AUX
bracis-19080	233	17	been	be	AUX
bracis-19080	233	18	pre	pre	VERB
bracis-19080	233	19	-	-	VERB
bracis-19080	233	20	trained	train	VERB
bracis-19080	233	21	in	in	ADP
bracis-19080	233	22	portuguese	portuguese	PROPN
bracis-19080	233	23	.	.	PUNCT
bracis-19080	234	1	remaining	remain	VERB
bracis-19080	234	2	models	model	NOUN
bracis-19080	234	3	are	be	AUX
bracis-19080	234	4	sbert	sbert	NOUN
bracis-19080	234	5	approaches	approach	NOUN
bracis-19080	234	6	.	.	PUNCT
bracis-19080	235	1	given	give	VERB
bracis-19080	235	2	the	the	DET
bracis-19080	235	3	compatibility	compatibility	NOUN
bracis-19080	235	4	between	between	ADP
bracis-19080	235	5	bert	bert	PROPN
bracis-19080	235	6	and	and	CCONJ
bracis-19080	235	7	sbert	sbert	NOUN
bracis-19080	235	8	,	,	PUNCT
bracis-19080	235	9	we	we	PRON
bracis-19080	235	10	include	include	VERB
bracis-19080	235	11	sbert	sbert	NOUN
bracis-19080	235	12	results	result	NOUN
bracis-19080	235	13	where	where	SCONJ
bracis-19080	235	14	weights	weight	NOUN
bracis-19080	235	15	were	be	AUX
bracis-19080	235	16	obtained	obtain	VERB
bracis-19080	235	17	from	from	ADP
bracis-19080	235	18	bertimbau	bertimbau	NOUN
bracis-19080	235	19	models	model	NOUN
bracis-19080	235	20	,	,	PUNCT
bracis-19080	235	21	which	which	PRON
bracis-19080	235	22	we	we	PRON
bracis-19080	235	23	refer	refer	VERB
bracis-19080	235	24	to	to	ADP
bracis-19080	235	25	as	as	ADP
bracis-19080	235	26	 	 	SPACE
bracis-19080	235	27	ptbert	ptbert	NOUN
bracis-19080	235	28	-	-	PUNCT
bracis-19080	235	29	base	base	NOUN
bracis-19080	235	30	 	 	SPACE
bracis-19080	235	31	(	(	PUNCT
bracis-19080	235	32	sbert	sbert	NOUN
bracis-19080	235	33	)	)	PUNCT
bracis-19080	235	34	and	and	CCONJ
bracis-19080	235	35	ptbert	ptbert	NOUN
bracis-19080	235	36	-	-	PUNCT
bracis-19080	235	37	large	large	ADJ
bracis-19080	235	38	 	 	SPACE
bracis-19080	235	39	(	(	PUNCT
bracis-19080	235	40	sbert	sbert	NOUN
bracis-19080	235	41	)	)	PUNCT
bracis-19080	235	42	.	.	PUNCT
bracis-19080	236	1	finally	finally	ADV
bracis-19080	236	2	,	,	PUNCT
bracis-19080	236	3	table	table	NOUN
bracis-19080	236	4	 	 	SPACE
bracis-19080	236	5	1	1	NUM
bracis-19080	236	6	lists	list	VERB
bracis-19080	236	7	the	the	DET
bracis-19080	236	8	multilingual	multilingual	ADJ
bracis-19080	236	9	sbert	sbert	NOUN
bracis-19080	236	10	models	model	NOUN
bracis-19080	236	11	,	,	PUNCT
bracis-19080	236	12	namely	namely	ADV
bracis-19080	236	13	\(\langle	\(\langle	VERB
bracis-19080	236	14	\textsf	\textsf	PROPN
bracis-19080	236	15	{	{	PUNCT
bracis-19080	236	16	xlm	xlm	PROPN
bracis-19080	236	17	-	-	PUNCT
bracis-19080	236	18	r	r	NOUN
bracis-19080	236	19	,	,	PUNCT
bracis-19080	236	20	sbert	sbert	NOUN
bracis-19080	236	21	,	,	PUNCT
bracis-19080	236	22	nli+sts}\rangle	nli+sts}\rangle	PROPN
bracis-19080	236	23	\	\	PROPN
bracis-19080	236	24	)	)	PUNCT
bracis-19080	236	25	,	,	PUNCT
bracis-19080	236	26	\(\langle	\(\langle	PROPN
bracis-19080	236	27	\textsf	\textsf	PROPN
bracis-19080	236	28	{	{	PUNCT
bracis-19080	236	29	xlm	xlm	PROPN
bracis-19080	236	30	-	-	PUNCT
bracis-19080	236	31	r	r	NOUN
bracis-19080	236	32	,	,	PUNCT
bracis-19080	236	33	sbert	sbert	NOUN
bracis-19080	236	34	,	,	PUNCT
bracis-19080	236	35	paraphrases}\rangle	paraphrases}\rangle	NUM
bracis-19080	236	36	\	\	NOUN
bracis-19080	236	37	)	)	PUNCT
bracis-19080	236	38	and	and	CCONJ
bracis-19080	236	39	\(\langle	\(\langle	NUM
bracis-19080	236	40	\textsf	\textsf	PROPN
bracis-19080	236	41	{	{	PUNCT
bracis-19080	236	42	xlm	xlm	PROPN
bracis-19080	236	43	-	-	PUNCT
bracis-19080	236	44	r	r	NOUN
bracis-19080	236	45	,	,	PUNCT
bracis-19080	236	46	sbert	sbert	NOUN
bracis-19080	236	47	,	,	PUNCT
bracis-19080	236	48	cross	cross	ADJ
bracis-19080	236	49	-	-	ADJ
bracis-19080	236	50	en	en	ADJ
bracis-19080	236	51	-	-	PUNCT
bracis-19080	236	52	de}\rangle	de}\rangle	ADJ
bracis-19080	236	53	\	\	NOUN
bracis-19080	236	54	)	)	PUNCT
bracis-19080	236	55	.	.	PUNCT
bracis-19080	237	1	next	next	ADV
bracis-19080	237	2	,	,	PUNCT
bracis-19080	237	3	we	we	PRON
bracis-19080	237	4	discuss	discuss	VERB
bracis-19080	237	5	the	the	DET
bracis-19080	237	6	main	main	ADJ
bracis-19080	237	7	insights	insight	NOUN
bracis-19080	237	8	observed	observe	VERB
bracis-19080	237	9	in	in	ADP
bracis-19080	237	10	this	this	DET
bracis-19080	237	11	part	part	NOUN
bracis-19080	237	12	of	of	ADP
bracis-19080	237	13	our	our	PRON
bracis-19080	237	14	investigation	investigation	NOUN
bracis-19080	237	15	.	.	PUNCT
bracis-19080	238	1	portuguese	portuguese	ADJ
bracis-19080	238	2	pre	pre	ADJ
bracis-19080	238	3	-	-	ADJ
bracis-19080	238	4	trained	train	VERB
bracis-19080	238	5	models	model	NOUN
bracis-19080	238	6	did	do	AUX
bracis-19080	238	7	not	not	PART
bracis-19080	238	8	produce	produce	VERB
bracis-19080	238	9	competitive	competitive	ADJ
bracis-19080	238	10	results	result	NOUN
bracis-19080	238	11	w.r.t	w.r.t	VERB
bracis-19080	238	12	.	.	PUNCT
bracis-19080	239	1	workshop	workshop	NOUN
bracis-19080	239	2	or	or	CCONJ
bracis-19080	239	3	literature	literature	NOUN
bracis-19080	239	4	results	result	NOUN
bracis-19080	239	5	.	.	PUNCT
bracis-19080	240	1	a	a	DET
bracis-19080	240	2	direct	direct	ADJ
bracis-19080	240	3	comparison	comparison	NOUN
bracis-19080	240	4	between	between	ADP
bracis-19080	240	5	models	model	NOUN
bracis-19080	240	6	that	that	PRON
bracis-19080	240	7	use	use	VERB
bracis-19080	240	8	bertimbau	bertimbau	NOUN
bracis-19080	240	9	weights	weight	NOUN
bracis-19080	240	10	showed	show	VERB
bracis-19080	240	11	that	that	SCONJ
bracis-19080	240	12	sbert	sbert	NOUN
bracis-19080	240	13	models	model	NOUN
bracis-19080	240	14	in	in	ADP
bracis-19080	240	15	general	general	ADJ
bracis-19080	240	16	achieved	achieve	VERB
bracis-19080	240	17	better	well	ADJ
bracis-19080	240	18	results	result	NOUN
bracis-19080	240	19	than	than	ADP
bracis-19080	240	20	their	their	PRON
bracis-19080	240	21	bert	bert	PROPN
bracis-19080	240	22	counterparts	counterpart	NOUN
bracis-19080	240	23	.	.	PUNCT
bracis-19080	241	1	this	this	DET
bracis-19080	241	2	result	result	NOUN
bracis-19080	241	3	is	be	AUX
bracis-19080	241	4	interesting	interesting	ADJ
bracis-19080	241	5	considering	consider	VERB
bracis-19080	241	6	that	that	SCONJ
bracis-19080	241	7	the	the	DET
bracis-19080	241	8	only	only	ADJ
bracis-19080	241	9	difference	difference	NOUN
bracis-19080	241	10	between	between	ADP
bracis-19080	241	11	them	they	PRON
bracis-19080	241	12	is	be	AUX
bracis-19080	241	13	that	that	SCONJ
bracis-19080	241	14	embeddings	embedding	NOUN
bracis-19080	241	15	from	from	ADP
bracis-19080	241	16	bert	bert	NOUN
bracis-19080	241	17	models	model	NOUN
bracis-19080	241	18	are	be	AUX
bracis-19080	241	19	obtained	obtain	VERB
bracis-19080	241	20	from	from	ADP
bracis-19080	241	21	the	the	DET
bracis-19080	241	22	[	[	X
bracis-19080	241	23	cls	cls	X
bracis-19080	241	24	]	]	X
bracis-19080	241	25	special	special	NOUN
bracis-19080	241	26	token	token	NOUN
bracis-19080	241	27	,	,	PUNCT
bracis-19080	241	28	whereas	whereas	SCONJ
bracis-19080	241	29	sbert	sbert	NOUN
bracis-19080	241	30	is	be	AUX
bracis-19080	241	31	average	average	ADJ
bracis-19080	241	32	pooling	pool	VERB
bracis-19080	241	33	all	all	DET
bracis-19080	241	34	bert	bert	ADJ
bracis-19080	241	35	output	output	NOUN
bracis-19080	241	36	embeddings	embedding	NOUN
bracis-19080	241	37	.	.	PUNCT
bracis-19080	242	1	another	another	DET
bracis-19080	242	2	important	important	ADJ
bracis-19080	242	3	insight	insight	NOUN
bracis-19080	242	4	is	be	AUX
bracis-19080	242	5	that	that	SCONJ
bracis-19080	242	6	not	not	PART
bracis-19080	242	7	always	always	ADV
bracis-19080	242	8	the	the	DET
bracis-19080	242	9	large	large	ADJ
bracis-19080	242	10	version	version	NOUN
bracis-19080	242	11	outperformed	outperform	VERB
bracis-19080	242	12	its	its	PRON
bracis-19080	242	13	base	base	NOUN
bracis-19080	242	14	counterpart	counterpart	NOUN
bracis-19080	242	15	.	.	PUNCT
bracis-19080	243	1	this	this	PRON
bracis-19080	243	2	is	be	AUX
bracis-19080	243	3	likely	likely	ADV
bracis-19080	243	4	explained	explain	VERB
bracis-19080	243	5	by	by	ADP
bracis-19080	243	6	the	the	DET
bracis-19080	243	7	checkpoints	checkpoint	NOUN
bracis-19080	243	8	used	use	VERB
bracis-19080	243	9	by	by	ADP
bracis-19080	243	10	neuralmind	neuralmind	NOUN
bracis-19080	243	11	for	for	ADP
bracis-19080	243	12	bertimbau	bertimbau	NOUN
bracis-19080	243	13	model	model	NOUN
bracis-19080	243	14	training	training	NOUN
bracis-19080	243	15	.	.	PUNCT
bracis-19080	244	1	as	as	SCONJ
bracis-19080	244	2	discussed	discuss	VERB
bracis-19080	244	3	in	in	ADP
bracis-19080	244	4	sect	sect	NOUN
bracis-19080	244	5	.	.	PUNCT
bracis-19080	244	6	 	 	SPACE
bracis-19080	245	1	4	4	NUM
bracis-19080	245	2	,	,	PUNCT
bracis-19080	245	3	a	a	DET
bracis-19080	245	4	multilingual	multilingual	ADJ
bracis-19080	245	5	bert	bert	NOUN
bracis-19080	245	6	checkpoint	checkpoint	NOUN
bracis-19080	245	7	was	be	AUX
bracis-19080	245	8	used	use	VERB
bracis-19080	245	9	to	to	PART
bracis-19080	245	10	warm	warm	ADJ
bracis-19080	245	11	start	start	NOUN
bracis-19080	245	12	base	base	NOUN
bracis-19080	245	13	training	training	NOUN
bracis-19080	245	14	,	,	PUNCT
bracis-19080	245	15	whereas	whereas	SCONJ
bracis-19080	245	16	large	large	ADJ
bracis-19080	245	17	models	model	NOUN
bracis-19080	245	18	were	be	AUX
bracis-19080	245	19	warm	warm	ADJ
bracis-19080	245	20	-	-	PUNCT
bracis-19080	245	21	started	start	VERB
bracis-19080	245	22	with	with	ADP
bracis-19080	245	23	an	an	DET
bracis-19080	245	24	english	english	ADJ
bracis-19080	245	25	bert	bert	PROPN
bracis-19080	245	26	checkpoint	checkpoint	NOUN
bracis-19080	245	27	.	.	PUNCT
bracis-19080	246	1	multilingual	multilingual	ADJ
bracis-19080	246	2	pre	pre	VERB
bracis-19080	246	3	-	-	ADJ
bracis-19080	246	4	trained	train	VERB
bracis-19080	246	5	models	model	NOUN
bracis-19080	246	6	produced	produce	VERB
bracis-19080	246	7	competitive	competitive	ADJ
bracis-19080	246	8	results	result	NOUN
bracis-19080	246	9	w.r.t	w.r.t	VERB
bracis-19080	246	10	.	.	PUNCT
bracis-19080	247	1	results	result	NOUN
bracis-19080	247	2	from	from	ADP
bracis-19080	247	3	the	the	DET
bracis-19080	247	4	participating	participate	VERB
bracis-19080	247	5	teams	team	NOUN
bracis-19080	247	6	of	of	ADP
bracis-19080	247	7	the	the	DET
bracis-19080	247	8	shared	share	VERB
bracis-19080	247	9	tasks	task	NOUN
bracis-19080	247	10	.	.	PUNCT
bracis-19080	248	1	this	this	PRON
bracis-19080	248	2	is	be	AUX
bracis-19080	248	3	true	true	ADJ
bracis-19080	248	4	for	for	ADP
bracis-19080	248	5	nearly	nearly	ADV
bracis-19080	248	6	all	all	PRON
bracis-19080	248	7	datasets	dataset	NOUN
bracis-19080	248	8	and	and	CCONJ
bracis-19080	248	9	multilingual	multilingual	ADJ
bracis-19080	248	10	models	model	NOUN
bracis-19080	248	11	considered	consider	VERB
bracis-19080	248	12	,	,	PUNCT
bracis-19080	248	13	with	with	ADP
bracis-19080	248	14	two	two	NUM
bracis-19080	248	15	major	major	ADJ
bracis-19080	248	16	exceptions	exception	NOUN
bracis-19080	248	17	.	.	PUNCT
bracis-19080	249	1	first	first	ADV
bracis-19080	249	2	,	,	PUNCT
bracis-19080	249	3	the	the	DET
bracis-19080	249	4	\(\langle	\(\langle	PROPN
bracis-19080	249	5	\textsf	\textsf	PROPN
bracis-19080	249	6	{	{	PUNCT
bracis-19080	249	7	xlm	xlm	PROPN
bracis-19080	249	8	-	-	PUNCT
bracis-19080	249	9	r	r	NOUN
bracis-19080	249	10	,	,	PUNCT
bracis-19080	249	11	sbert	sbert	NOUN
bracis-19080	249	12	,	,	PUNCT
bracis-19080	249	13	nli+sts}\rangle	nli+sts}\rangle	PROPN
bracis-19080	249	14	\	\	NOUN
bracis-19080	249	15	)	)	PUNCT
bracis-19080	249	16	model	model	NOUN
bracis-19080	249	17	is	be	AUX
bracis-19080	249	18	unable	unable	ADJ
bracis-19080	249	19	to	to	PART
bracis-19080	249	20	match	match	VERB
bracis-19080	249	21	workshop	workshop	NOUN
bracis-19080	249	22	results	result	NOUN
bracis-19080	249	23	for	for	ADP
bracis-19080	249	24	any	any	PRON
bracis-19080	249	25	of	of	ADP
bracis-19080	249	26	the	the	DET
bracis-19080	249	27	datasets	dataset	NOUN
bracis-19080	249	28	considered	consider	VERB
bracis-19080	249	29	,	,	PUNCT
bracis-19080	249	30	though	though	SCONJ
bracis-19080	249	31	its	its	PRON
bracis-19080	249	32	teacher	teacher	NOUN
bracis-19080	249	33	-	-	PUNCT
bracis-19080	249	34	model	model	NOUN
bracis-19080	249	35	had	have	AUX
bracis-19080	249	36	been	be	AUX
bracis-19080	249	37	originally	originally	ADV
bracis-19080	249	38	trained	train	VERB
bracis-19080	249	39	for	for	ADP
bracis-19080	249	40	nli	nli	ADJ
bracis-19080	249	41	and	and	CCONJ
bracis-19080	249	42	sts	st	VERB
bracis-19080	249	43	tasks	task	NOUN
bracis-19080	249	44	.	.	PUNCT
bracis-19080	250	1	second	second	ADJ
bracis-19080	250	2	,	,	PUNCT
bracis-19080	250	3	no	no	DET
bracis-19080	250	4	model	model	NOUN
bracis-19080	250	5	is	be	AUX
bracis-19080	250	6	able	able	ADJ
bracis-19080	250	7	to	to	PART
bracis-19080	250	8	match	match	VERB
bracis-19080	250	9	the	the	DET
bracis-19080	250	10	performance	performance	NOUN
bracis-19080	250	11	of	of	ADP
bracis-19080	250	12	assin	assin	PROPN
bracis-19080	250	13	 	 	SPACE
bracis-19080	250	14	2	2	NUM
bracis-19080	250	15	workshop	workshop	NOUN
bracis-19080	250	16	results	result	NOUN
bracis-19080	250	17	.	.	PUNCT
bracis-19080	251	1	among	among	ADP
bracis-19080	251	2	the	the	DET
bracis-19080	251	3	sbert	sbert	NOUN
bracis-19080	251	4	versions	version	NOUN
bracis-19080	251	5	,	,	PUNCT
bracis-19080	251	6	the	the	DET
bracis-19080	251	7	one	one	NOUN
bracis-19080	251	8	that	that	PRON
bracis-19080	251	9	shows	show	VERB
bracis-19080	251	10	the	the	DET
bracis-19080	251	11	most	most	ADV
bracis-19080	251	12	consistent	consistent	ADJ
bracis-19080	251	13	performance	performance	NOUN
bracis-19080	251	14	is	be	AUX
bracis-19080	251	15	the	the	DET
bracis-19080	251	16	one	one	NUM
bracis-19080	251	17	fine	fine	ADV
bracis-19080	251	18	-	-	PUNCT
bracis-19080	251	19	tuned	tune	VERB
bracis-19080	251	20	for	for	ADP
bracis-19080	251	21	cross	cross	ADJ
bracis-19080	251	22	-	-	ADJ
bracis-19080	251	23	en	en	ADJ
bracis-19080	251	24	-	-	PUNCT
bracis-19080	251	25	de	de	X
bracis-19080	251	26	sts	st	NOUN
bracis-19080	251	27	.	.	PUNCT
bracis-19080	252	1	this	this	PRON
bracis-19080	252	2	indicates	indicate	VERB
bracis-19080	252	3	that	that	SCONJ
bracis-19080	252	4	fine	fine	NOUN
bracis-19080	252	5	-	-	PUNCT
bracis-19080	252	6	tuning	tuning	NOUN
bracis-19080	252	7	for	for	ADP
bracis-19080	252	8	multilingual	multilingual	ADJ
bracis-19080	252	9	tasks	task	NOUN
bracis-19080	252	10	is	be	AUX
bracis-19080	252	11	helpful	helpful	ADJ
bracis-19080	252	12	,	,	PUNCT
bracis-19080	252	13	as	as	SCONJ
bracis-19080	252	14	expected	expect	VERB
bracis-19080	252	15	.	.	PUNCT
bracis-19080	253	1	yet	yet	ADV
bracis-19080	253	2	,	,	PUNCT
bracis-19080	253	3	without	without	ADP
bracis-19080	253	4	a	a	DET
bracis-19080	253	5	specific	specific	ADJ
bracis-19080	253	6	fine	fine	ADV
bracis-19080	253	7	-	-	PUNCT
bracis-19080	253	8	tuning	tuning	NOUN
bracis-19080	253	9	for	for	ADP
bracis-19080	253	10	sts	st	NOUN
bracis-19080	253	11	in	in	ADP
bracis-19080	253	12	portuguese	portuguese	ADJ
bracis-19080	253	13	,	,	PUNCT
bracis-19080	253	14	even	even	ADV
bracis-19080	253	15	this	this	DET
bracis-19080	253	16	model	model	NOUN
bracis-19080	253	17	can	can	AUX
bracis-19080	253	18	not	not	PART
bracis-19080	253	19	match	match	VERB
bracis-19080	253	20	state	state	NOUN
bracis-19080	253	21	-	-	PUNCT
bracis-19080	253	22	of	of	ADP
bracis-19080	253	23	-	-	PUNCT
bracis-19080	253	24	the	the	DET
bracis-19080	253	25	-	-	PUNCT
bracis-19080	253	26	art	art	NOUN
bracis-19080	253	27	results	result	NOUN
bracis-19080	253	28	 	 	SPACE
bracis-19080	253	29	[	[	X
bracis-19080	253	30	8	8	NUM
bracis-19080	253	31	]	]	PUNCT
bracis-19080	253	32	.	.	PUNCT
bracis-19080	254	1	indeed	indeed	ADV
bracis-19080	254	2	,	,	PUNCT
bracis-19080	254	3	in	in	ADP
bracis-19080	254	4	this	this	DET
bracis-19080	254	5	part	part	NOUN
bracis-19080	254	6	of	of	ADP
bracis-19080	254	7	the	the	DET
bracis-19080	254	8	investigation	investigation	NOUN
bracis-19080	254	9	all	all	DET
bracis-19080	254	10	sbert	sbert	NOUN
bracis-19080	254	11	results	result	NOUN
bracis-19080	254	12	fall	fall	VERB
bracis-19080	254	13	short	short	ADV
bracis-19080	254	14	of	of	ADP
bracis-19080	254	15	the	the	DET
bracis-19080	254	16	state	state	NOUN
bracis-19080	254	17	-	-	PUNCT
bracis-19080	254	18	of	of	ADP
bracis-19080	254	19	-	-	PUNCT
bracis-19080	254	20	the	the	DET
bracis-19080	254	21	-	-	PUNCT
bracis-19080	254	22	art	art	NOUN
bracis-19080	254	23	for	for	ADP
bracis-19080	254	24	all	all	DET
bracis-19080	254	25	datasets	dataset	NOUN
bracis-19080	254	26	considered	consider	VERB
bracis-19080	254	27	.	.	PUNCT
bracis-19080	255	1	5.2	5.2	NUM
bracis-19080	255	2	pre	pre	ADJ
bracis-19080	255	3	-	-	ADJ
bracis-19080	255	4	trained	train	VERB
bracis-19080	255	5	models	model	NOUN
bracis-19080	255	6	with	with	ADP
bracis-19080	255	7	 	 	SPACE
bracis-19080	255	8	fine	fine	ADV
bracis-19080	255	9	-	-	PUNCT
bracis-19080	255	10	tuning	tune	VERB
bracis-19080	255	11	results	result	NOUN
bracis-19080	255	12	obtained	obtain	VERB
bracis-19080	255	13	with	with	ADP
bracis-19080	255	14	fine	fine	ADV
bracis-19080	255	15	-	-	PUNCT
bracis-19080	255	16	tuned	tune	VERB
bracis-19080	255	17	models	model	NOUN
bracis-19080	255	18	are	be	AUX
bracis-19080	255	19	displayed	display	VERB
bracis-19080	255	20	at	at	ADP
bracis-19080	255	21	the	the	DET
bracis-19080	255	22	bottom	bottom	ADJ
bracis-19080	255	23	part	part	NOUN
bracis-19080	255	24	of	of	ADP
bracis-19080	255	25	table	table	NOUN
bracis-19080	255	26	 	 	SPACE
bracis-19080	255	27	1	1	NUM
bracis-19080	255	28	.	.	PUNCT
bracis-19080	255	29	models	model	NOUN
bracis-19080	255	30	are	be	AUX
bracis-19080	255	31	given	give	VERB
bracis-19080	255	32	in	in	ADP
bracis-19080	255	33	the	the	DET
bracis-19080	255	34	same	same	ADJ
bracis-19080	255	35	order	order	NOUN
bracis-19080	255	36	as	as	ADP
bracis-19080	255	37	the	the	DET
bracis-19080	255	38	pre	pre	ADJ
bracis-19080	255	39	-	-	ADJ
bracis-19080	255	40	trained	train	VERB
bracis-19080	255	41	models	model	NOUN
bracis-19080	255	42	.	.	PUNCT
bracis-19080	256	1	below	below	ADV
bracis-19080	256	2	,	,	PUNCT
bracis-19080	256	3	we	we	PRON
bracis-19080	256	4	discuss	discuss	VERB
bracis-19080	256	5	the	the	DET
bracis-19080	256	6	main	main	ADJ
bracis-19080	256	7	insights	insight	NOUN
bracis-19080	256	8	from	from	ADP
bracis-19080	256	9	this	this	DET
bracis-19080	256	10	analysis	analysis	NOUN
bracis-19080	256	11	.	.	PUNCT
bracis-19080	257	1	benefits	benefit	NOUN
bracis-19080	257	2	of	of	ADP
bracis-19080	257	3	fine	fine	ADV
bracis-19080	257	4	-	-	PUNCT
bracis-19080	257	5	tuning	tuning	NOUN
bracis-19080	257	6	.	.	PUNCT
bracis-19080	258	1	the	the	DET
bracis-19080	258	2	performance	performance	NOUN
bracis-19080	258	3	gains	gain	NOUN
bracis-19080	258	4	provided	provide	VERB
bracis-19080	258	5	by	by	ADP
bracis-19080	258	6	fine	fine	ADJ
bracis-19080	258	7	-	-	PUNCT
bracis-19080	258	8	tuning	tuning	NOUN
bracis-19080	258	9	are	be	AUX
bracis-19080	258	10	considerable	considerable	ADJ
bracis-19080	258	11	for	for	ADP
bracis-19080	258	12	all	all	DET
bracis-19080	258	13	algorithms	algorithm	NOUN
bracis-19080	258	14	.	.	PUNCT
bracis-19080	259	1	in	in	ADP
bracis-19080	259	2	fact	fact	NOUN
bracis-19080	259	3	,	,	PUNCT
bracis-19080	259	4	all	all	DET
bracis-19080	259	5	models	model	NOUN
bracis-19080	259	6	now	now	ADV
bracis-19080	259	7	(	(	PUNCT
bracis-19080	259	8	numerically	numerically	ADV
bracis-19080	259	9	)	)	PUNCT
bracis-19080	259	10	outperform	outperform	VERB
bracis-19080	259	11	the	the	DET
bracis-19080	259	12	best	good	ADJ
bracis-19080	259	13	results	result	NOUN
bracis-19080	259	14	achieved	achieve	VERB
bracis-19080	259	15	during	during	ADP
bracis-19080	259	16	the	the	DET
bracis-19080	259	17	assin	assin	PROPN
bracis-19080	259	18	and	and	CCONJ
bracis-19080	259	19	assin	assin	PROPN
bracis-19080	259	20	 	 	SPACE
bracis-19080	259	21	2	2	NUM
bracis-19080	259	22	tasks	task	NOUN
bracis-19080	259	23	.	.	PUNCT
bracis-19080	260	1	to	to	PART
bracis-19080	260	2	evidence	evidence	VERB
bracis-19080	260	3	the	the	DET
bracis-19080	260	4	performance	performance	NOUN
bracis-19080	260	5	gains	gain	NOUN
bracis-19080	260	6	provided	provide	VERB
bracis-19080	260	7	by	by	ADP
bracis-19080	260	8	fine	fine	ADJ
bracis-19080	260	9	-	-	PUNCT
bracis-19080	260	10	tuning	tuning	NOUN
bracis-19080	260	11	,	,	PUNCT
bracis-19080	260	12	the	the	DET
bracis-19080	260	13	boxplots	boxplot	NOUN
bracis-19080	260	14	in	in	ADP
bracis-19080	260	15	fig	fig	NOUN
bracis-19080	260	16	.	.	PUNCT
bracis-19080	260	17	 	 	SPACE
bracis-19080	261	1	3	3	NUM
bracis-19080	261	2	show	show	VERB
bracis-19080	261	3	the	the	DET
bracis-19080	261	4	average	average	ADJ
bracis-19080	261	5	performance	performance	NOUN
bracis-19080	261	6	of	of	ADP
bracis-19080	261	7	all	all	DET
bracis-19080	261	8	sbert	sbert	NOUN
bracis-19080	261	9	models	model	NOUN
bracis-19080	261	10	with	with	ADP
bracis-19080	261	11	and	and	CCONJ
bracis-19080	261	12	without	without	ADP
bracis-19080	261	13	fine	fine	ADV
bracis-19080	261	14	-	-	PUNCT
bracis-19080	261	15	tuning	tuning	NOUN
bracis-19080	261	16	,	,	PUNCT
bracis-19080	261	17	grouped	group	VERB
bracis-19080	261	18	by	by	ADP
bracis-19080	261	19	dataset	dataset	NOUN
bracis-19080	261	20	.	.	PUNCT
bracis-19080	262	1	in	in	ADP
bracis-19080	262	2	addition	addition	NOUN
bracis-19080	262	3	to	to	ADP
bracis-19080	262	4	improving	improve	VERB
bracis-19080	262	5	average	average	ADJ
bracis-19080	262	6	performance	performance	NOUN
bracis-19080	262	7	,	,	PUNCT
bracis-19080	262	8	fine	fine	ADV
bracis-19080	262	9	-	-	PUNCT
bracis-19080	262	10	tuning	tuning	NOUN
bracis-19080	262	11	also	also	ADV
bracis-19080	262	12	reduces	reduce	VERB
bracis-19080	262	13	variability	variability	NOUN
bracis-19080	262	14	to	to	ADP
bracis-19080	262	15	a	a	DET
bracis-19080	262	16	significant	significant	ADJ
bracis-19080	262	17	extent	extent	NOUN
bracis-19080	262	18	,	,	PUNCT
bracis-19080	262	19	especially	especially	ADV
bracis-19080	262	20	on	on	ADP
bracis-19080	262	21	assin	assin	PROPN
bracis-19080	262	22	pt	pt	PROPN
bracis-19080	262	23	-	-	PUNCT
bracis-19080	262	24	pt	pt	PROPN
bracis-19080	262	25	.	.	PROPN
bracis-19080	262	26	comparison	comparison	NOUN
bracis-19080	262	27	with	with	ADP
bracis-19080	262	28	the	the	DET
bracis-19080	262	29	state	state	NOUN
bracis-19080	262	30	-	-	PUNCT
bracis-19080	262	31	of	of	ADP
bracis-19080	262	32	-	-	PUNCT
bracis-19080	262	33	the	the	DET
bracis-19080	262	34	-	-	PUNCT
bracis-19080	262	35	art	art	NOUN
bracis-19080	262	36	.	.	PUNCT
bracis-19080	263	1	results	result	NOUN
bracis-19080	263	2	differ	differ	VERB
bracis-19080	263	3	as	as	ADP
bracis-19080	263	4	a	a	DET
bracis-19080	263	5	function	function	NOUN
bracis-19080	263	6	of	of	ADP
bracis-19080	263	7	shared	share	VERB
bracis-19080	263	8	task	task	NOUN
bracis-19080	263	9	.	.	PUNCT
bracis-19080	264	1	multilingual	multilingual	ADJ
bracis-19080	264	2	sbert	sbert	NOUN
bracis-19080	264	3	models	model	NOUN
bracis-19080	264	4	numerically	numerically	ADV
bracis-19080	264	5	surpass	surpass	VERB
bracis-19080	264	6	the	the	DET
bracis-19080	264	7	state	state	NOUN
bracis-19080	264	8	-	-	PUNCT
bracis-19080	264	9	of	of	ADP
bracis-19080	264	10	-	-	PUNCT
bracis-19080	264	11	the	the	DET
bracis-19080	264	12	-	-	PUNCT
bracis-19080	264	13	art	art	NOUN
bracis-19080	264	14	results	result	NOUN
bracis-19080	264	15	according	accord	VERB
bracis-19080	264	16	to	to	ADP
bracis-19080	264	17	pearson	pearson	PROPN
bracis-19080	264	18	correlation	correlation	NOUN
bracis-19080	264	19	and/or	and/or	CCONJ
bracis-19080	264	20	mse	mse	PROPN
bracis-19080	264	21	on	on	ADP
bracis-19080	264	22	assin	assin	PROPN
bracis-19080	264	23	 	 	SPACE
bracis-19080	264	24	2	2	NUM
bracis-19080	264	25	.	.	PUNCT
bracis-19080	264	26	remaining	remain	VERB
bracis-19080	264	27	models	model	NOUN
bracis-19080	264	28	also	also	ADV
bracis-19080	264	29	get	get	VERB
bracis-19080	264	30	closer	close	ADJ
bracis-19080	264	31	to	to	ADP
bracis-19080	264	32	the	the	DET
bracis-19080	264	33	state	state	NOUN
bracis-19080	264	34	-	-	PUNCT
bracis-19080	264	35	of	of	ADP
bracis-19080	264	36	-	-	PUNCT
bracis-19080	264	37	the	the	DET
bracis-19080	264	38	-	-	PUNCT
bracis-19080	264	39	art	art	NOUN
bracis-19080	264	40	for	for	ADP
bracis-19080	264	41	this	this	DET
bracis-19080	264	42	task	task	NOUN
bracis-19080	264	43	,	,	PUNCT
bracis-19080	264	44	confirming	confirm	VERB
bracis-19080	264	45	the	the	DET
bracis-19080	264	46	expected	expect	VERB
bracis-19080	264	47	benefits	benefit	NOUN
bracis-19080	264	48	from	from	ADP
bracis-19080	264	49	sbert	sbert	NOUN
bracis-19080	264	50	and	and	CCONJ
bracis-19080	264	51	fine	fine	ADV
bracis-19080	264	52	-	-	PUNCT
bracis-19080	264	53	tuning	tuning	NOUN
bracis-19080	264	54	.	.	PUNCT
bracis-19080	265	1	for	for	ADP
bracis-19080	265	2	assin	assin	PROPN
bracis-19080	265	3	pt	pt	PROPN
bracis-19080	265	4	-	-	PUNCT
bracis-19080	265	5	br	br	PROPN
bracis-19080	265	6	and	and	CCONJ
bracis-19080	265	7	assin	assin	PROPN
bracis-19080	265	8	pt	pt	PROPN
bracis-19080	265	9	-	-	PUNCT
bracis-19080	265	10	pt	pt	NOUN
bracis-19080	265	11	,	,	PUNCT
bracis-19080	265	12	the	the	DET
bracis-19080	265	13	best	good	ADJ
bracis-19080	265	14	results	result	NOUN
bracis-19080	265	15	showed	show	VERB
bracis-19080	265	16	by	by	ADP
bracis-19080	265	17	sbert	sbert	NOUN
bracis-19080	265	18	models	model	NOUN
bracis-19080	265	19	are	be	AUX
bracis-19080	265	20	not	not	PART
bracis-19080	265	21	yet	yet	ADV
bracis-19080	265	22	competitive	competitive	ADJ
bracis-19080	265	23	.	.	PUNCT
bracis-19080	266	1	homogeneity	homogeneity	NOUN
bracis-19080	266	2	of	of	ADP
bracis-19080	266	3	results	result	NOUN
bracis-19080	266	4	.	.	PUNCT
bracis-19080	267	1	in	in	ADP
bracis-19080	267	2	the	the	DET
bracis-19080	267	3	absence	absence	NOUN
bracis-19080	267	4	of	of	ADP
bracis-19080	267	5	fine	fine	ADV
bracis-19080	267	6	-	-	PUNCT
bracis-19080	267	7	tuning	tuning	NOUN
bracis-19080	267	8	(	(	PUNCT
bracis-19080	267	9	pre	pre	ADJ
bracis-19080	267	10	-	-	ADJ
bracis-19080	267	11	trained	train	VERB
bracis-19080	267	12	section	section	NOUN
bracis-19080	267	13	in	in	ADP
bracis-19080	267	14	table	table	NOUN
bracis-19080	267	15	 	 	SPACE
bracis-19080	267	16	1	1	NUM
bracis-19080	267	17	)	)	PUNCT
bracis-19080	267	18	results	result	NOUN
bracis-19080	267	19	were	be	AUX
bracis-19080	267	20	very	very	ADV
bracis-19080	267	21	contrasting	contrasting	ADJ
bracis-19080	267	22	among	among	ADP
bracis-19080	267	23	different	different	ADJ
bracis-19080	267	24	models	model	NOUN
bracis-19080	267	25	.	.	PUNCT
bracis-19080	268	1	conversely	conversely	ADV
bracis-19080	268	2	,	,	PUNCT
bracis-19080	268	3	all	all	DET
bracis-19080	268	4	fine	fine	ADJ
bracis-19080	268	5	-	-	PUNCT
bracis-19080	268	6	tuned	tune	VERB
bracis-19080	268	7	models	model	NOUN
bracis-19080	268	8	have	have	AUX
bracis-19080	268	9	produced	produce	VERB
bracis-19080	268	10	results	result	NOUN
bracis-19080	268	11	that	that	PRON
bracis-19080	268	12	are	be	AUX
bracis-19080	268	13	much	much	ADV
bracis-19080	268	14	closer	close	ADJ
bracis-19080	268	15	to	to	ADP
bracis-19080	268	16	each	each	DET
bracis-19080	268	17	other	other	ADJ
bracis-19080	268	18	.	.	PUNCT
bracis-19080	269	1	we	we	PRON
bracis-19080	269	2	illustrate	illustrate	VERB
bracis-19080	269	3	this	this	DET
bracis-19080	269	4	increased	increase	VERB
bracis-19080	269	5	homogeneity	homogeneity	NOUN
bracis-19080	269	6	for	for	ADP
bracis-19080	269	7	assin	assin	PROPN
bracis-19080	269	8	 	 	SPACE
bracis-19080	269	9	2	2	NUM
bracis-19080	269	10	results	result	NOUN
bracis-19080	269	11	in	in	ADP
bracis-19080	269	12	fig	fig	NOUN
bracis-19080	269	13	.	.	PUNCT
bracis-19080	269	14	 	 	SPACE
bracis-19080	269	15	4	4	NUM
bracis-19080	269	16	,	,	PUNCT
bracis-19080	269	17	where	where	SCONJ
bracis-19080	269	18	average	average	ADJ
bracis-19080	269	19	pre	pre	ADJ
bracis-19080	269	20	-	-	ADJ
bracis-19080	269	21	trained	train	VERB
bracis-19080	269	22	model	model	NOUN
bracis-19080	269	23	performance	performance	NOUN
bracis-19080	269	24	is	be	AUX
bracis-19080	269	25	given	give	VERB
bracis-19080	269	26	in	in	ADP
bracis-19080	269	27	red	red	ADJ
bracis-19080	269	28	and	and	CCONJ
bracis-19080	269	29	average	average	ADJ
bracis-19080	269	30	fine	fine	ADV
bracis-19080	269	31	-	-	PUNCT
bracis-19080	269	32	tuned	tune	VERB
bracis-19080	269	33	model	model	NOUN
bracis-19080	269	34	performance	performance	NOUN
bracis-19080	269	35	is	be	AUX
bracis-19080	269	36	given	give	VERB
bracis-19080	269	37	in	in	ADP
bracis-19080	269	38	blue	blue	ADJ
bracis-19080	269	39	,	,	PUNCT
bracis-19080	269	40	grouped	group	VERB
bracis-19080	269	41	by	by	ADP
bracis-19080	269	42	model	model	NOUN
bracis-19080	269	43	and	and	CCONJ
bracis-19080	269	44	sorted	sort	VERB
bracis-19080	269	45	by	by	ADP
bracis-19080	269	46	fine	fine	ADV
bracis-19080	269	47	-	-	PUNCT
bracis-19080	269	48	tuned	tune	VERB
bracis-19080	269	49	performance	performance	NOUN
bracis-19080	269	50	.	.	PUNCT
bracis-19080	270	1	for	for	ADP
bracis-19080	270	2	this	this	DET
bracis-19080	270	3	dataset	dataset	NOUN
bracis-19080	270	4	,	,	PUNCT
bracis-19080	270	5	even	even	ADV
bracis-19080	270	6	bertimbau	bertimbau	NOUN
bracis-19080	270	7	models	model	NOUN
bracis-19080	270	8	lead	lead	VERB
bracis-19080	270	9	to	to	ADP
bracis-19080	270	10	a	a	DET
bracis-19080	270	11	competitive	competitive	ADJ
bracis-19080	270	12	performance	performance	NOUN
bracis-19080	270	13	after	after	ADP
bracis-19080	270	14	fine	fine	ADV
bracis-19080	270	15	-	-	PUNCT
bracis-19080	270	16	tuning	tuning	NOUN
bracis-19080	270	17	.	.	PUNCT
bracis-19080	271	1	fig	fig	NOUN
bracis-19080	271	2	.	.	PUNCT
bracis-19080	272	1	3	3	X
bracis-19080	272	2	.	.	X
bracis-19080	272	3	performances	performance	NOUN
bracis-19080	272	4	of	of	ADP
bracis-19080	272	5	pre	pre	ADJ
bracis-19080	272	6	-	-	ADJ
bracis-19080	272	7	trained	train	VERB
bracis-19080	272	8	models	model	NOUN
bracis-19080	272	9	with	with	ADP
bracis-19080	272	10	and	and	CCONJ
bracis-19080	272	11	without	without	ADP
bracis-19080	272	12	fine	fine	ADV
bracis-19080	272	13	-	-	PUNCT
bracis-19080	272	14	tuning	tuning	NOUN
bracis-19080	272	15	.	.	PUNCT
bracis-19080	273	1	full	full	ADJ
bracis-19080	273	2	size	size	NOUN
bracis-19080	273	3	image	image	NOUN
bracis-19080	273	4	fig	fig	NOUN
bracis-19080	273	5	.	.	PUNCT
bracis-19080	274	1	4	4	X
bracis-19080	274	2	.	.	X
bracis-19080	274	3	homogeneity	homogeneity	NOUN
bracis-19080	274	4	of	of	ADP
bracis-19080	274	5	results	result	NOUN
bracis-19080	274	6	with	with	ADP
bracis-19080	274	7	 	 	SPACE
bracis-19080	274	8	(	(	PUNCT
bracis-19080	274	9	blue	blue	ADJ
bracis-19080	274	10	)	)	PUNCT
bracis-19080	274	11	and	and	CCONJ
bracis-19080	274	12	without	without	ADP
bracis-19080	274	13	 	 	SPACE
bracis-19080	274	14	(	(	PUNCT
bracis-19080	274	15	red	red	ADJ
bracis-19080	274	16	)	)	PUNCT
bracis-19080	274	17	fine	fine	ADJ
bracis-19080	274	18	-	-	PUNCT
bracis-19080	274	19	tuning	tuning	NOUN
bracis-19080	274	20	(	(	PUNCT
bracis-19080	274	21	color	color	NOUN
bracis-19080	274	22	figure	figure	NOUN
bracis-19080	274	23	olnine	olnine	ADJ
bracis-19080	274	24	)	)	PUNCT
bracis-19080	274	25	full	full	ADJ
bracis-19080	274	26	size	size	NOUN
bracis-19080	274	27	image	image	NOUN
bracis-19080	274	28	overall	overall	ADV
bracis-19080	274	29	,	,	PUNCT
bracis-19080	274	30	results	result	NOUN
bracis-19080	274	31	discussed	discuss	VERB
bracis-19080	274	32	in	in	ADP
bracis-19080	274	33	this	this	DET
bracis-19080	274	34	section	section	NOUN
bracis-19080	274	35	confirmed	confirm	VERB
bracis-19080	274	36	the	the	DET
bracis-19080	274	37	benefits	benefit	NOUN
bracis-19080	274	38	of	of	ADP
bracis-19080	274	39	adopting	adopt	VERB
bracis-19080	274	40	sbert	sbert	NOUN
bracis-19080	274	41	and	and	CCONJ
bracis-19080	274	42	fine	fine	ADV
bracis-19080	274	43	-	-	PUNCT
bracis-19080	274	44	tuning	tuning	NOUN
bracis-19080	274	45	for	for	ADP
bracis-19080	274	46	sts	st	NOUN
bracis-19080	274	47	in	in	ADP
bracis-19080	274	48	portuguese	portuguese	PROPN
bracis-19080	274	49	.	.	PUNCT
bracis-19080	275	1	for	for	ADP
bracis-19080	275	2	the	the	DET
bracis-19080	275	3	assin	assin	PROPN
bracis-19080	275	4	 	 	SPACE
bracis-19080	275	5	2	2	NUM
bracis-19080	275	6	dataset	dataset	NOUN
bracis-19080	275	7	,	,	PUNCT
bracis-19080	275	8	multilingual	multilingual	ADJ
bracis-19080	275	9	models	model	NOUN
bracis-19080	275	10	even	even	ADV
bracis-19080	275	11	numerically	numerically	ADV
bracis-19080	275	12	surpass	surpass	VERB
bracis-19080	275	13	the	the	DET
bracis-19080	275	14	current	current	ADJ
bracis-19080	275	15	state	state	NOUN
bracis-19080	275	16	-	-	PUNCT
bracis-19080	275	17	of	of	ADP
bracis-19080	275	18	-	-	PUNCT
bracis-19080	275	19	the	the	DET
bracis-19080	275	20	-	-	PUNCT
bracis-19080	275	21	art	art	NOUN
bracis-19080	275	22	.	.	PUNCT
bracis-19080	276	1	in	in	ADP
bracis-19080	276	2	the	the	DET
bracis-19080	276	3	next	next	ADJ
bracis-19080	276	4	section	section	NOUN
bracis-19080	276	5	,	,	PUNCT
bracis-19080	276	6	we	we	PRON
bracis-19080	276	7	further	far	ADV
bracis-19080	276	8	investigate	investigate	VERB
bracis-19080	276	9	results	result	NOUN
bracis-19080	276	10	to	to	PART
bracis-19080	276	11	understand	understand	VERB
bracis-19080	276	12	these	these	DET
bracis-19080	276	13	benefits	benefit	NOUN
bracis-19080	276	14	.	.	PUNCT
bracis-19080	277	1	6	6	NUM
bracis-19080	277	2	further	further	ADJ
bracis-19080	277	3	analysis	analysis	NOUN
bracis-19080	277	4	to	to	PART
bracis-19080	277	5	better	well	ADV
bracis-19080	277	6	understand	understand	VERB
bracis-19080	277	7	some	some	PRON
bracis-19080	277	8	of	of	ADP
bracis-19080	277	9	the	the	DET
bracis-19080	277	10	insights	insight	NOUN
bracis-19080	277	11	observed	observe	VERB
bracis-19080	277	12	in	in	ADP
bracis-19080	277	13	the	the	DET
bracis-19080	277	14	previous	previous	ADJ
bracis-19080	277	15	section	section	NOUN
bracis-19080	277	16	,	,	PUNCT
bracis-19080	277	17	we	we	PRON
bracis-19080	277	18	initially	initially	ADV
bracis-19080	277	19	conduct	conduct	VERB
bracis-19080	277	20	a	a	DET
bracis-19080	277	21	qualitative	qualitative	ADJ
bracis-19080	277	22	assessment	assessment	NOUN
bracis-19080	277	23	,	,	PUNCT
bracis-19080	277	24	regarding	regard	VERB
bracis-19080	277	25	both	both	DET
bracis-19080	277	26	(	(	PUNCT
bracis-19080	277	27	i	i	NOUN
bracis-19080	277	28	)	)	PUNCT
bracis-19080	277	29	 	 	SPACE
bracis-19080	277	30	fine	fine	ADV
bracis-19080	277	31	-	-	PUNCT
bracis-19080	277	32	tuning	tuning	NOUN
bracis-19080	277	33	and	and	CCONJ
bracis-19080	277	34	(	(	PUNCT
bracis-19080	277	35	ii	ii	NOUN
bracis-19080	277	36	)	)	PUNCT
bracis-19080	277	37	 	 	SPACE
bracis-19080	277	38	language	language	NOUN
bracis-19080	277	39	variant	variant	NOUN
bracis-19080	277	40	effects	effect	NOUN
bracis-19080	277	41	.	.	PUNCT
bracis-19080	278	1	moreover	moreover	ADV
bracis-19080	278	2	,	,	PUNCT
bracis-19080	278	3	we	we	PRON
bracis-19080	278	4	run	run	VERB
bracis-19080	278	5	additional	additional	ADJ
bracis-19080	278	6	experiments	experiment	NOUN
bracis-19080	278	7	to	to	PART
bracis-19080	278	8	assess	assess	VERB
bracis-19080	278	9	the	the	DET
bracis-19080	278	10	benefits	benefit	NOUN
bracis-19080	278	11	of	of	ADP
bracis-19080	278	12	multiple	multiple	ADJ
bracis-19080	278	13	language	language	NOUN
bracis-19080	278	14	variants	variant	NOUN
bracis-19080	278	15	in	in	ADP
bracis-19080	278	16	the	the	DET
bracis-19080	278	17	fine	fine	ADJ
bracis-19080	278	18	-	-	PUNCT
bracis-19080	278	19	tuning	tuning	NOUN
bracis-19080	278	20	setup	setup	NOUN
bracis-19080	278	21	.	.	PUNCT
bracis-19080	279	1	table	table	NOUN
bracis-19080	279	2	2	2	NUM
bracis-19080	279	3	.	.	PUNCT
bracis-19080	280	1	assin	assin	NOUN
bracis-19080	280	2	 	 	SPACE
bracis-19080	280	3	2	2	NUM
bracis-19080	280	4	sentence	sentence	NOUN
bracis-19080	280	5	pair	pair	NOUN
bracis-19080	280	6	examples	example	NOUN
bracis-19080	280	7	with	with	ADP
bracis-19080	280	8	their	their	PRON
bracis-19080	280	9	scores	score	NOUN
bracis-19080	280	10	illustrating	illustrate	VERB
bracis-19080	280	11	changes	change	NOUN
bracis-19080	280	12	in	in	ADP
bracis-19080	280	13	cosine	cosine	NOUN
bracis-19080	280	14	similarity	similarity	NOUN
bracis-19080	280	15	before	before	ADV
bracis-19080	280	16	and	and	CCONJ
bracis-19080	280	17	after	after	ADP
bracis-19080	280	18	fine	fine	ADV
bracis-19080	280	19	-	-	PUNCT
bracis-19080	280	20	tuning	tune	VERB
bracis-19080	280	21	\(\langle	\(\langle	NOUN
bracis-19080	280	22	\textsf	\textsf	PROPN
bracis-19080	280	23	{	{	PUNCT
bracis-19080	280	24	xlm	xlm	PROPN
bracis-19080	280	25	-	-	PUNCT
bracis-19080	280	26	r	r	NOUN
bracis-19080	280	27	,	,	PUNCT
bracis-19080	280	28	sbert	sbert	NOUN
bracis-19080	280	29	,	,	PUNCT
bracis-19080	280	30	paraphrases}\rangle	paraphrases}\rangle	NUM
bracis-19080	280	31	\).full	\).full	NOUN
bracis-19080	280	32	size	size	NOUN
bracis-19080	280	33	table	table	NOUN
bracis-19080	280	34	6.1	6.1	NUM
bracis-19080	280	35	fine	fine	ADV
bracis-19080	280	36	-	-	PUNCT
bracis-19080	280	37	tuning	tuning	NOUN
bracis-19080	280	38	effects	effect	NOUN
bracis-19080	280	39	we	we	PRON
bracis-19080	280	40	select	select	VERB
bracis-19080	280	41	a	a	DET
bracis-19080	280	42	few	few	ADJ
bracis-19080	280	43	sentence	sentence	NOUN
bracis-19080	280	44	pairs	pair	NOUN
bracis-19080	280	45	to	to	PART
bracis-19080	280	46	illustrate	illustrate	VERB
bracis-19080	280	47	fine	fine	ADV
bracis-19080	280	48	-	-	PUNCT
bracis-19080	280	49	tuning	tuning	NOUN
bracis-19080	280	50	effects	effect	NOUN
bracis-19080	280	51	on	on	ADP
bracis-19080	280	52	\(\langle	\(\langle	PROPN
bracis-19080	280	53	\textsf	\textsf	PROPN
bracis-19080	280	54	{	{	PUNCT
bracis-19080	280	55	xlm	xlm	PROPN
bracis-19080	280	56	-	-	PUNCT
bracis-19080	280	57	r	r	NOUN
bracis-19080	280	58	,	,	PUNCT
bracis-19080	280	59	sbert	sbert	NOUN
bracis-19080	280	60	,	,	PUNCT
bracis-19080	280	61	paraphrases}\rangle	paraphrases}\rangle	NUM
bracis-19080	280	62	\	\	NOUN
bracis-19080	280	63	)	)	PUNCT
bracis-19080	280	64	results	result	NOUN
bracis-19080	280	65	,	,	PUNCT
bracis-19080	280	66	the	the	DET
bracis-19080	280	67	sbert	sbert	NOUN
bracis-19080	280	68	model	model	NOUN
bracis-19080	280	69	that	that	PRON
bracis-19080	280	70	performed	perform	VERB
bracis-19080	280	71	best	well	ADV
bracis-19080	280	72	in	in	ADP
bracis-19080	280	73	our	our	PRON
bracis-19080	280	74	experiments	experiment	NOUN
bracis-19080	280	75	.	.	PUNCT
bracis-19080	281	1	table	table	NOUN
bracis-19080	281	2	 	 	SPACE
bracis-19080	281	3	2	2	NUM
bracis-19080	281	4	gives	give	VERB
bracis-19080	281	5	six	six	NUM
bracis-19080	281	6	sentence	sentence	NOUN
bracis-19080	281	7	pair	pair	NOUN
bracis-19080	281	8	examples	example	NOUN
bracis-19080	281	9	with	with	ADP
bracis-19080	281	10	their	their	PRON
bracis-19080	281	11	expected	expect	VERB
bracis-19080	281	12	scores	score	NOUN
bracis-19080	281	13	and	and	CCONJ
bracis-19080	281	14	cosine	cosine	NOUN
bracis-19080	281	15	similarities	similarity	NOUN
bracis-19080	281	16	before	before	ADV
bracis-19080	281	17	and	and	CCONJ
bracis-19080	281	18	after	after	ADP
bracis-19080	281	19	fine	fine	ADV
bracis-19080	281	20	-	-	PUNCT
bracis-19080	281	21	tuning	tuning	NOUN
bracis-19080	281	22	for	for	ADP
bracis-19080	281	23	sts	st	NOUN
bracis-19080	281	24	in	in	ADP
bracis-19080	281	25	portuguese	portuguese	PROPN
bracis-19080	281	26	.	.	PUNCT
bracis-19080	282	1	for	for	ADP
bracis-19080	282	2	the	the	DET
bracis-19080	282	3	first	first	ADJ
bracis-19080	282	4	three	three	NUM
bracis-19080	282	5	pairs	pair	NOUN
bracis-19080	282	6	,	,	PUNCT
bracis-19080	282	7	results	result	NOUN
bracis-19080	282	8	were	be	AUX
bracis-19080	282	9	improved	improve	VERB
bracis-19080	282	10	by	by	ADP
bracis-19080	282	11	fine	fine	ADV
bracis-19080	282	12	-	-	PUNCT
bracis-19080	282	13	tuning	tuning	NOUN
bracis-19080	282	14	.	.	PUNCT
bracis-19080	283	1	indeed	indeed	ADV
bracis-19080	283	2	,	,	PUNCT
bracis-19080	283	3	fine	fine	ADV
bracis-19080	283	4	-	-	PUNCT
bracis-19080	283	5	tuned	tune	VERB
bracis-19080	283	6	results	result	NOUN
bracis-19080	283	7	become	become	VERB
bracis-19080	283	8	very	very	ADV
bracis-19080	283	9	close	close	ADJ
bracis-19080	283	10	to	to	ADP
bracis-19080	283	11	the	the	DET
bracis-19080	283	12	expected	expect	VERB
bracis-19080	283	13	score	score	NOUN
bracis-19080	283	14	,	,	PUNCT
bracis-19080	283	15	whereas	whereas	SCONJ
bracis-19080	283	16	pre	pre	ADJ
bracis-19080	283	17	-	-	ADJ
bracis-19080	283	18	trained	train	VERB
bracis-19080	283	19	results	result	NOUN
bracis-19080	283	20	were	be	AUX
bracis-19080	283	21	very	very	ADV
bracis-19080	283	22	off	off	ADV
bracis-19080	283	23	.	.	PUNCT
bracis-19080	284	1	by	by	ADP
bracis-19080	284	2	contrast	contrast	NOUN
bracis-19080	284	3	,	,	PUNCT
bracis-19080	284	4	results	result	NOUN
bracis-19080	284	5	obtained	obtain	VERB
bracis-19080	284	6	with	with	ADP
bracis-19080	284	7	fine	fine	ADV
bracis-19080	284	8	-	-	PUNCT
bracis-19080	284	9	tuning	tuning	NOUN
bracis-19080	284	10	for	for	ADP
bracis-19080	284	11	the	the	DET
bracis-19080	284	12	latter	latter	ADJ
bracis-19080	284	13	three	three	NUM
bracis-19080	284	14	sentence	sentence	NOUN
bracis-19080	284	15	pairs	pair	NOUN
bracis-19080	284	16	in	in	ADP
bracis-19080	284	17	table	table	NOUN
bracis-19080	284	18	 	 	SPACE
bracis-19080	284	19	2	2	NUM
bracis-19080	284	20	are	be	AUX
bracis-19080	284	21	worse	bad	ADJ
bracis-19080	284	22	than	than	ADP
bracis-19080	284	23	when	when	SCONJ
bracis-19080	284	24	fine	fine	ADJ
bracis-19080	284	25	-	-	PUNCT
bracis-19080	284	26	tuning	tuning	NOUN
bracis-19080	284	27	is	be	AUX
bracis-19080	284	28	not	not	PART
bracis-19080	284	29	adopted	adopt	VERB
bracis-19080	284	30	.	.	PUNCT
bracis-19080	285	1	in	in	ADP
bracis-19080	285	2	the	the	DET
bracis-19080	285	3	first	first	ADJ
bracis-19080	285	4	deterioration	deterioration	NOUN
bracis-19080	285	5	example	example	NOUN
bracis-19080	285	6	,	,	PUNCT
bracis-19080	285	7	we	we	PRON
bracis-19080	285	8	believe	believe	VERB
bracis-19080	285	9	the	the	DET
bracis-19080	285	10	model	model	NOUN
bracis-19080	285	11	misinterprets	misinterpret	VERB
bracis-19080	285	12	the	the	DET
bracis-19080	285	13	terms	term	NOUN
bracis-19080	285	14	“	"	PUNCT
bracis-19080	285	15	colocando	colocando	NOUN
bracis-19080	285	16	”	"	PUNCT
bracis-19080	285	17	(	(	PUNCT
bracis-19080	285	18	putting	put	VERB
bracis-19080	285	19	)	)	PUNCT
bracis-19080	285	20	e	e	NOUN
bracis-19080	285	21	“	"	PUNCT
bracis-19080	285	22	tirando	tirando	PROPN
bracis-19080	285	23	”	"	PUNCT
bracis-19080	285	24	 	 	SPACE
bracis-19080	285	25	(	(	PUNCT
bracis-19080	285	26	taking	taking	NOUN
bracis-19080	285	27	)	)	PUNCT
bracis-19080	285	28	,	,	PUNCT
bracis-19080	285	29	which	which	PRON
bracis-19080	285	30	can	can	AUX
bracis-19080	285	31	be	be	AUX
bracis-19080	285	32	used	use	VERB
bracis-19080	285	33	as	as	ADP
bracis-19080	285	34	antonyms	antonyms	NOUN
bracis-19080	285	35	but	but	CCONJ
bracis-19080	285	36	not	not	PART
bracis-19080	285	37	in	in	ADP
bracis-19080	285	38	the	the	DET
bracis-19080	285	39	context	context	NOUN
bracis-19080	285	40	“	"	PUNCT
bracis-19080	285	41	tirando	tirando	PROPN
bracis-19080	285	42	uma	uma	PROPN
bracis-19080	285	43	foto	foto	PROPN
bracis-19080	285	44	”	"	PUNCT
bracis-19080	285	45	 	 	SPACE
bracis-19080	285	46	(	(	PUNCT
bracis-19080	285	47	taking	take	VERB
bracis-19080	285	48	a	a	DET
bracis-19080	285	49	picture	picture	NOUN
bracis-19080	285	50	)	)	PUNCT
bracis-19080	285	51	.	.	PUNCT
bracis-19080	286	1	in	in	ADP
bracis-19080	286	2	the	the	DET
bracis-19080	286	3	second	second	ADJ
bracis-19080	286	4	example	example	NOUN
bracis-19080	286	5	,	,	PUNCT
bracis-19080	286	6	we	we	PRON
bracis-19080	286	7	believe	believe	VERB
bracis-19080	286	8	the	the	DET
bracis-19080	286	9	model	model	NOUN
bracis-19080	286	10	misinterprets	misinterpret	VERB
bracis-19080	286	11	a	a	DET
bracis-19080	286	12	negation	negation	NOUN
bracis-19080	286	13	quantifier	quantifier	NOUN
bracis-19080	286	14	“	"	PUNCT
bracis-19080	286	15	não	não	PROPN
bracis-19080	286	16	tem	tem	PROPN
bracis-19080	286	17	nenhuma	nenhuma	PROPN
bracis-19080	286	18	mulher	mulher	NOUN
bracis-19080	286	19	”	"	PUNCT
bracis-19080	286	20	(	(	PUNCT
bracis-19080	286	21	no	no	DET
bracis-19080	286	22	woman	woman	NOUN
bracis-19080	286	23	)	)	PUNCT
bracis-19080	286	24	with	with	ADP
bracis-19080	286	25	an	an	DET
bracis-19080	286	26	opposition	opposition	NOUN
bracis-19080	286	27	adverb	adverb	NOUN
bracis-19080	286	28	 	 	SPACE
bracis-19080	286	29	(	(	PUNCT
bracis-19080	286	30	not	not	PART
bracis-19080	286	31	a	a	DET
bracis-19080	286	32	woman	woman	NOUN
bracis-19080	286	33	)	)	PUNCT
bracis-19080	286	34	,	,	PUNCT
bracis-19080	286	35	which	which	PRON
bracis-19080	286	36	renders	render	VERB
bracis-19080	286	37	both	both	DET
bracis-19080	286	38	sentences	sentence	NOUN
bracis-19080	286	39	less	less	ADV
bracis-19080	286	40	semantically	semantically	ADV
bracis-19080	286	41	different	different	ADJ
bracis-19080	286	42	.	.	PUNCT
bracis-19080	287	1	future	future	ADJ
bracis-19080	287	2	analysis	analysis	NOUN
bracis-19080	287	3	could	could	AUX
bracis-19080	287	4	also	also	ADV
bracis-19080	287	5	investigate	investigate	VERB
bracis-19080	287	6	the	the	DET
bracis-19080	287	7	impact	impact	NOUN
bracis-19080	287	8	of	of	ADP
bracis-19080	287	9	double	double	ADJ
bracis-19080	287	10	negatives	negative	NOUN
bracis-19080	287	11	in	in	ADP
bracis-19080	287	12	this	this	DET
bracis-19080	287	13	case	case	NOUN
bracis-19080	287	14	,	,	PUNCT
bracis-19080	287	15	as	as	SCONJ
bracis-19080	287	16	multilingual	multilingual	ADJ
bracis-19080	287	17	models	model	NOUN
bracis-19080	287	18	need	need	VERB
bracis-19080	287	19	to	to	PART
bracis-19080	287	20	deal	deal	VERB
bracis-19080	287	21	with	with	ADP
bracis-19080	287	22	such	such	ADJ
bracis-19080	287	23	differences	difference	NOUN
bracis-19080	287	24	between	between	ADP
bracis-19080	287	25	languages	language	NOUN
bracis-19080	287	26	.	.	PUNCT
bracis-19080	288	1	finally	finally	ADV
bracis-19080	288	2	,	,	PUNCT
bracis-19080	288	3	in	in	ADP
bracis-19080	288	4	the	the	DET
bracis-19080	288	5	last	last	ADJ
bracis-19080	288	6	example	example	NOUN
bracis-19080	288	7	the	the	DET
bracis-19080	288	8	only	only	ADJ
bracis-19080	288	9	differences	difference	NOUN
bracis-19080	288	10	between	between	ADP
bracis-19080	288	11	sentences	sentence	NOUN
bracis-19080	288	12	regard	regard	NOUN
bracis-19080	288	13	(	(	PUNCT
bracis-19080	288	14	i	i	NOUN
bracis-19080	288	15	)	)	PUNCT
bracis-19080	288	16	 	 	SPACE
bracis-19080	288	17	active	active	ADJ
bracis-19080	288	18	versus	versus	ADP
bracis-19080	288	19	passive	passive	ADJ
bracis-19080	288	20	sentence	sentence	NOUN
bracis-19080	288	21	form	form	NOUN
bracis-19080	288	22	,	,	PUNCT
bracis-19080	288	23	and	and	CCONJ
bracis-19080	288	24	(	(	PUNCT
bracis-19080	288	25	ii	ii	NOUN
bracis-19080	288	26	)	)	PUNCT
bracis-19080	288	27	 	 	SPACE
bracis-19080	288	28	whether	whether	SCONJ
bracis-19080	288	29	“	"	PUNCT
bracis-19080	288	30	montando	montando	X
bracis-19080	288	31	”	"	PUNCT
bracis-19080	288	32	refers	refer	VERB
bracis-19080	288	33	to	to	ADP
bracis-19080	288	34	getting	get	VERB
bracis-19080	288	35	on	on	ADP
bracis-19080	288	36	or	or	CCONJ
bracis-19080	288	37	riding	ride	VERB
bracis-19080	288	38	an	an	DET
bracis-19080	288	39	animal	animal	NOUN
bracis-19080	288	40	.	.	PUNCT
bracis-19080	289	1	since	since	SCONJ
bracis-19080	289	2	both	both	DET
bracis-19080	289	3	interpretations	interpretation	NOUN
bracis-19080	289	4	are	be	AUX
bracis-19080	289	5	possible	possible	ADJ
bracis-19080	289	6	,	,	PUNCT
bracis-19080	289	7	we	we	PRON
bracis-19080	289	8	believe	believe	VERB
bracis-19080	289	9	that	that	SCONJ
bracis-19080	289	10	the	the	DET
bracis-19080	289	11	assin	assin	PROPN
bracis-19080	289	12	 	 	SPACE
bracis-19080	289	13	2	2	NUM
bracis-19080	289	14	example	example	NOUN
bracis-19080	289	15	could	could	AUX
bracis-19080	289	16	be	be	AUX
bracis-19080	289	17	considered	consider	VERB
bracis-19080	289	18	noisy	noisy	ADJ
bracis-19080	289	19	.	.	PUNCT
bracis-19080	290	1	6.2	6.2	NUM
bracis-19080	290	2	language	language	NOUN
bracis-19080	290	3	variant	variant	NOUN
bracis-19080	290	4	effects	effect	NOUN
bracis-19080	290	5	with	with	ADP
bracis-19080	290	6	the	the	DET
bracis-19080	290	7	exception	exception	NOUN
bracis-19080	290	8	of	of	ADP
bracis-19080	290	9	the	the	DET
bracis-19080	290	10	state	state	NOUN
bracis-19080	290	11	-	-	PUNCT
bracis-19080	290	12	of	of	ADP
bracis-19080	290	13	-	-	PUNCT
bracis-19080	290	14	the	the	DET
bracis-19080	290	15	-	-	PUNCT
bracis-19080	290	16	art	art	NOUN
bracis-19080	290	17	,	,	PUNCT
bracis-19080	290	18	all	all	DET
bracis-19080	290	19	algorithms	algorithm	NOUN
bracis-19080	290	20	assessed	assess	VERB
bracis-19080	290	21	for	for	ADP
bracis-19080	290	22	assin	assin	PROPN
bracis-19080	290	23	datasets	dataset	NOUN
bracis-19080	290	24	in	in	ADP
bracis-19080	290	25	the	the	DET
bracis-19080	290	26	literature	literature	NOUN
bracis-19080	290	27	and	and	CCONJ
bracis-19080	290	28	also	also	ADV
bracis-19080	290	29	here	here	ADV
bracis-19080	290	30	are	be	AUX
bracis-19080	290	31	unable	unable	ADJ
bracis-19080	290	32	to	to	PART
bracis-19080	290	33	perform	perform	VERB
bracis-19080	290	34	their	their	PRON
bracis-19080	290	35	best	good	ADJ
bracis-19080	290	36	for	for	ADP
bracis-19080	290	37	the	the	DET
bracis-19080	290	38	european	european	ADJ
bracis-19080	290	39	and	and	CCONJ
bracis-19080	290	40	brazilian	brazilian	ADJ
bracis-19080	290	41	portuguese	portuguese	ADJ
bracis-19080	290	42	variants	variant	NOUN
bracis-19080	290	43	at	at	ADP
bracis-19080	290	44	the	the	DET
bracis-19080	290	45	same	same	ADJ
bracis-19080	290	46	time	time	NOUN
bracis-19080	290	47	.	.	PUNCT
bracis-19080	291	1	we	we	PRON
bracis-19080	291	2	illustrate	illustrate	VERB
bracis-19080	291	3	this	this	PRON
bracis-19080	291	4	with	with	ADP
bracis-19080	291	5	sentence	sentence	NOUN
bracis-19080	291	6	pairs	pair	NOUN
bracis-19080	291	7	given	give	VERB
bracis-19080	291	8	in	in	ADP
bracis-19080	291	9	table	table	NOUN
bracis-19080	291	10	 	 	SPACE
bracis-19080	291	11	3	3	NUM
bracis-19080	291	12	.	.	PUNCT
bracis-19080	292	1	the	the	DET
bracis-19080	292	2	two	two	NUM
bracis-19080	292	3	top	top	ADV
bracis-19080	292	4	-	-	PUNCT
bracis-19080	292	5	most	most	ADJ
bracis-19080	292	6	sentence	sentence	NOUN
bracis-19080	292	7	pairs	pair	NOUN
bracis-19080	292	8	are	be	AUX
bracis-19080	292	9	obtained	obtain	VERB
bracis-19080	292	10	from	from	ADP
bracis-19080	292	11	assin	assin	PROPN
bracis-19080	292	12	pt	pt	PROPN
bracis-19080	292	13	-	-	PUNCT
bracis-19080	292	14	br	br	NOUN
bracis-19080	292	15	,	,	PUNCT
bracis-19080	292	16	and	and	CCONJ
bracis-19080	292	17	the	the	DET
bracis-19080	292	18	two	two	NUM
bracis-19080	292	19	bottom	bottom	ADV
bracis-19080	292	20	-	-	PUNCT
bracis-19080	292	21	most	most	ADJ
bracis-19080	292	22	from	from	ADP
bracis-19080	292	23	assin	assin	PROPN
bracis-19080	292	24	pt	pt	PROPN
bracis-19080	292	25	-	-	PUNCT
bracis-19080	292	26	pt	pt	PROPN
bracis-19080	292	27	.	.	PROPN
bracis-19080	292	28	sentences	sentence	NOUN
bracis-19080	292	29	selected	select	VERB
bracis-19080	292	30	in	in	ADP
bracis-19080	292	31	european	european	ADJ
bracis-19080	292	32	portuguese	portuguese	PROPN
bracis-19080	292	33	respectively	respectively	ADV
bracis-19080	292	34	illustrate	illustrate	VERB
bracis-19080	292	35	vocabulary	vocabulary	NOUN
bracis-19080	292	36	and	and	CCONJ
bracis-19080	292	37	grammatical	grammatical	ADJ
bracis-19080	292	38	differences	difference	NOUN
bracis-19080	292	39	between	between	ADP
bracis-19080	292	40	the	the	DET
bracis-19080	292	41	language	language	NOUN
bracis-19080	292	42	variants	variant	NOUN
bracis-19080	292	43	.	.	PUNCT
bracis-19080	293	1	beside	beside	ADP
bracis-19080	293	2	expected	expect	VERB
bracis-19080	293	3	scores	score	NOUN
bracis-19080	293	4	,	,	PUNCT
bracis-19080	293	5	similarities	similarity	NOUN
bracis-19080	293	6	computed	compute	VERB
bracis-19080	293	7	by	by	ADP
bracis-19080	293	8	\(\langle	\(\langle	PROPN
bracis-19080	293	9	\textsf	\textsf	PROPN
bracis-19080	293	10	{	{	PUNCT
bracis-19080	293	11	xlm	xlm	PROPN
bracis-19080	293	12	-	-	PUNCT
bracis-19080	293	13	r	r	NOUN
bracis-19080	293	14	,	,	PUNCT
bracis-19080	293	15	sbert	sbert	NOUN
bracis-19080	293	16	,	,	PUNCT
bracis-19080	293	17	paraphrases}\rangle	paraphrases}\rangle	NUM
bracis-19080	293	18	\	\	NOUN
bracis-19080	293	19	)	)	PUNCT
bracis-19080	293	20	when	when	SCONJ
bracis-19080	293	21	fine	fine	ADV
bracis-19080	293	22	-	-	PUNCT
bracis-19080	293	23	tuned	tune	VERB
bracis-19080	293	24	for	for	ADP
bracis-19080	293	25	assin	assin	PROPN
bracis-19080	293	26	pt	pt	PROPN
bracis-19080	293	27	-	-	PUNCT
bracis-19080	293	28	br	br	PROPN
bracis-19080	293	29	and	and	CCONJ
bracis-19080	293	30	assin	assin	PROPN
bracis-19080	293	31	pt	pt	PROPN
bracis-19080	293	32	-	-	PUNCT
bracis-19080	293	33	pt	pt	NOUN
bracis-19080	293	34	are	be	AUX
bracis-19080	293	35	also	also	ADV
bracis-19080	293	36	given	give	VERB
bracis-19080	293	37	.	.	PUNCT
bracis-19080	294	1	results	result	NOUN
bracis-19080	294	2	for	for	ADP
bracis-19080	294	3	the	the	DET
bracis-19080	294	4	model	model	NOUN
bracis-19080	294	5	fine	fine	ADV
bracis-19080	294	6	-	-	PUNCT
bracis-19080	294	7	tuned	tune	VERB
bracis-19080	294	8	for	for	ADP
bracis-19080	294	9	assin	assin	PROPN
bracis-19080	294	10	pt	pt	PROPN
bracis-19080	294	11	-	-	PUNCT
bracis-19080	294	12	br	br	NOUN
bracis-19080	294	13	are	be	AUX
bracis-19080	294	14	better	well	ADJ
bracis-19080	294	15	on	on	ADP
bracis-19080	294	16	that	that	DET
bracis-19080	294	17	dataset	dataset	NOUN
bracis-19080	294	18	.	.	PUNCT
bracis-19080	295	1	the	the	DET
bracis-19080	295	2	symmetrical	symmetrical	ADJ
bracis-19080	295	3	situation	situation	NOUN
bracis-19080	295	4	is	be	AUX
bracis-19080	295	5	observed	observe	VERB
bracis-19080	295	6	for	for	ADP
bracis-19080	295	7	the	the	DET
bracis-19080	295	8	assin	assin	PROPN
bracis-19080	295	9	pt	pt	PROPN
bracis-19080	295	10	-	-	PUNCT
bracis-19080	295	11	pt	pt	NOUN
bracis-19080	295	12	dataset	dataset	NOUN
bracis-19080	295	13	.	.	PUNCT
bracis-19080	296	1	the	the	DET
bracis-19080	296	2	better	well	ADJ
bracis-19080	296	3	performance	performance	NOUN
bracis-19080	296	4	of	of	ADP
bracis-19080	296	5	models	model	NOUN
bracis-19080	296	6	fine	fine	ADV
bracis-19080	296	7	-	-	PUNCT
bracis-19080	296	8	tuned	tune	VERB
bracis-19080	296	9	for	for	SCONJ
bracis-19080	296	10	the	the	DET
bracis-19080	296	11	given	give	VERB
bracis-19080	296	12	language	language	NOUN
bracis-19080	296	13	variant	variant	NOUN
bracis-19080	296	14	is	be	AUX
bracis-19080	296	15	expected	expect	VERB
bracis-19080	296	16	.	.	PUNCT
bracis-19080	297	1	we	we	PRON
bracis-19080	297	2	then	then	ADV
bracis-19080	297	3	conduct	conduct	VERB
bracis-19080	297	4	additional	additional	ADJ
bracis-19080	297	5	experiments	experiment	NOUN
bracis-19080	297	6	to	to	PART
bracis-19080	297	7	understand	understand	VERB
bracis-19080	297	8	the	the	DET
bracis-19080	297	9	impact	impact	NOUN
bracis-19080	297	10	of	of	ADP
bracis-19080	297	11	fine	fine	ADV
bracis-19080	297	12	-	-	PUNCT
bracis-19080	297	13	tuning	tune	VERB
bracis-19080	297	14	\(\langle	\(\langle	NOUN
bracis-19080	297	15	\textsf	\textsf	PROPN
bracis-19080	297	16	{	{	PUNCT
bracis-19080	297	17	xlm	xlm	PROPN
bracis-19080	297	18	-	-	PUNCT
bracis-19080	297	19	r	r	NOUN
bracis-19080	297	20	,	,	PUNCT
bracis-19080	297	21	sbert	sbert	NOUN
bracis-19080	297	22	,	,	PUNCT
bracis-19080	297	23	paraphrases}\rangle	paraphrases}\rangle	NUM
bracis-19080	297	24	\	\	NOUN
bracis-19080	297	25	)	)	PUNCT
bracis-19080	297	26	for	for	ADP
bracis-19080	297	27	the	the	DET
bracis-19080	297	28	three	three	NUM
bracis-19080	297	29	assin	assin	NOUN
bracis-19080	297	30	and	and	CCONJ
bracis-19080	297	31	assin	assin	PROPN
bracis-19080	297	32	 	 	SPACE
bracis-19080	297	33	2	2	NUM
bracis-19080	297	34	datasets	dataset	NOUN
bracis-19080	297	35	altogether	altogether	ADV
bracis-19080	297	36	.	.	PUNCT
bracis-19080	298	1	we	we	PRON
bracis-19080	298	2	report	report	VERB
bracis-19080	298	3	mean	mean	VERB
bracis-19080	298	4	values	value	NOUN
bracis-19080	298	5	for	for	ADP
bracis-19080	298	6	ten	ten	NUM
bracis-19080	298	7	repetitions	repetition	NOUN
bracis-19080	298	8	,	,	PUNCT
bracis-19080	298	9	and	and	CCONJ
bracis-19080	298	10	provide	provide	VERB
bracis-19080	298	11	standard	standard	ADJ
bracis-19080	298	12	deviations	deviation	NOUN
bracis-19080	298	13	as	as	ADP
bracis-19080	298	14	supplementary	supplementary	ADJ
bracis-19080	298	15	material	material	NOUN
bracis-19080	298	16	 	 	SPACE
bracis-19080	299	1	[	[	X
bracis-19080	299	2	2	2	NUM
bracis-19080	299	3	]	]	PUNCT
bracis-19080	299	4	given	give	VERB
bracis-19080	299	5	the	the	DET
bracis-19080	299	6	very	very	ADV
bracis-19080	299	7	low	low	ADJ
bracis-19080	299	8	variability	variability	NOUN
bracis-19080	299	9	observed	observe	VERB
bracis-19080	299	10	in	in	ADP
bracis-19080	299	11	the	the	DET
bracis-19080	299	12	results	result	NOUN
bracis-19080	299	13	.	.	PUNCT
bracis-19080	300	1	results	result	NOUN
bracis-19080	300	2	differ	differ	VERB
bracis-19080	300	3	as	as	ADP
bracis-19080	300	4	a	a	DET
bracis-19080	300	5	function	function	NOUN
bracis-19080	300	6	of	of	ADP
bracis-19080	300	7	the	the	DET
bracis-19080	300	8	target	target	NOUN
bracis-19080	300	9	testing	testing	NOUN
bracis-19080	300	10	dataset	dataset	NOUN
bracis-19080	300	11	.	.	PUNCT
bracis-19080	301	1	for	for	ADP
bracis-19080	301	2	assin	assin	PROPN
bracis-19080	301	3	pt	pt	PROPN
bracis-19080	301	4	-	-	PUNCT
bracis-19080	301	5	br	br	NOUN
bracis-19080	301	6	and	and	CCONJ
bracis-19080	301	7	assin	assin	PROPN
bracis-19080	301	8	 	 	SPACE
bracis-19080	301	9	2	2	NUM
bracis-19080	301	10	,	,	PUNCT
bracis-19080	301	11	results	result	NOUN
bracis-19080	301	12	are	be	AUX
bracis-19080	301	13	not	not	PART
bracis-19080	301	14	changed	change	VERB
bracis-19080	301	15	,	,	PUNCT
bracis-19080	301	16	except	except	SCONJ
bracis-19080	301	17	for	for	ADP
bracis-19080	301	18	mse	mse	NOUN
bracis-19080	301	19	on	on	ADP
bracis-19080	301	20	the	the	DET
bracis-19080	301	21	latter	latter	ADJ
bracis-19080	301	22	,	,	PUNCT
bracis-19080	301	23	reduced	reduce	VERB
bracis-19080	301	24	from	from	ADP
bracis-19080	301	25	0.43	0.43	NUM
bracis-19080	301	26	to	to	ADP
bracis-19080	301	27	0.41	0.41	NUM
bracis-19080	301	28	.	.	PUNCT
bracis-19080	302	1	regarding	regard	VERB
bracis-19080	302	2	assin	assin	PROPN
bracis-19080	302	3	pt	pt	PROPN
bracis-19080	302	4	-	-	PUNCT
bracis-19080	302	5	pt	pt	NOUN
bracis-19080	302	6	,	,	PUNCT
bracis-19080	302	7	results	result	NOUN
bracis-19080	302	8	are	be	AUX
bracis-19080	302	9	strongly	strongly	ADV
bracis-19080	302	10	improved	improve	VERB
bracis-19080	302	11	,	,	PUNCT
bracis-19080	302	12	with	with	ADP
bracis-19080	302	13	pearson	pearson	PROPN
bracis-19080	302	14	coefficient	coefficient	NOUN
bracis-19080	302	15	correlation	correlation	NOUN
bracis-19080	302	16	increasing	increase	VERB
bracis-19080	302	17	from	from	ADP
bracis-19080	302	18	0.77	0.77	NUM
bracis-19080	302	19	to	to	ADP
bracis-19080	302	20	0.81	0.81	NUM
bracis-19080	302	21	and	and	CCONJ
bracis-19080	302	22	mse	mse	NOUN
bracis-19080	302	23	reducing	reduce	VERB
bracis-19080	302	24	from	from	ADP
bracis-19080	302	25	0.52	0.52	NUM
bracis-19080	302	26	to	to	ADP
bracis-19080	302	27	0.49	0.49	NUM
bracis-19080	302	28	.	.	PUNCT
bracis-19080	303	1	table	table	NOUN
bracis-19080	303	2	3	3	PROPN
bracis-19080	303	3	.	.	PUNCT
bracis-19080	304	1	assin	assin	PROPN
bracis-19080	304	2	pt	pt	PROPN
bracis-19080	304	3	-	-	PUNCT
bracis-19080	304	4	br	br	NOUN
bracis-19080	304	5	 	 	SPACE
bracis-19080	304	6	(	(	PUNCT
bracis-19080	304	7	top	top	NOUN
bracis-19080	304	8	)	)	PUNCT
bracis-19080	304	9	and	and	CCONJ
bracis-19080	304	10	pt	pt	PROPN
bracis-19080	304	11	-	-	PUNCT
bracis-19080	304	12	pt	pt	NOUN
bracis-19080	304	13	 	 	SPACE
bracis-19080	304	14	(	(	PUNCT
bracis-19080	304	15	bottom	bottom	ADJ
bracis-19080	304	16	)	)	PUNCT
bracis-19080	304	17	pairs	pair	NOUN
bracis-19080	304	18	with	with	ADP
bracis-19080	304	19	their	their	PRON
bracis-19080	304	20	scores	score	NOUN
bracis-19080	304	21	illustrating	illustrate	VERB
bracis-19080	304	22	fine	fine	ADV
bracis-19080	304	23	-	-	PUNCT
bracis-19080	304	24	tuning	tune	VERB
bracis-19080	304	25	\(\langle	\(\langle	NOUN
bracis-19080	304	26	\textsf	\textsf	PROPN
bracis-19080	304	27	{	{	PUNCT
bracis-19080	304	28	xlm	xlm	PROPN
bracis-19080	304	29	-	-	PUNCT
bracis-19080	304	30	r	r	NOUN
bracis-19080	304	31	,	,	PUNCT
bracis-19080	304	32	sbert	sbert	NOUN
bracis-19080	304	33	,	,	PUNCT
bracis-19080	304	34	paraphrases}\rangle	paraphrases}\rangle	NUM
bracis-19080	304	35	\	\	NOUN
bracis-19080	304	36	)	)	PUNCT
bracis-19080	304	37	for	for	ADP
bracis-19080	304	38	different	different	ADJ
bracis-19080	304	39	language	language	NOUN
bracis-19080	304	40	variants.full	variants.full	ADJ
bracis-19080	304	41	size	size	NOUN
bracis-19080	304	42	table	table	NOUN
bracis-19080	304	43	the	the	DET
bracis-19080	304	44	qualitative	qualitative	ADJ
bracis-19080	304	45	assessment	assessment	NOUN
bracis-19080	304	46	conducted	conduct	VERB
bracis-19080	304	47	in	in	ADP
bracis-19080	304	48	this	this	DET
bracis-19080	304	49	section	section	NOUN
bracis-19080	304	50	helped	help	VERB
bracis-19080	304	51	further	far	ADV
bracis-19080	304	52	understand	understand	VERB
bracis-19080	304	53	the	the	DET
bracis-19080	304	54	benefits	benefit	NOUN
bracis-19080	304	55	of	of	ADP
bracis-19080	304	56	fine	fine	ADV
bracis-19080	304	57	-	-	PUNCT
bracis-19080	304	58	tuning	tuning	NOUN
bracis-19080	304	59	and	and	CCONJ
bracis-19080	304	60	the	the	DET
bracis-19080	304	61	potential	potential	ADJ
bracis-19080	304	62	impacts	impact	NOUN
bracis-19080	304	63	of	of	ADP
bracis-19080	304	64	different	different	ADJ
bracis-19080	304	65	language	language	NOUN
bracis-19080	304	66	variants	variant	NOUN
bracis-19080	304	67	.	.	PUNCT
bracis-19080	305	1	even	even	ADV
bracis-19080	305	2	if	if	SCONJ
bracis-19080	305	3	improved	improved	ADJ
bracis-19080	305	4	results	result	NOUN
bracis-19080	305	5	did	do	AUX
bracis-19080	305	6	not	not	PART
bracis-19080	305	7	match	match	VERB
bracis-19080	305	8	the	the	DET
bracis-19080	305	9	state	state	NOUN
bracis-19080	305	10	-	-	PUNCT
bracis-19080	305	11	of	of	ADP
bracis-19080	305	12	-	-	PUNCT
bracis-19080	305	13	the	the	DET
bracis-19080	305	14	-	-	PUNCT
bracis-19080	305	15	art	art	NOUN
bracis-19080	305	16	for	for	ADP
bracis-19080	305	17	the	the	DET
bracis-19080	305	18	assin	assin	PROPN
bracis-19080	305	19	pt	pt	PROPN
bracis-19080	305	20	-	-	PUNCT
bracis-19080	305	21	pt	pt	NOUN
bracis-19080	305	22	dataset	dataset	NOUN
bracis-19080	305	23	,	,	PUNCT
bracis-19080	305	24	they	they	PRON
bracis-19080	305	25	shed	shed	VERB
bracis-19080	305	26	light	light	NOUN
bracis-19080	305	27	into	into	ADP
bracis-19080	305	28	promising	promise	VERB
bracis-19080	305	29	future	future	ADJ
bracis-19080	305	30	work	work	NOUN
bracis-19080	305	31	directions	direction	NOUN
bracis-19080	305	32	.	.	PUNCT
bracis-19080	306	1	7	7	NUM
bracis-19080	306	2	conclusions	conclusion	NOUN
bracis-19080	306	3	semantic	semantic	ADJ
bracis-19080	306	4	textual	textual	ADJ
bracis-19080	306	5	similarity	similarity	NOUN
bracis-19080	306	6	 	 	SPACE
bracis-19080	306	7	(	(	PUNCT
bracis-19080	306	8	sts	st	VERB
bracis-19080	306	9	 	 	SPACE
bracis-19080	306	10	[	[	PUNCT
bracis-19080	306	11	11	11	NUM
bracis-19080	306	12	]	]	PUNCT
bracis-19080	306	13	)	)	PUNCT
bracis-19080	306	14	is	be	AUX
bracis-19080	306	15	a	a	DET
bracis-19080	306	16	core	core	NOUN
bracis-19080	306	17	research	research	NOUN
bracis-19080	306	18	problem	problem	NOUN
bracis-19080	306	19	in	in	ADP
bracis-19080	306	20	natural	natural	ADJ
bracis-19080	306	21	language	language	NOUN
bracis-19080	306	22	processings	processing	NOUN
bracis-19080	306	23	 	 	SPACE
bracis-19080	306	24	(	(	PUNCT
bracis-19080	306	25	nlp	nlp	ADJ
bracis-19080	306	26	)	)	PUNCT
bracis-19080	306	27	studies	study	NOUN
bracis-19080	306	28	.	.	PUNCT
bracis-19080	307	1	this	this	PRON
bracis-19080	307	2	is	be	AUX
bracis-19080	307	3	evidenced	evidence	VERB
bracis-19080	307	4	by	by	ADP
bracis-19080	307	5	the	the	DET
bracis-19080	307	6	(	(	PUNCT
bracis-19080	307	7	i	i	NOUN
bracis-19080	307	8	)	)	PUNCT
bracis-19080	307	9	 	 	SPACE
bracis-19080	307	10	international	international	ADJ
bracis-19080	307	11	efforts	effort	NOUN
bracis-19080	307	12	promoting	promote	VERB
bracis-19080	307	13	sts	st	NOUN
bracis-19080	307	14	shared	share	VERB
bracis-19080	307	15	tasks	task	NOUN
bracis-19080	307	16	 	 	SPACE
bracis-19080	308	1	[	[	X
bracis-19080	308	2	5	5	NUM
bracis-19080	308	3	,	,	PUNCT
bracis-19080	308	4	12	12	NUM
bracis-19080	308	5	]	]	PUNCT
bracis-19080	308	6	and	and	CCONJ
bracis-19080	308	7	(	(	PUNCT
bracis-19080	308	8	ii	ii	NOUN
bracis-19080	308	9	)	)	PUNCT
bracis-19080	308	10	 	 	SPACE
bracis-19080	309	1	the	the	DET
bracis-19080	309	2	fact	fact	NOUN
bracis-19080	309	3	that	that	SCONJ
bracis-19080	309	4	several	several	ADJ
bracis-19080	309	5	novel	novel	ADJ
bracis-19080	309	6	contextual	contextual	ADJ
bracis-19080	309	7	embedddings	embeddding	NOUN
bracis-19080	309	8	recently	recently	ADV
bracis-19080	309	9	proposed	propose	VERB
bracis-19080	309	10	have	have	AUX
bracis-19080	309	11	been	be	AUX
bracis-19080	309	12	benchmarked	benchmarke	VERB
bracis-19080	309	13	on	on	ADP
bracis-19080	309	14	this	this	DET
bracis-19080	309	15	problem	problem	NOUN
bracis-19080	309	16	 	 	SPACE
bracis-19080	310	1	[	[	X
bracis-19080	310	2	17	17	NUM
bracis-19080	310	3	]	]	PUNCT
bracis-19080	310	4	.	.	PUNCT
bracis-19080	311	1	in	in	ADP
bracis-19080	311	2	the	the	DET
bracis-19080	311	3	context	context	NOUN
bracis-19080	311	4	of	of	ADP
bracis-19080	311	5	the	the	DET
bracis-19080	311	6	portuguese	portuguese	ADJ
bracis-19080	311	7	language	language	NOUN
bracis-19080	311	8	and	and	CCONJ
bracis-19080	311	9	its	its	PRON
bracis-19080	311	10	variants	variant	NOUN
bracis-19080	311	11	,	,	PUNCT
bracis-19080	311	12	assin	assin	PROPN
bracis-19080	311	13	and	and	CCONJ
bracis-19080	311	14	assin	assin	PROPN
bracis-19080	311	15	 	 	SPACE
bracis-19080	311	16	2	2	NUM
bracis-19080	311	17	are	be	AUX
bracis-19080	311	18	the	the	DET
bracis-19080	311	19	most	most	ADV
bracis-19080	311	20	relevant	relevant	ADJ
bracis-19080	311	21	shared	share	VERB
bracis-19080	311	22	tasks	task	NOUN
bracis-19080	311	23	identified	identify	VERB
bracis-19080	311	24	 	 	SPACE
bracis-19080	312	1	[	[	X
bracis-19080	312	2	9	9	NUM
bracis-19080	312	3	,	,	PUNCT
bracis-19080	312	4	15	15	NUM
bracis-19080	312	5	]	]	PUNCT
bracis-19080	312	6	,	,	PUNCT
bracis-19080	312	7	for	for	ADP
bracis-19080	312	8	which	which	PRON
bracis-19080	312	9	bert	bert	NOUN
bracis-19080	312	10	embeddings	embedding	NOUN
bracis-19080	312	11	comprise	comprise	VERB
bracis-19080	312	12	the	the	DET
bracis-19080	312	13	current	current	ADJ
bracis-19080	312	14	state	state	NOUN
bracis-19080	312	15	-	-	PUNCT
bracis-19080	312	16	of	of	ADP
bracis-19080	312	17	-	-	PUNCT
bracis-19080	312	18	the	the	DET
bracis-19080	312	19	-	-	PUNCT
bracis-19080	312	20	art	art	NOUN
bracis-19080	312	21	 	 	SPACE
bracis-19080	312	22	[	[	X
bracis-19080	312	23	8	8	NUM
bracis-19080	312	24	]	]	PUNCT
bracis-19080	312	25	.	.	PUNCT
bracis-19080	313	1	in	in	ADP
bracis-19080	313	2	this	this	DET
bracis-19080	313	3	work	work	NOUN
bracis-19080	313	4	,	,	PUNCT
bracis-19080	313	5	we	we	PRON
bracis-19080	313	6	have	have	AUX
bracis-19080	313	7	investigated	investigate	VERB
bracis-19080	313	8	the	the	DET
bracis-19080	313	9	application	application	NOUN
bracis-19080	313	10	of	of	ADP
bracis-19080	313	11	the	the	DET
bracis-19080	313	12	sentence	sentence	NOUN
bracis-19080	313	13	-	-	PUNCT
bracis-19080	313	14	bert	bert	PROPN
bracis-19080	313	15	 	 	SPACE
bracis-19080	313	16	(	(	PUNCT
bracis-19080	313	17	sbert	sbert	NOUN
bracis-19080	313	18	 	 	SPACE
bracis-19080	314	1	[	[	X
bracis-19080	314	2	17	17	NUM
bracis-19080	314	3	]	]	PUNCT
bracis-19080	314	4	)	)	PUNCT
bracis-19080	314	5	architecture	architecture	NOUN
bracis-19080	314	6	to	to	PART
bracis-19080	314	7	sts	sts	VERB
bracis-19080	314	8	in	in	ADP
bracis-19080	314	9	portuguese	portuguese	PROPN
bracis-19080	314	10	.	.	PUNCT
bracis-19080	315	1	sbert	sbert	NOUN
bracis-19080	315	2	addresses	address	NOUN
bracis-19080	315	3	scalability	scalability	NOUN
bracis-19080	315	4	issues	issue	NOUN
bracis-19080	315	5	in	in	ADP
bracis-19080	315	6	bert	bert	PROPN
bracis-19080	315	7	networks	network	NOUN
bracis-19080	315	8	,	,	PUNCT
bracis-19080	315	9	and	and	CCONJ
bracis-19080	315	10	has	have	AUX
bracis-19080	315	11	been	be	AUX
bracis-19080	315	12	successfully	successfully	ADV
bracis-19080	315	13	applied	apply	VERB
bracis-19080	315	14	and	and	CCONJ
bracis-19080	315	15	open	open	ADJ
bracis-19080	315	16	-	-	PUNCT
bracis-19080	315	17	source	source	NOUN
bracis-19080	315	18	also	also	ADV
bracis-19080	315	19	for	for	ADP
bracis-19080	315	20	multilingual	multilingual	ADJ
bracis-19080	315	21	applications	application	NOUN
bracis-19080	315	22	 	 	SPACE
bracis-19080	316	1	[	[	X
bracis-19080	316	2	18	18	NUM
bracis-19080	316	3	]	]	PUNCT
bracis-19080	316	4	.	.	PUNCT
bracis-19080	317	1	our	our	PRON
bracis-19080	317	2	contributions	contribution	NOUN
bracis-19080	317	3	were	be	AUX
bracis-19080	317	4	two	two	NUM
bracis-19080	317	5	-	-	PUNCT
bracis-19080	317	6	fold	fold	ADJ
bracis-19080	317	7	.	.	PUNCT
bracis-19080	318	1	in	in	ADP
bracis-19080	318	2	the	the	DET
bracis-19080	318	3	first	first	ADJ
bracis-19080	318	4	part	part	NOUN
bracis-19080	318	5	of	of	ADP
bracis-19080	318	6	this	this	DET
bracis-19080	318	7	work	work	NOUN
bracis-19080	318	8	,	,	PUNCT
bracis-19080	318	9	we	we	PRON
bracis-19080	318	10	assessed	assess	VERB
bracis-19080	318	11	portuguese	portuguese	ADJ
bracis-19080	318	12	and	and	CCONJ
bracis-19080	318	13	multilingual	multilingual	ADJ
bracis-19080	318	14	sbert	sbert	NOUN
bracis-19080	318	15	models	model	NOUN
bracis-19080	318	16	,	,	PUNCT
bracis-19080	318	17	demonstrating	demonstrate	VERB
bracis-19080	318	18	that	that	SCONJ
bracis-19080	318	19	the	the	DET
bracis-19080	318	20	latter	latter	ADJ
bracis-19080	318	21	present	present	ADJ
bracis-19080	318	22	competitive	competitive	ADJ
bracis-19080	318	23	performance	performance	NOUN
bracis-19080	318	24	w.r.t	w.r.t	NOUN
bracis-19080	318	25	.	.	PUNCT
bracis-19080	319	1	shared	share	VERB
bracis-19080	319	2	task	task	NOUN
bracis-19080	319	3	results	result	NOUN
bracis-19080	319	4	even	even	ADV
bracis-19080	319	5	without	without	ADP
bracis-19080	319	6	fine	fine	ADV
bracis-19080	319	7	-	-	PUNCT
bracis-19080	319	8	tuning	tuning	NOUN
bracis-19080	319	9	.	.	PUNCT
bracis-19080	320	1	this	this	PRON
bracis-19080	320	2	can	can	AUX
bracis-19080	320	3	be	be	AUX
bracis-19080	320	4	very	very	ADV
bracis-19080	320	5	useful	useful	ADJ
bracis-19080	320	6	to	to	ADP
bracis-19080	320	7	data	data	NOUN
bracis-19080	320	8	professionals	professional	NOUN
bracis-19080	320	9	who	who	PRON
bracis-19080	320	10	need	need	VERB
bracis-19080	320	11	to	to	PART
bracis-19080	320	12	address	address	VERB
bracis-19080	320	13	this	this	DET
bracis-19080	320	14	problem	problem	NOUN
bracis-19080	320	15	in	in	ADP
bracis-19080	320	16	unsupervised	unsupervised	ADJ
bracis-19080	320	17	learning	learning	NOUN
bracis-19080	320	18	scenarios	scenario	NOUN
bracis-19080	320	19	,	,	PUNCT
bracis-19080	320	20	as	as	ADP
bracis-19080	320	21	typical	typical	ADJ
bracis-19080	320	22	in	in	ADP
bracis-19080	320	23	real	real	ADJ
bracis-19080	320	24	-	-	PUNCT
bracis-19080	320	25	world	world	NOUN
bracis-19080	320	26	applications	application	NOUN
bracis-19080	320	27	.	.	PUNCT
bracis-19080	321	1	in	in	ADP
bracis-19080	321	2	the	the	DET
bracis-19080	321	3	second	second	ADJ
bracis-19080	321	4	part	part	NOUN
bracis-19080	321	5	,	,	PUNCT
bracis-19080	321	6	we	we	PRON
bracis-19080	321	7	demonstrated	demonstrate	VERB
bracis-19080	321	8	that	that	SCONJ
bracis-19080	321	9	fine	fine	ADJ
bracis-19080	321	10	tuning	tuning	NOUN
bracis-19080	321	11	sbert	sbert	NOUN
bracis-19080	321	12	led	lead	VERB
bracis-19080	321	13	to	to	ADP
bracis-19080	321	14	improvements	improvement	NOUN
bracis-19080	321	15	in	in	ADP
bracis-19080	321	16	all	all	DET
bracis-19080	321	17	previous	previous	ADJ
bracis-19080	321	18	results	result	NOUN
bracis-19080	321	19	.	.	PUNCT
bracis-19080	322	1	the	the	DET
bracis-19080	322	2	multilingual	multilingual	ADJ
bracis-19080	322	3	models	model	NOUN
bracis-19080	322	4	of	of	ADP
bracis-19080	322	5	sbert	sbert	NOUN
bracis-19080	322	6	once	once	ADV
bracis-19080	322	7	again	again	ADV
bracis-19080	322	8	stood	stand	VERB
bracis-19080	322	9	out	out	ADP
bracis-19080	322	10	and	and	CCONJ
bracis-19080	322	11	this	this	DET
bracis-19080	322	12	time	time	NOUN
bracis-19080	322	13	surpassed	surpass	VERB
bracis-19080	322	14	the	the	DET
bracis-19080	322	15	state	state	NOUN
bracis-19080	322	16	-	-	PUNCT
bracis-19080	322	17	of	of	ADP
bracis-19080	322	18	-	-	PUNCT
bracis-19080	322	19	the	the	DET
bracis-19080	322	20	-	-	PUNCT
bracis-19080	322	21	art	art	NOUN
bracis-19080	322	22	for	for	ADP
bracis-19080	322	23	the	the	DET
bracis-19080	322	24	assin	assin	PROPN
bracis-19080	322	25	 	 	SPACE
bracis-19080	322	26	2	2	NUM
bracis-19080	322	27	dataset	dataset	NOUN
bracis-19080	322	28	.	.	PUNCT
bracis-19080	323	1	further	further	ADJ
bracis-19080	323	2	analysis	analysis	NOUN
bracis-19080	323	3	indicated	indicate	VERB
bracis-19080	323	4	promising	promise	VERB
bracis-19080	323	5	future	future	ADJ
bracis-19080	323	6	work	work	NOUN
bracis-19080	323	7	possibilities	possibility	NOUN
bracis-19080	323	8	.	.	PUNCT
bracis-19080	324	1	in	in	ADP
bracis-19080	324	2	detail	detail	NOUN
bracis-19080	324	3	,	,	PUNCT
bracis-19080	324	4	we	we	PRON
bracis-19080	324	5	have	have	VERB
bracis-19080	324	6	observed	observe	VERB
bracis-19080	324	7	situations	situation	NOUN
bracis-19080	324	8	where	where	SCONJ
bracis-19080	324	9	fine	fine	ADV
bracis-19080	324	10	-	-	PUNCT
bracis-19080	324	11	tuning	tuning	NOUN
bracis-19080	324	12	worsened	worsen	VERB
bracis-19080	324	13	results	result	NOUN
bracis-19080	324	14	,	,	PUNCT
bracis-19080	324	15	an	an	DET
bracis-19080	324	16	indication	indication	NOUN
bracis-19080	324	17	that	that	SCONJ
bracis-19080	324	18	pre	pre	ADJ
bracis-19080	324	19	-	-	ADJ
bracis-19080	324	20	training	training	ADJ
bracis-19080	324	21	sbert	sbert	NOUN
bracis-19080	324	22	models	model	NOUN
bracis-19080	324	23	in	in	ADP
bracis-19080	324	24	portuguese	portuguese	PROPN
bracis-19080	324	25	could	could	AUX
bracis-19080	324	26	lead	lead	VERB
bracis-19080	324	27	to	to	ADP
bracis-19080	324	28	yet	yet	ADV
bracis-19080	324	29	better	well	ADJ
bracis-19080	324	30	results	result	NOUN
bracis-19080	324	31	.	.	PUNCT
bracis-19080	325	1	more	more	ADV
bracis-19080	325	2	importantly	importantly	ADV
bracis-19080	325	3	,	,	PUNCT
bracis-19080	325	4	we	we	PRON
bracis-19080	325	5	have	have	AUX
bracis-19080	325	6	discussed	discuss	VERB
bracis-19080	325	7	the	the	DET
bracis-19080	325	8	impact	impact	NOUN
bracis-19080	325	9	of	of	ADP
bracis-19080	325	10	language	language	NOUN
bracis-19080	325	11	variants	variant	NOUN
bracis-19080	325	12	,	,	PUNCT
bracis-19080	325	13	which	which	PRON
bracis-19080	325	14	we	we	PRON
bracis-19080	325	15	believe	believe	VERB
bracis-19080	325	16	should	should	AUX
bracis-19080	325	17	be	be	AUX
bracis-19080	325	18	taken	take	VERB
bracis-19080	325	19	into	into	ADP
bracis-19080	325	20	consideration	consideration	NOUN
bracis-19080	325	21	either	either	CCONJ
bracis-19080	325	22	for	for	ADP
bracis-19080	325	23	pre	pre	ADJ
bracis-19080	325	24	-	-	ADJ
bracis-19080	325	25	training	training	ADJ
bracis-19080	325	26	or	or	CCONJ
bracis-19080	325	27	fine	fine	ADV
bracis-19080	325	28	-	-	PUNCT
bracis-19080	325	29	tuning	tuning	NOUN
bracis-19080	325	30	.	.	PUNCT
bracis-19080	326	1	we	we	PRON
bracis-19080	326	2	evidenced	evidence	VERB
bracis-19080	326	3	this	this	DET
bracis-19080	326	4	possibility	possibility	NOUN
bracis-19080	326	5	with	with	ADP
bracis-19080	326	6	additional	additional	ADJ
bracis-19080	326	7	experiments	experiment	NOUN
bracis-19080	326	8	that	that	PRON
bracis-19080	326	9	further	far	ADV
bracis-19080	326	10	improved	improve	VERB
bracis-19080	326	11	the	the	DET
bracis-19080	326	12	performance	performance	NOUN
bracis-19080	326	13	on	on	ADP
bracis-19080	326	14	european	european	ADJ
bracis-19080	326	15	portuguese	portuguese	PROPN
bracis-19080	326	16	for	for	ADP
bracis-19080	326	17	the	the	DET
bracis-19080	326	18	best	well	ADV
bracis-19080	326	19	-	-	PUNCT
bracis-19080	326	20	performing	perform	VERB
bracis-19080	326	21	sbert	sbert	NOUN
bracis-19080	326	22	multilingual	multilingual	ADJ
bracis-19080	326	23	model	model	NOUN
bracis-19080	326	24	.	.	PUNCT
bracis-19080	327	1	the	the	DET
bracis-19080	327	2	contributions	contribution	NOUN
bracis-19080	327	3	discussed	discuss	VERB
bracis-19080	327	4	in	in	ADP
bracis-19080	327	5	this	this	DET
bracis-19080	327	6	paper	paper	NOUN
bracis-19080	327	7	further	far	ADV
bracis-19080	327	8	highlight	highlight	VERB
bracis-19080	327	9	that	that	SCONJ
bracis-19080	327	10	other	other	ADJ
bracis-19080	327	11	contextual	contextual	ADJ
bracis-19080	327	12	embeddings	embedding	NOUN
bracis-19080	327	13	such	such	ADJ
bracis-19080	327	14	as	as	ADP
bracis-19080	327	15	xlnet	xlnet	PROPN
bracis-19080	327	16	 	 	SPACE
bracis-19080	328	1	[	[	X
bracis-19080	328	2	24	24	NUM
bracis-19080	328	3	]	]	PUNCT
bracis-19080	328	4	need	need	VERB
bracis-19080	328	5	to	to	PART
bracis-19080	328	6	be	be	AUX
bracis-19080	328	7	assessed	assess	VERB
bracis-19080	328	8	in	in	ADP
bracis-19080	328	9	the	the	DET
bracis-19080	328	10	context	context	NOUN
bracis-19080	328	11	of	of	ADP
bracis-19080	328	12	sts	st	NOUN
bracis-19080	328	13	in	in	ADP
bracis-19080	328	14	portuguese	portuguese	PROPN
bracis-19080	328	15	.	.	PUNCT
bracis-19080	329	1	in	in	ADP
bracis-19080	329	2	this	this	DET
bracis-19080	329	3	sense	sense	NOUN
bracis-19080	329	4	,	,	PUNCT
bracis-19080	329	5	we	we	PRON
bracis-19080	329	6	believe	believe	VERB
bracis-19080	329	7	our	our	PRON
bracis-19080	329	8	work	work	NOUN
bracis-19080	329	9	is	be	AUX
bracis-19080	329	10	instrumental	instrumental	ADJ
bracis-19080	329	11	to	to	PART
bracis-19080	329	12	motivate	motivate	VERB
bracis-19080	329	13	more	more	ADJ
bracis-19080	329	14	research	research	NOUN
bracis-19080	329	15	groups	group	NOUN
bracis-19080	329	16	addressing	address	VERB
bracis-19080	329	17	nlp	nlp	ADJ
bracis-19080	329	18	problems	problem	NOUN
bracis-19080	329	19	in	in	ADP
bracis-19080	329	20	portuguese	portuguese	PROPN
bracis-19080	329	21	.	.	PUNCT
bracis-19080	330	1	this	this	PRON
bracis-19080	330	2	is	be	AUX
bracis-19080	330	3	the	the	DET
bracis-19080	330	4	purpose	purpose	NOUN
bracis-19080	330	5	of	of	ADP
bracis-19080	330	6	shared	share	VERB
bracis-19080	330	7	tasks	task	NOUN
bracis-19080	330	8	,	,	PUNCT
bracis-19080	330	9	especially	especially	ADV
bracis-19080	330	10	given	give	VERB
bracis-19080	330	11	the	the	DET
bracis-19080	330	12	computational	computational	ADJ
bracis-19080	330	13	cost	cost	NOUN
bracis-19080	330	14	of	of	ADP
bracis-19080	330	15	the	the	DET
bracis-19080	330	16	experimental	experimental	ADJ
bracis-19080	330	17	campaigns	campaign	NOUN
bracis-19080	330	18	involved	involve	VERB
bracis-19080	330	19	in	in	ADP
bracis-19080	330	20	deep	deep	ADJ
bracis-19080	330	21	learning	learning	NOUN
bracis-19080	330	22	assessment	assessment	NOUN
bracis-19080	330	23	for	for	ADP
bracis-19080	330	24	nlp	nlp	NOUN
bracis-19080	330	25	applications	application	NOUN
bracis-19080	330	26	.	.	PUNCT
bracis-19080	331	1	even	even	ADV
bracis-19080	331	2	more	more	ADV
bracis-19080	331	3	interesting	interesting	ADJ
bracis-19080	331	4	is	be	AUX
bracis-19080	331	5	stirring	stir	VERB
bracis-19080	331	6	the	the	DET
bracis-19080	331	7	interest	interest	NOUN
bracis-19080	331	8	of	of	ADP
bracis-19080	331	9	researchers	researcher	NOUN
bracis-19080	331	10	proposing	propose	VERB
bracis-19080	331	11	novel	novel	ADJ
bracis-19080	331	12	nlp	nlp	NOUN
bracis-19080	331	13	models	model	NOUN
bracis-19080	331	14	to	to	ADP
bracis-19080	331	15	the	the	DET
bracis-19080	331	16	portuguese	portuguese	ADJ
bracis-19080	331	17	language	language	NOUN
bracis-19080	331	18	,	,	PUNCT
bracis-19080	331	19	given	give	VERB
bracis-19080	331	20	that	that	SCONJ
bracis-19080	331	21	the	the	DET
bracis-19080	331	22	differences	difference	NOUN
bracis-19080	331	23	between	between	ADP
bracis-19080	331	24	portuguese	portuguese	ADJ
bracis-19080	331	25	variants	variant	NOUN
bracis-19080	331	26	impacts	impact	NOUN
bracis-19080	331	27	on	on	ADP
bracis-19080	331	28	the	the	DET
bracis-19080	331	29	results	result	NOUN
bracis-19080	331	30	for	for	ADP
bracis-19080	331	31	most	most	ADJ
bracis-19080	331	32	models	model	NOUN
bracis-19080	331	33	considered	consider	VERB
bracis-19080	331	34	.	.	PUNCT
bracis-19080	332	1	notes	note	NOUN
bracis-19080	332	2	1.for	1.for	ADP
bracis-19080	332	3	brevity	brevity	NOUN
bracis-19080	332	4	,	,	PUNCT
bracis-19080	332	5	other	other	ADJ
bracis-19080	332	6	relevant	relevant	ADJ
bracis-19080	332	7	works	work	NOUN
bracis-19080	332	8	such	such	ADJ
bracis-19080	332	9	as	as	ADP
bracis-19080	332	10	 	 	SPACE
bracis-19080	332	11	[	[	X
bracis-19080	332	12	19	19	NUM
bracis-19080	332	13	]	]	PUNCT
bracis-19080	332	14	comparing	compare	VERB
bracis-19080	332	15	word2vec	word2vec	NOUN
bracis-19080	332	16	,	,	PUNCT
bracis-19080	332	17	fasttext	fasttext	PROPN
bracis-19080	332	18	,	,	PUNCT
bracis-19080	332	19	elmo	elmo	PROPN
bracis-19080	332	20	,	,	PUNCT
bracis-19080	332	21	and	and	CCONJ
bracis-19080	332	22	bert	bert	PROPN
bracis-19080	332	23	on	on	ADP
bracis-19080	332	24	assin	assin	PROPN
bracis-19080	332	25	are	be	AUX
bracis-19080	332	26	not	not	PART
bracis-19080	332	27	included	include	VERB
bracis-19080	332	28	as	as	SCONJ
bracis-19080	332	29	their	their	PRON
bracis-19080	332	30	results	result	NOUN
bracis-19080	332	31	are	be	AUX
bracis-19080	332	32	surpassed	surpass	VERB
bracis-19080	332	33	by	by	ADP
bracis-19080	332	34	 	 	SPACE
bracis-19080	332	35	[	[	X
bracis-19080	332	36	8	8	NUM
bracis-19080	332	37	]	]	PUNCT
bracis-19080	332	38	.	.	PUNCT
bracis-19080	333	1	2.though	2.though	NUM
bracis-19080	333	2	models	model	NOUN
bracis-19080	333	3	based	base	VERB
bracis-19080	333	4	on	on	ADP
bracis-19080	333	5	other	other	ADJ
bracis-19080	333	6	relevant	relevant	ADJ
bracis-19080	333	7	architectures	architecture	NOUN
bracis-19080	333	8	such	such	ADJ
bracis-19080	333	9	as	as	ADP
bracis-19080	333	10	the	the	DET
bracis-19080	333	11	multilingual	multilingual	ADJ
bracis-19080	333	12	universal	universal	ADJ
bracis-19080	333	13	sentence	sentence	NOUN
bracis-19080	333	14	encoder	encoder	NOUN
bracis-19080	333	15	 	 	SPACE
bracis-19080	334	1	[	[	X
bracis-19080	334	2	23	23	NUM
bracis-19080	334	3	]	]	PUNCT
bracis-19080	334	4	were	be	AUX
bracis-19080	334	5	available	available	ADJ
bracis-19080	334	6	at	at	ADP
bracis-19080	334	7	the	the	DET
bracis-19080	334	8	sbert	sbert	NOUN
bracis-19080	334	9	repository	repository	NOUN
bracis-19080	334	10	,	,	PUNCT
bracis-19080	334	11	we	we	PRON
bracis-19080	334	12	did	do	AUX
bracis-19080	334	13	not	not	PART
bracis-19080	334	14	include	include	VERB
bracis-19080	334	15	them	they	PRON
bracis-19080	334	16	in	in	ADP
bracis-19080	334	17	our	our	PRON
bracis-19080	334	18	work	work	NOUN
bracis-19080	334	19	due	due	ADP
bracis-19080	334	20	to	to	ADP
bracis-19080	334	21	the	the	DET
bracis-19080	334	22	lack	lack	NOUN
bracis-19080	334	23	of	of	ADP
bracis-19080	334	24	training	training	NOUN
bracis-19080	334	25	setup	setup	NOUN
bracis-19080	334	26	details	detail	NOUN
bracis-19080	334	27	.	.	PUNCT
bracis-19080	335	1	references	reference	NOUN
bracis-19080	335	2	agirre	agirre	PROPN
bracis-19080	335	3	,	,	PUNCT
bracis-19080	335	4	e.	e.	PROPN
bracis-19080	335	5	,	,	PUNCT
bracis-19080	335	6	diab	diab	PROPN
bracis-19080	335	7	,	,	PUNCT
bracis-19080	335	8	m.	m.	NOUN
bracis-19080	335	9	,	,	PUNCT
bracis-19080	335	10	cer	cer	PROPN
bracis-19080	335	11	,	,	PUNCT
bracis-19080	335	12	d.	d.	PROPN
bracis-19080	335	13	,	,	PUNCT
bracis-19080	335	14	gonzalez	gonzalez	PROPN
bracis-19080	335	15	-	-	PUNCT
bracis-19080	335	16	agirre	agirre	PROPN
bracis-19080	335	17	,	,	PUNCT
bracis-19080	335	18	a.	a.	NOUN
bracis-19080	335	19	:	:	PUNCT
bracis-19080	335	20	semeval-2012	semeval-2012	ADJ
bracis-19080	335	21	task	task	NOUN
bracis-19080	335	22	6	6	NUM
bracis-19080	335	23	:	:	PUNCT
bracis-19080	335	24	a	a	DET
bracis-19080	335	25	pilot	pilot	NOUN
bracis-19080	335	26	on	on	ADP
bracis-19080	335	27	semantic	semantic	ADJ
bracis-19080	335	28	textual	textual	ADJ
bracis-19080	335	29	similarity	similarity	NOUN
bracis-19080	335	30	.	.	PUNCT
bracis-19080	336	1	in	in	ADP
bracis-19080	336	2	:	:	PUNCT
bracis-19080	336	3	semeval	semeval	NOUN
bracis-19080	336	4	,	,	PUNCT
bracis-19080	336	5	pp	pp	ADV
bracis-19080	336	6	.	.	PUNCT
bracis-19080	337	1	385–393	385–393	NUM
bracis-19080	337	2	.	.	PUNCT
bracis-19080	338	1	acl	acl	PROPN
bracis-19080	338	2	,	,	PUNCT
bracis-19080	338	3	usa	usa	PROPN
bracis-19080	338	4	(	(	PUNCT
bracis-19080	338	5	2012	2012	NUM
bracis-19080	338	6	)	)	PUNCT
bracis-19080	338	7	google	google	PROPN
bracis-19080	338	8	scholar	scholar	NOUN
bracis-19080	338	9	  	  	SPACE
bracis-19080	338	10	andrade	andrade	PROPN
bracis-19080	338	11	,	,	PUNCT
bracis-19080	338	12	j.	j.	PROPN
bracis-19080	338	13	,	,	PUNCT
bracis-19080	338	14	bezerra	bezerra	PROPN
bracis-19080	338	15	,	,	PUNCT
bracis-19080	338	16	l.c.t	l.c.t	PROPN
bracis-19080	338	17	.	.	PUNCT
bracis-19080	338	18	,	,	PUNCT
bracis-19080	338	19	cardoso	cardoso	PROPN
bracis-19080	338	20	-	-	PUNCT
bracis-19080	338	21	silva	silva	NOUN
bracis-19080	338	22	,	,	PUNCT
bracis-19080	338	23	j.	j.	PROPN
bracis-19080	338	24	:	:	PUNCT
bracis-19080	338	25	comparing	compare	VERB
bracis-19080	338	26	contextual	contextual	ADJ
bracis-19080	338	27	embeddings	embedding	NOUN
bracis-19080	338	28	for	for	ADP
bracis-19080	338	29	semantic	semantic	ADJ
bracis-19080	338	30	textual	textual	ADJ
bracis-19080	338	31	similarity	similarity	NOUN
bracis-19080	338	32	in	in	ADP
bracis-19080	338	33	portuguese	portuguese	PROPN
bracis-19080	338	34	(	(	PUNCT
bracis-19080	338	35	supplementary	supplementary	ADJ
bracis-19080	338	36	material	material	NOUN
bracis-19080	338	37	)	)	PUNCT
bracis-19080	338	38	(	(	PUNCT
bracis-19080	338	39	2021	2021	NUM
bracis-19080	338	40	)	)	PUNCT
bracis-19080	338	41	.	.	PUNCT
bracis-19080	339	1	https://github.com/andradejunior/bracis-2021-supp-material	https://github.com/andradejunior/bracis-2021-supp-material	PROPN
bracis-19080	339	2	bengio	bengio	PROPN
bracis-19080	339	3	,	,	PUNCT
bracis-19080	339	4	y.	y.	PROPN
bracis-19080	339	5	,	,	PUNCT
bracis-19080	339	6	ducharme	ducharme	PROPN
bracis-19080	339	7	,	,	PUNCT
bracis-19080	339	8	r.	r.	PROPN
bracis-19080	339	9	,	,	PUNCT
bracis-19080	339	10	vincent	vincent	PROPN
bracis-19080	339	11	,	,	PUNCT
bracis-19080	339	12	p.	p.	PROPN
bracis-19080	339	13	,	,	PUNCT
bracis-19080	339	14	janvin	janvin	PROPN
bracis-19080	339	15	,	,	PUNCT
bracis-19080	339	16	c.	c.	PROPN
bracis-19080	339	17	:	:	PUNCT
bracis-19080	339	18	a	a	DET
bracis-19080	339	19	neural	neural	ADJ
bracis-19080	339	20	probabilistic	probabilistic	ADJ
bracis-19080	339	21	language	language	NOUN
bracis-19080	339	22	model	model	NOUN
bracis-19080	339	23	.	.	PUNCT
bracis-19080	340	1	j.	j.	PROPN
bracis-19080	340	2	mach	mach	PROPN
bracis-19080	340	3	.	.	PUNCT
bracis-19080	341	1	learn	learn	VERB
bracis-19080	341	2	.	.	PUNCT
bracis-19080	342	1	res	re	NOUN
bracis-19080	342	2	.	.	PROPN
bracis-19080	343	1	3	3	NUM
bracis-19080	343	2	,	,	PUNCT
bracis-19080	343	3	1137–1155	1137–1155	NUM
bracis-19080	343	4	(	(	PUNCT
bracis-19080	343	5	2003	2003	NUM
bracis-19080	343	6	)	)	PUNCT
bracis-19080	343	7	math	math	NOUN
bracis-19080	343	8	  	  	SPACE
bracis-19080	343	9	google	google	PROPN
bracis-19080	343	10	scholar	scholar	NOUN
bracis-19080	343	11	  	  	SPACE
bracis-19080	343	12	bojanowski	bojanowski	NOUN
bracis-19080	343	13	,	,	PUNCT
bracis-19080	343	14	p.	p.	NOUN
bracis-19080	343	15	,	,	PUNCT
bracis-19080	343	16	grave	grave	NOUN
bracis-19080	343	17	,	,	PUNCT
bracis-19080	343	18	e.	e.	PROPN
bracis-19080	343	19	,	,	PUNCT
bracis-19080	343	20	joulin	joulin	PROPN
bracis-19080	343	21	,	,	PUNCT
bracis-19080	343	22	a.	a.	NOUN
bracis-19080	343	23	,	,	PUNCT
bracis-19080	343	24	mikolov	mikolov	ADV
bracis-19080	343	25	,	,	PUNCT
bracis-19080	343	26	t.	t.	PROPN
bracis-19080	343	27	:	:	PUNCT
bracis-19080	343	28	enriching	enrich	VERB
bracis-19080	343	29	word	word	NOUN
bracis-19080	343	30	vectors	vector	NOUN
bracis-19080	343	31	with	with	ADP
bracis-19080	343	32	subword	subword	PROPN
bracis-19080	343	33	information	information	PROPN
bracis-19080	343	34	.	.	PUNCT
bracis-19080	344	1	trans	trans	PROPN
bracis-19080	344	2	.	.	PUNCT
bracis-19080	345	1	acl	acl	PROPN
bracis-19080	345	2	5	5	NUM
bracis-19080	345	3	,	,	PUNCT
bracis-19080	345	4	135–146	135–146	NUM
bracis-19080	345	5	(	(	PUNCT
bracis-19080	345	6	2017	2017	NUM
bracis-19080	345	7	)	)	PUNCT
bracis-19080	345	8	google	google	PROPN
bracis-19080	345	9	scholar	scholar	NOUN
bracis-19080	345	10	  	  	SPACE
bracis-19080	345	11	cer	cer	PROPN
bracis-19080	345	12	,	,	PUNCT
bracis-19080	345	13	d.	d.	PROPN
bracis-19080	345	14	,	,	PUNCT
bracis-19080	345	15	diab	diab	PROPN
bracis-19080	345	16	,	,	PUNCT
bracis-19080	345	17	m.	m.	NOUN
bracis-19080	345	18	,	,	PUNCT
bracis-19080	345	19	agirre	agirre	PROPN
bracis-19080	345	20	,	,	PUNCT
bracis-19080	345	21	e.	e.	PROPN
bracis-19080	345	22	,	,	PUNCT
bracis-19080	345	23	lopez	lopez	NOUN
bracis-19080	345	24	-	-	PUNCT
bracis-19080	345	25	gazpio	gazpio	NOUN
bracis-19080	345	26	,	,	PUNCT
bracis-19080	345	27	i.	i.	NOUN
bracis-19080	345	28	,	,	PUNCT
bracis-19080	345	29	specia	specia	NOUN
bracis-19080	345	30	,	,	PUNCT
bracis-19080	345	31	l.	l.	PROPN
bracis-19080	345	32	:	:	PUNCT
bracis-19080	345	33	semeval-2017	semeval-2017	PROPN
bracis-19080	345	34	task	task	NOUN
bracis-19080	345	35	1	1	NUM
bracis-19080	345	36	:	:	PUNCT
bracis-19080	345	37	semantic	semantic	ADJ
bracis-19080	345	38	textual	textual	ADJ
bracis-19080	345	39	similarity	similarity	NOUN
bracis-19080	345	40	multilingual	multilingual	ADJ
bracis-19080	345	41	and	and	CCONJ
bracis-19080	345	42	crosslingual	crosslingual	ADJ
bracis-19080	345	43	focused	focused	ADJ
bracis-19080	345	44	evaluation	evaluation	NOUN
bracis-19080	345	45	.	.	PUNCT
bracis-19080	346	1	in	in	ADP
bracis-19080	346	2	:	:	PUNCT
bracis-19080	346	3	semeval	semeval	NOUN
bracis-19080	346	4	,	,	PUNCT
bracis-19080	346	5	pp	pp	PROPN
bracis-19080	346	6	.	.	PUNCT
bracis-19080	347	1	1–14	1–14	PROPN
bracis-19080	347	2	.	.	PUNCT
bracis-19080	348	1	acl	acl	PROPN
bracis-19080	348	2	,	,	PUNCT
bracis-19080	348	3	vancouver	vancouver	PROPN
bracis-19080	348	4	(	(	PUNCT
bracis-19080	348	5	2017	2017	NUM
bracis-19080	348	6	)	)	PUNCT
bracis-19080	348	7	google	google	NOUN
bracis-19080	348	8	scholar	scholar	NOUN
bracis-19080	348	9	  	  	SPACE
bracis-19080	348	10	conneau	conneau	NOUN
bracis-19080	348	11	,	,	PUNCT
bracis-19080	348	12	a.	a.	PROPN
bracis-19080	348	13	,	,	PUNCT
bracis-19080	348	14	et	et	PROPN
bracis-19080	348	15	al	al	PROPN
bracis-19080	348	16	.	.	PROPN
bracis-19080	348	17	:	:	PUNCT
bracis-19080	349	1	unsupervised	unsupervised	ADJ
bracis-19080	349	2	cross	cross	ADJ
bracis-19080	349	3	-	-	ADJ
bracis-19080	349	4	lingual	lingual	ADJ
bracis-19080	349	5	representation	representation	NOUN
bracis-19080	349	6	learning	learn	VERB
bracis-19080	349	7	at	at	ADP
bracis-19080	349	8	scale	scale	NOUN
bracis-19080	349	9	.	.	PUNCT
bracis-19080	350	1	in	in	ADP
bracis-19080	350	2	:	:	PUNCT
bracis-19080	350	3	acl	acl	PROPN
bracis-19080	350	4	,	,	PUNCT
bracis-19080	350	5	pp	pp	PROPN
bracis-19080	350	6	.	.	PUNCT
bracis-19080	350	7	8440–8451	8440–8451	NUM
bracis-19080	350	8	.	.	PUNCT
bracis-19080	351	1	acl	acl	PROPN
bracis-19080	351	2	(	(	PUNCT
bracis-19080	351	3	2020	2020	NUM
bracis-19080	351	4	)	)	PUNCT
bracis-19080	351	5	google	google	PROPN
bracis-19080	351	6	scholar	scholar	NOUN
bracis-19080	351	7	  	  	SPACE
bracis-19080	351	8	devlin	devlin	PROPN
bracis-19080	351	9	,	,	PUNCT
bracis-19080	351	10	j.	j.	PROPN
bracis-19080	351	11	,	,	PUNCT
bracis-19080	351	12	chang	chang	PROPN
bracis-19080	351	13	,	,	PUNCT
bracis-19080	351	14	m.w	m.w	PROPN
bracis-19080	351	15	.	.	PROPN
bracis-19080	351	16	,	,	PUNCT
bracis-19080	351	17	lee	lee	PROPN
bracis-19080	351	18	,	,	PUNCT
bracis-19080	351	19	k.	k.	PROPN
bracis-19080	351	20	,	,	PUNCT
bracis-19080	351	21	toutanova	toutanova	PROPN
bracis-19080	351	22	,	,	PUNCT
bracis-19080	351	23	k.	k.	PROPN
bracis-19080	351	24	:	:	PUNCT
bracis-19080	352	1	bert	bert	PROPN
bracis-19080	352	2	:	:	PUNCT
bracis-19080	352	3	pre	pre	ADJ
bracis-19080	352	4	-	-	NOUN
bracis-19080	352	5	training	training	NOUN
bracis-19080	352	6	of	of	ADP
bracis-19080	352	7	deep	deep	ADJ
bracis-19080	352	8	bidirectional	bidirectional	ADJ
bracis-19080	352	9	transformers	transformer	NOUN
bracis-19080	352	10	for	for	ADP
bracis-19080	352	11	language	language	NOUN
bracis-19080	352	12	understanding	understanding	NOUN
bracis-19080	352	13	.	.	PUNCT
bracis-19080	353	1	in	in	ADP
bracis-19080	353	2	:	:	PUNCT
bracis-19080	353	3	naacl	naacl	PROPN
bracis-19080	353	4	,	,	PUNCT
bracis-19080	353	5	pp	pp	ADJ
bracis-19080	353	6	.	.	PUNCT
bracis-19080	354	1	4171–4186	4171–4186	NUM
bracis-19080	354	2	.	.	PUNCT
bracis-19080	355	1	acl	acl	PROPN
bracis-19080	355	2	,	,	PUNCT
bracis-19080	355	3	minneapolis	minneapolis	PROPN
bracis-19080	355	4	(	(	PUNCT
bracis-19080	355	5	2019	2019	NUM
bracis-19080	355	6	)	)	PUNCT
bracis-19080	355	7	google	google	NOUN
bracis-19080	355	8	scholar	scholar	NOUN
bracis-19080	355	9	  	  	SPACE
bracis-19080	355	10	fialho	fialho	NOUN
bracis-19080	355	11	,	,	PUNCT
bracis-19080	355	12	p.	p.	PROPN
bracis-19080	355	13	,	,	PUNCT
bracis-19080	355	14	coheur	coheur	PROPN
bracis-19080	355	15	,	,	PUNCT
bracis-19080	355	16	l.	l.	PROPN
bracis-19080	355	17	,	,	PUNCT
bracis-19080	355	18	quaresma	quaresma	PROPN
bracis-19080	355	19	,	,	PUNCT
bracis-19080	355	20	p.	p.	NOUN
bracis-19080	355	21	:	:	PUNCT
bracis-19080	355	22	benchmarking	benchmarke	VERB
bracis-19080	355	23	natural	natural	ADJ
bracis-19080	355	24	language	language	NOUN
bracis-19080	355	25	inference	inference	NOUN
bracis-19080	355	26	and	and	CCONJ
bracis-19080	355	27	semantic	semantic	ADJ
bracis-19080	355	28	textual	textual	ADJ
bracis-19080	355	29	similarity	similarity	NOUN
bracis-19080	355	30	for	for	ADP
bracis-19080	355	31	portuguese	portuguese	PROPN
bracis-19080	355	32	.	.	PUNCT
bracis-19080	356	1	information	information	NOUN
bracis-19080	356	2	11	11	NUM
bracis-19080	356	3	,	,	PUNCT
bracis-19080	356	4	484	484	NUM
bracis-19080	356	5	(	(	PUNCT
bracis-19080	356	6	2020	2020	NUM
bracis-19080	356	7	)	)	PUNCT
bracis-19080	356	8	article	article	NOUN
bracis-19080	356	9	  	  	SPACE
bracis-19080	356	10	google	google	PROPN
bracis-19080	356	11	scholar	scholar	NOUN
bracis-19080	356	12	  	  	SPACE
bracis-19080	356	13	fonseca	fonseca	PROPN
bracis-19080	356	14	,	,	PUNCT
bracis-19080	356	15	e.r	e.r	PROPN
bracis-19080	356	16	.	.	PROPN
bracis-19080	356	17	,	,	PUNCT
bracis-19080	356	18	borges	borges	PROPN
bracis-19080	356	19	dos	dos	PROPN
bracis-19080	356	20	santos	santos	PROPN
bracis-19080	356	21	,	,	PUNCT
bracis-19080	356	22	l.	l.	PROPN
bracis-19080	356	23	,	,	PUNCT
bracis-19080	356	24	criscuolo	criscuolo	PROPN
bracis-19080	356	25	,	,	PUNCT
bracis-19080	356	26	m.	m.	NOUN
bracis-19080	356	27	,	,	PUNCT
bracis-19080	356	28	aluísio	aluísio	NOUN
bracis-19080	356	29	,	,	PUNCT
bracis-19080	356	30	s.m	s.m	PROPN
bracis-19080	356	31	.	.	PROPN
bracis-19080	356	32	:	:	PUNCT
bracis-19080	357	1	visão	visão	PROPN
bracis-19080	357	2	geral	geral	PROPN
bracis-19080	357	3	da	da	PROPN
bracis-19080	357	4	avaliação	avaliação	PROPN
bracis-19080	357	5	de	de	PROPN
bracis-19080	357	6	similaridade	similaridade	PROPN
bracis-19080	357	7	semântica	semântica	PROPN
bracis-19080	357	8	e	e	PROPN
bracis-19080	357	9	inferência	inferência	PROPN
bracis-19080	357	10	textual	textual	ADJ
bracis-19080	357	11	.	.	PUNCT
bracis-19080	358	1	linguamática	linguamática	NOUN
bracis-19080	358	2	8(2	8(2	NUM
bracis-19080	358	3	)	)	PUNCT
bracis-19080	358	4	,	,	PUNCT
bracis-19080	358	5	3–13	3–13	X
bracis-19080	358	6	(	(	PUNCT
bracis-19080	358	7	2016	2016	NUM
bracis-19080	358	8	)	)	PUNCT
bracis-19080	358	9	google	google	PROPN
bracis-19080	358	10	scholar	scholar	NOUN
bracis-19080	358	11	  	  	SPACE
bracis-19080	358	12	gou	gou	PROPN
bracis-19080	358	13	,	,	PUNCT
bracis-19080	358	14	j.	j.	PROPN
bracis-19080	358	15	,	,	PUNCT
bracis-19080	358	16	yu	yu	PROPN
bracis-19080	358	17	,	,	PUNCT
bracis-19080	358	18	b.	b.	PROPN
bracis-19080	358	19	,	,	PUNCT
bracis-19080	358	20	maybank	maybank	PROPN
bracis-19080	358	21	,	,	PUNCT
bracis-19080	358	22	s.j	s.j	PROPN
bracis-19080	358	23	.	.	PROPN
bracis-19080	358	24	,	,	PUNCT
bracis-19080	358	25	tao	tao	PROPN
bracis-19080	358	26	,	,	PUNCT
bracis-19080	358	27	d.	d.	PROPN
bracis-19080	358	28	:	:	PUNCT
bracis-19080	358	29	knowledge	knowledge	NOUN
bracis-19080	358	30	distillation	distillation	NOUN
bracis-19080	358	31	:	:	PUNCT
bracis-19080	358	32	a	a	DET
bracis-19080	358	33	survey	survey	NOUN
bracis-19080	358	34	.	.	PUNCT
bracis-19080	359	1	int	int	NOUN
bracis-19080	359	2	.	.	PUNCT
bracis-19080	360	1	j.	j.	PROPN
bracis-19080	360	2	comput	comput	PROPN
bracis-19080	360	3	.	.	PUNCT
bracis-19080	361	1	vis	vis	X
bracis-19080	361	2	.	.	PROPN
bracis-19080	361	3	129(6	129(6	NUM
bracis-19080	361	4	)	)	PUNCT
bracis-19080	361	5	,	,	PUNCT
bracis-19080	361	6	1789–1819	1789–1819	NUM
bracis-19080	361	7	(	(	PUNCT
bracis-19080	361	8	2021	2021	NUM
bracis-19080	361	9	)	)	PUNCT
bracis-19080	361	10	article	article	NOUN
bracis-19080	361	11	  	  	SPACE
bracis-19080	361	12	google	google	PROPN
bracis-19080	361	13	scholar	scholar	NOUN
bracis-19080	361	14	  	  	SPACE
bracis-19080	361	15	jurafsky	jurafsky	PROPN
bracis-19080	361	16	,	,	PUNCT
bracis-19080	361	17	d.	d.	PROPN
bracis-19080	361	18	,	,	PUNCT
bracis-19080	361	19	martin	martin	PROPN
bracis-19080	361	20	,	,	PUNCT
bracis-19080	361	21	j.h	j.h	PROPN
bracis-19080	361	22	.	.	PROPN
bracis-19080	361	23	:	:	PUNCT
bracis-19080	361	24	speech	speech	NOUN
bracis-19080	361	25	and	and	CCONJ
bracis-19080	361	26	language	language	NOUN
bracis-19080	361	27	processing	processing	NOUN
bracis-19080	361	28	,	,	PUNCT
bracis-19080	361	29	2nd	2nd	ADJ
bracis-19080	361	30	edn	edn	PROPN
bracis-19080	361	31	.	.	PUNCT
bracis-19080	362	1	prentice	prentice	NOUN
bracis-19080	362	2	-	-	PUNCT
bracis-19080	362	3	hall	hall	PROPN
bracis-19080	362	4	inc	inc	PROPN
bracis-19080	362	5	.	.	PROPN
bracis-19080	362	6	,	,	PUNCT
bracis-19080	362	7	upper	upper	ADJ
bracis-19080	362	8	saddle	saddle	NOUN
bracis-19080	362	9	river	river	NOUN
bracis-19080	362	10	(	(	PUNCT
bracis-19080	362	11	2009	2009	NUM
bracis-19080	362	12	)	)	PUNCT
bracis-19080	362	13	google	google	PROPN
bracis-19080	362	14	scholar	scholar	NOUN
bracis-19080	362	15	  	  	SPACE
bracis-19080	362	16	marelli	marelli	PROPN
bracis-19080	362	17	,	,	PUNCT
bracis-19080	362	18	m.	m.	NOUN
bracis-19080	362	19	,	,	PUNCT
bracis-19080	362	20	bentivogli	bentivogli	PROPN
bracis-19080	362	21	,	,	PUNCT
bracis-19080	362	22	l.	l.	PROPN
bracis-19080	362	23	,	,	PUNCT
bracis-19080	362	24	baroni	baroni	PROPN
bracis-19080	362	25	,	,	PUNCT
bracis-19080	362	26	m.	m.	NOUN
bracis-19080	362	27	,	,	PUNCT
bracis-19080	362	28	bernardi	bernardi	PROPN
bracis-19080	362	29	,	,	PUNCT
bracis-19080	362	30	r.	r.	PROPN
bracis-19080	362	31	,	,	PUNCT
bracis-19080	362	32	menini	menini	PROPN
bracis-19080	362	33	,	,	PUNCT
bracis-19080	362	34	s.	s.	PROPN
bracis-19080	362	35	,	,	PUNCT
bracis-19080	362	36	zamparelli	zamparelli	PROPN
bracis-19080	362	37	,	,	PUNCT
bracis-19080	362	38	r.	r.	PROPN
bracis-19080	362	39	:	:	PUNCT
bracis-19080	362	40	semeval-2014	semeval-2014	NOUN
bracis-19080	362	41	task	task	NOUN
bracis-19080	362	42	1	1	NUM
bracis-19080	362	43	:	:	PUNCT
bracis-19080	362	44	evaluation	evaluation	NOUN
bracis-19080	362	45	of	of	ADP
bracis-19080	362	46	compositional	compositional	ADJ
bracis-19080	362	47	distributional	distributional	ADJ
bracis-19080	362	48	semantic	semantic	ADJ
bracis-19080	362	49	models	model	NOUN
bracis-19080	362	50	on	on	ADP
bracis-19080	362	51	full	full	ADJ
bracis-19080	362	52	sentences	sentence	NOUN
bracis-19080	362	53	through	through	ADP
bracis-19080	362	54	semantic	semantic	ADJ
bracis-19080	362	55	relatedness	relatedness	NOUN
bracis-19080	362	56	and	and	CCONJ
bracis-19080	362	57	textual	textual	ADJ
bracis-19080	362	58	entailment	entailment	NOUN
bracis-19080	362	59	.	.	PUNCT
bracis-19080	363	1	in	in	ADP
bracis-19080	363	2	:	:	PUNCT
bracis-19080	363	3	semeval	semeval	NOUN
bracis-19080	363	4	,	,	PUNCT
bracis-19080	363	5	pp	pp	ADP
bracis-19080	363	6	.	.	PUNCT
bracis-19080	364	1	1–8	1–8	X
bracis-19080	364	2	.	.	PUNCT
bracis-19080	365	1	acl	acl	PROPN
bracis-19080	365	2	,	,	PUNCT
bracis-19080	365	3	dublin	dublin	PROPN
bracis-19080	365	4	(	(	PUNCT
bracis-19080	365	5	2014	2014	NUM
bracis-19080	365	6	)	)	PUNCT
bracis-19080	365	7	google	google	PROPN
bracis-19080	365	8	scholar	scholar	NOUN
bracis-19080	365	9	  	  	SPACE
bracis-19080	365	10	mikolov	mikolov	NOUN
bracis-19080	365	11	,	,	PUNCT
bracis-19080	365	12	t.	t.	PROPN
bracis-19080	365	13	,	,	PUNCT
bracis-19080	365	14	sutskever	sutskever	PROPN
bracis-19080	365	15	,	,	PUNCT
bracis-19080	365	16	i.	i.	PROPN
bracis-19080	365	17	,	,	PUNCT
bracis-19080	365	18	chen	chen	PROPN
bracis-19080	365	19	,	,	PUNCT
bracis-19080	365	20	k.	k.	PROPN
bracis-19080	365	21	,	,	PUNCT
bracis-19080	365	22	corrado	corrado	PROPN
bracis-19080	365	23	,	,	PUNCT
bracis-19080	365	24	g.	g.	PROPN
bracis-19080	365	25	,	,	PUNCT
bracis-19080	365	26	dean	dean	PROPN
bracis-19080	365	27	,	,	PUNCT
bracis-19080	365	28	j.	j.	PROPN
bracis-19080	365	29	:	:	PUNCT
bracis-19080	365	30	distributed	distribute	VERB
bracis-19080	365	31	representations	representation	NOUN
bracis-19080	365	32	of	of	ADP
bracis-19080	365	33	words	word	NOUN
bracis-19080	365	34	and	and	CCONJ
bracis-19080	365	35	phrases	phrase	NOUN
bracis-19080	365	36	and	and	CCONJ
bracis-19080	365	37	their	their	PRON
bracis-19080	365	38	compositionality	compositionality	NOUN
bracis-19080	365	39	.	.	PUNCT
bracis-19080	366	1	in	in	ADP
bracis-19080	366	2	:	:	PUNCT
bracis-19080	366	3	neurips	neurip	NOUN
bracis-19080	366	4	,	,	PUNCT
bracis-19080	366	5	pp	pp	ADV
bracis-19080	366	6	.	.	PUNCT
bracis-19080	366	7	3111–3119	3111–3119	NUM
bracis-19080	366	8	.	.	PUNCT
bracis-19080	367	1	curran	curran	PROPN
bracis-19080	367	2	associates	associates	PROPN
bracis-19080	367	3	inc	inc	PROPN
bracis-19080	367	4	.	.	PROPN
bracis-19080	367	5	,	,	PUNCT
bracis-19080	367	6	red	red	ADJ
bracis-19080	367	7	hook	hook	NOUN
bracis-19080	367	8	(	(	PUNCT
bracis-19080	367	9	2013	2013	NUM
bracis-19080	367	10	)	)	PUNCT
bracis-19080	367	11	google	google	PROPN
bracis-19080	367	12	scholar	scholar	NOUN
bracis-19080	367	13	  	  	SPACE
bracis-19080	367	14	pennington	pennington	PROPN
bracis-19080	367	15	,	,	PUNCT
bracis-19080	367	16	j.	j.	PROPN
bracis-19080	367	17	,	,	PUNCT
bracis-19080	367	18	socher	socher	PROPN
bracis-19080	367	19	,	,	PUNCT
bracis-19080	367	20	r.	r.	PROPN
bracis-19080	367	21	,	,	PUNCT
bracis-19080	367	22	manning	manning	NOUN
bracis-19080	367	23	,	,	PUNCT
bracis-19080	367	24	c.	c.	NOUN
bracis-19080	367	25	:	:	PUNCT
bracis-19080	367	26	glove	glove	NOUN
bracis-19080	367	27	:	:	PUNCT
bracis-19080	367	28	global	global	ADJ
bracis-19080	367	29	vectors	vector	NOUN
bracis-19080	367	30	for	for	ADP
bracis-19080	367	31	word	word	NOUN
bracis-19080	367	32	representation	representation	NOUN
bracis-19080	367	33	.	.	PUNCT
bracis-19080	368	1	in	in	ADP
bracis-19080	368	2	:	:	PUNCT
bracis-19080	368	3	emnlp	emnlp	ADV
bracis-19080	368	4	,	,	PUNCT
bracis-19080	368	5	pp	pp	ADJ
bracis-19080	368	6	.	.	PUNCT
bracis-19080	368	7	1532–1543	1532–1543	NUM
bracis-19080	368	8	.	.	PUNCT
bracis-19080	369	1	acl	acl	PROPN
bracis-19080	369	2	,	,	PUNCT
bracis-19080	369	3	doha	doha	PROPN
bracis-19080	369	4	,	,	PUNCT
bracis-19080	369	5	qatar	qatar	PROPN
bracis-19080	369	6	(	(	PUNCT
bracis-19080	369	7	2014	2014	NUM
bracis-19080	369	8	)	)	PUNCT
bracis-19080	369	9	google	google	NOUN
bracis-19080	369	10	scholar	scholar	NOUN
bracis-19080	369	11	  	  	SPACE
bracis-19080	369	12	real	real	ADJ
bracis-19080	369	13	,	,	PUNCT
bracis-19080	369	14	l.	l.	PROPN
bracis-19080	369	15	,	,	PUNCT
bracis-19080	369	16	fonseca	fonseca	PROPN
bracis-19080	369	17	,	,	PUNCT
bracis-19080	369	18	e.	e.	PROPN
bracis-19080	369	19	,	,	PUNCT
bracis-19080	369	20	gonçalo	gonçalo	PROPN
bracis-19080	369	21	oliveira	oliveira	PROPN
bracis-19080	369	22	,	,	PUNCT
bracis-19080	369	23	h.	h.	PROPN
bracis-19080	369	24	:	:	PUNCT
bracis-19080	369	25	the	the	DET
bracis-19080	369	26	assin	assin	PROPN
bracis-19080	369	27	2	2	NUM
bracis-19080	369	28	shared	share	VERB
bracis-19080	369	29	task	task	NOUN
bracis-19080	369	30	:	:	PUNCT
bracis-19080	369	31	a	a	DET
bracis-19080	369	32	quick	quick	ADJ
bracis-19080	369	33	overview	overview	NOUN
bracis-19080	369	34	.	.	PUNCT
bracis-19080	370	1	in	in	ADP
bracis-19080	370	2	:	:	PUNCT
bracis-19080	370	3	quaresma	quaresma	PROPN
bracis-19080	370	4	,	,	PUNCT
bracis-19080	370	5	p.	p.	PROPN
bracis-19080	370	6	,	,	PUNCT
bracis-19080	370	7	vieira	vieira	PROPN
bracis-19080	370	8	,	,	PUNCT
bracis-19080	370	9	r.	r.	PROPN
bracis-19080	370	10	,	,	PUNCT
bracis-19080	370	11	aluísio	aluísio	NOUN
bracis-19080	370	12	,	,	PUNCT
bracis-19080	370	13	s.	s.	PROPN
bracis-19080	370	14	,	,	PUNCT
bracis-19080	370	15	moniz	moniz	PROPN
bracis-19080	370	16	,	,	PUNCT
bracis-19080	370	17	h.	h.	PROPN
bracis-19080	370	18	,	,	PUNCT
bracis-19080	370	19	batista	batista	PROPN
bracis-19080	370	20	,	,	PUNCT
bracis-19080	370	21	f.	f.	PROPN
bracis-19080	370	22	,	,	PUNCT
bracis-19080	370	23	gonçalves	gonçalves	PROPN
bracis-19080	370	24	,	,	PUNCT
bracis-19080	370	25	t.	t.	PROPN
bracis-19080	370	26	(	(	PUNCT
bracis-19080	370	27	eds	eds	PROPN
bracis-19080	370	28	.	.	PUNCT
bracis-19080	370	29	)	)	PUNCT
bracis-19080	370	30	propor	propor	ADJ
bracis-19080	370	31	2020	2020	NUM
bracis-19080	370	32	.	.	PUNCT
bracis-19080	371	1	lncs	lncs	PROPN
bracis-19080	371	2	(	(	PUNCT
bracis-19080	371	3	lnai	lnai	ADJ
bracis-19080	371	4	)	)	PUNCT
bracis-19080	371	5	,	,	PUNCT
bracis-19080	371	6	vol	vol	NOUN
bracis-19080	371	7	.	.	PROPN
bracis-19080	371	8	12037	12037	NUM
bracis-19080	371	9	,	,	PUNCT
bracis-19080	371	10	pp	pp	ADP
bracis-19080	371	11	.	.	PUNCT
bracis-19080	372	1	406–412	406–412	NUM
bracis-19080	372	2	.	.	PUNCT
bracis-19080	372	3	springer	springer	NOUN
bracis-19080	372	4	,	,	PUNCT
bracis-19080	372	5	cham	cham	PROPN
bracis-19080	372	6	(	(	PUNCT
bracis-19080	372	7	2020	2020	NUM
bracis-19080	372	8	)	)	PUNCT
bracis-19080	372	9	.	.	PUNCT
bracis-19080	373	1	https://doi.org/10.1007/978-3-030-41505-1_39	https://doi.org/10.1007/978-3-030-41505-1_39	PROPN
bracis-19080	373	2	chapter	chapter	NOUN
bracis-19080	373	3	  	  	SPACE
bracis-19080	373	4	google	google	PROPN
bracis-19080	373	5	scholar	scholar	NOUN
bracis-19080	373	6	  	  	SPACE
bracis-19080	373	7	real	real	ADJ
bracis-19080	373	8	,	,	PUNCT
bracis-19080	373	9	l.	l.	PROPN
bracis-19080	373	10	,	,	PUNCT
bracis-19080	373	11	et	et	PROPN
bracis-19080	373	12	al	al	PROPN
bracis-19080	373	13	.	.	PROPN
bracis-19080	373	14	:	:	PUNCT
bracis-19080	373	15	sick	sick	ADJ
bracis-19080	373	16	-	-	PUNCT
bracis-19080	373	17	br	br	NOUN
bracis-19080	373	18	:	:	PUNCT
bracis-19080	373	19	a	a	DET
bracis-19080	373	20	portuguese	portuguese	ADJ
bracis-19080	373	21	corpus	corpus	NOUN
bracis-19080	373	22	for	for	ADP
bracis-19080	373	23	inference	inference	NOUN
bracis-19080	373	24	.	.	PUNCT
bracis-19080	374	1	in	in	ADP
bracis-19080	374	2	:	:	PUNCT
bracis-19080	374	3	villavicencio	villavicencio	NOUN
bracis-19080	374	4	,	,	PUNCT
bracis-19080	374	5	a.	a.	NOUN
bracis-19080	374	6	(	(	PUNCT
bracis-19080	374	7	ed	ed	NOUN
bracis-19080	374	8	.	.	PUNCT
bracis-19080	374	9	)	)	PUNCT
bracis-19080	374	10	propor	propor	ADJ
bracis-19080	374	11	2018	2018	NUM
bracis-19080	374	12	.	.	PUNCT
bracis-19080	375	1	lncs	lncs	PROPN
bracis-19080	375	2	(	(	PUNCT
bracis-19080	375	3	lnai	lnai	ADJ
bracis-19080	375	4	)	)	PUNCT
bracis-19080	375	5	,	,	PUNCT
bracis-19080	375	6	vol	vol	NOUN
bracis-19080	375	7	.	.	PROPN
bracis-19080	375	8	11122	11122	NUM
bracis-19080	375	9	,	,	PUNCT
bracis-19080	375	10	pp	pp	ADV
bracis-19080	375	11	.	.	PUNCT
bracis-19080	376	1	303–312	303–312	NUM
bracis-19080	376	2	.	.	PUNCT
bracis-19080	376	3	springer	springer	NOUN
bracis-19080	376	4	,	,	PUNCT
bracis-19080	376	5	cham	cham	PROPN
bracis-19080	376	6	(	(	PUNCT
bracis-19080	376	7	2018	2018	NUM
bracis-19080	376	8	)	)	PUNCT
bracis-19080	376	9	.	.	PUNCT
bracis-19080	377	1	https://doi.org/10.1007/978-3-319-99722-3_31	https://doi.org/10.1007/978-3-319-99722-3_31	PROPN
bracis-19080	377	2	chapter	chapter	NOUN
bracis-19080	377	3	  	  	SPACE
bracis-19080	377	4	google	google	PROPN
bracis-19080	377	5	scholar	scholar	NOUN
bracis-19080	377	6	  	  	SPACE
bracis-19080	377	7	reimers	reimer	NOUN
bracis-19080	377	8	,	,	PUNCT
bracis-19080	377	9	n.	n.	NOUN
bracis-19080	377	10	,	,	PUNCT
bracis-19080	377	11	gurevych	gurevych	NOUN
bracis-19080	377	12	,	,	PUNCT
bracis-19080	377	13	i.	i.	NOUN
bracis-19080	377	14	:	:	PUNCT
bracis-19080	377	15	sentence	sentence	NOUN
bracis-19080	377	16	-	-	PUNCT
bracis-19080	377	17	bert	bert	PROPN
bracis-19080	377	18	:	:	PUNCT
bracis-19080	377	19	sentence	sentence	NOUN
bracis-19080	377	20	embeddings	embedding	NOUN
bracis-19080	377	21	using	use	VERB
bracis-19080	377	22	siamese	siamese	ADJ
bracis-19080	377	23	bert	bert	NOUN
bracis-19080	377	24	-	-	PUNCT
bracis-19080	377	25	networks	network	NOUN
bracis-19080	377	26	.	.	PUNCT
bracis-19080	378	1	in	in	ADP
bracis-19080	378	2	:	:	PUNCT
bracis-19080	378	3	emnlp	emnlp	ADV
bracis-19080	378	4	,	,	PUNCT
bracis-19080	378	5	pp	pp	ADV
bracis-19080	378	6	.	.	PUNCT
bracis-19080	378	7	3973–3983	3973–3983	NUM
bracis-19080	378	8	.	.	PUNCT
bracis-19080	379	1	acl	acl	PROPN
bracis-19080	379	2	(	(	PUNCT
bracis-19080	379	3	2019	2019	NUM
bracis-19080	379	4	)	)	PUNCT
bracis-19080	379	5	google	google	NOUN
bracis-19080	379	6	scholar	scholar	NOUN
bracis-19080	379	7	  	  	SPACE
bracis-19080	379	8	reimers	reimer	NOUN
bracis-19080	379	9	,	,	PUNCT
bracis-19080	379	10	n.	n.	NOUN
bracis-19080	379	11	,	,	PUNCT
bracis-19080	379	12	gurevych	gurevych	NOUN
bracis-19080	379	13	,	,	PUNCT
bracis-19080	379	14	i.	i.	NOUN
bracis-19080	379	15	:	:	PUNCT
bracis-19080	379	16	making	make	VERB
bracis-19080	379	17	monolingual	monolingual	ADJ
bracis-19080	379	18	sentence	sentence	NOUN
bracis-19080	379	19	embeddings	embedding	NOUN
bracis-19080	379	20	multilingual	multilingual	ADJ
bracis-19080	379	21	using	use	VERB
bracis-19080	379	22	knowledge	knowledge	NOUN
bracis-19080	379	23	distillation	distillation	NOUN
bracis-19080	379	24	.	.	PUNCT
bracis-19080	380	1	in	in	ADP
bracis-19080	380	2	:	:	PUNCT
bracis-19080	380	3	emnlp	emnlp	ADV
bracis-19080	380	4	,	,	PUNCT
bracis-19080	380	5	pp	pp	PROPN
bracis-19080	380	6	.	.	PUNCT
bracis-19080	380	7	4512–4525	4512–4525	NUM
bracis-19080	380	8	.	.	PUNCT
bracis-19080	381	1	acl	acl	PROPN
bracis-19080	381	2	(	(	PUNCT
bracis-19080	381	3	2020	2020	NUM
bracis-19080	381	4	)	)	PUNCT
bracis-19080	381	5	google	google	PROPN
bracis-19080	381	6	scholar	scholar	NOUN
bracis-19080	381	7	  	  	SPACE
bracis-19080	381	8	rodrigues	rodrigue	NOUN
bracis-19080	381	9	,	,	PUNCT
bracis-19080	381	10	r.c	r.c	PROPN
bracis-19080	381	11	.	.	PROPN
bracis-19080	381	12	,	,	PUNCT
bracis-19080	381	13	rodrigues	rodrigues	PROPN
bracis-19080	381	14	,	,	PUNCT
bracis-19080	381	15	j.	j.	PROPN
bracis-19080	381	16	,	,	PUNCT
bracis-19080	381	17	de	de	PROPN
bracis-19080	381	18	castro	castro	PROPN
bracis-19080	381	19	,	,	PUNCT
bracis-19080	381	20	p.v.q	p.v.q	PROPN
bracis-19080	381	21	.	.	PROPN
bracis-19080	381	22	,	,	PUNCT
bracis-19080	381	23	da	da	PROPN
bracis-19080	381	24	silva	silva	PROPN
bracis-19080	381	25	,	,	PUNCT
bracis-19080	381	26	n.f.f	n.f.f	PROPN
bracis-19080	381	27	.	.	PROPN
bracis-19080	381	28	,	,	PUNCT
bracis-19080	381	29	soares	soares	PROPN
bracis-19080	381	30	,	,	PUNCT
bracis-19080	381	31	a.	a.	NOUN
bracis-19080	381	32	:	:	PUNCT
bracis-19080	381	33	portuguese	portuguese	ADJ
bracis-19080	381	34	language	language	NOUN
bracis-19080	381	35	models	model	NOUN
bracis-19080	381	36	and	and	CCONJ
bracis-19080	381	37	word	word	NOUN
bracis-19080	381	38	embeddings	embedding	NOUN
bracis-19080	381	39	:	:	PUNCT
bracis-19080	381	40	evaluating	evaluate	VERB
bracis-19080	381	41	on	on	ADP
bracis-19080	381	42	semantic	semantic	ADJ
bracis-19080	381	43	similarity	similarity	NOUN
bracis-19080	381	44	tasks	task	NOUN
bracis-19080	381	45	.	.	PUNCT
bracis-19080	382	1	in	in	ADP
bracis-19080	382	2	:	:	PUNCT
bracis-19080	382	3	quaresma	quaresma	PROPN
bracis-19080	382	4	,	,	PUNCT
bracis-19080	382	5	p.	p.	PROPN
bracis-19080	382	6	,	,	PUNCT
bracis-19080	382	7	vieira	vieira	PROPN
bracis-19080	382	8	,	,	PUNCT
bracis-19080	382	9	r.	r.	PROPN
bracis-19080	382	10	,	,	PUNCT
bracis-19080	382	11	aluísio	aluísio	NOUN
bracis-19080	382	12	,	,	PUNCT
bracis-19080	382	13	s.	s.	PROPN
bracis-19080	382	14	,	,	PUNCT
bracis-19080	382	15	moniz	moniz	PROPN
bracis-19080	382	16	,	,	PUNCT
bracis-19080	382	17	h.	h.	PROPN
bracis-19080	382	18	,	,	PUNCT
bracis-19080	382	19	batista	batista	PROPN
bracis-19080	382	20	,	,	PUNCT
bracis-19080	382	21	f.	f.	PROPN
bracis-19080	382	22	,	,	PUNCT
bracis-19080	382	23	gonçalves	gonçalves	PROPN
bracis-19080	382	24	,	,	PUNCT
bracis-19080	382	25	t.	t.	PROPN
bracis-19080	382	26	(	(	PUNCT
bracis-19080	382	27	eds	eds	PROPN
bracis-19080	382	28	.	.	PUNCT
bracis-19080	382	29	)	)	PUNCT
bracis-19080	382	30	propor	propor	ADJ
bracis-19080	382	31	2020	2020	NUM
bracis-19080	382	32	.	.	PUNCT
bracis-19080	383	1	lncs	lncs	PROPN
bracis-19080	383	2	(	(	PUNCT
bracis-19080	383	3	lnai	lnai	ADJ
bracis-19080	383	4	)	)	PUNCT
bracis-19080	383	5	,	,	PUNCT
bracis-19080	383	6	vol	vol	NOUN
bracis-19080	383	7	.	.	PROPN
bracis-19080	383	8	12037	12037	NUM
bracis-19080	383	9	,	,	PUNCT
bracis-19080	383	10	pp	pp	ADP
bracis-19080	383	11	.	.	PUNCT
bracis-19080	384	1	239–248	239–248	NUM
bracis-19080	384	2	.	.	PUNCT
bracis-19080	384	3	springer	springer	NOUN
bracis-19080	384	4	,	,	PUNCT
bracis-19080	384	5	cham	cham	PROPN
bracis-19080	384	6	(	(	PUNCT
bracis-19080	384	7	2020	2020	NUM
bracis-19080	384	8	)	)	PUNCT
bracis-19080	384	9	.	.	PUNCT
bracis-19080	385	1	https://doi.org/10.1007/978-3-030-41505-1_23	https://doi.org/10.1007/978-3-030-41505-1_23	PROPN
bracis-19080	385	2	chapter	chapter	NOUN
bracis-19080	385	3	  	  	SPACE
bracis-19080	385	4	google	google	PROPN
bracis-19080	385	5	scholar	scholar	NOUN
bracis-19080	385	6	  	  	SPACE
bracis-19080	385	7	souza	souza	PROPN
bracis-19080	385	8	,	,	PUNCT
bracis-19080	385	9	f.	f.	PROPN
bracis-19080	385	10	,	,	PUNCT
bracis-19080	385	11	nogueira	nogueira	PROPN
bracis-19080	385	12	,	,	PUNCT
bracis-19080	385	13	r.	r.	PROPN
bracis-19080	385	14	,	,	PUNCT
bracis-19080	385	15	lotufo	lotufo	PROPN
bracis-19080	385	16	,	,	PUNCT
bracis-19080	385	17	r.	r.	PROPN
bracis-19080	385	18	:	:	PUNCT
bracis-19080	385	19	bertimbau	bertimbau	PROPN
bracis-19080	385	20	:	:	PUNCT
bracis-19080	385	21	pretrained	pretraine	VERB
bracis-19080	385	22	bert	bert	NOUN
bracis-19080	385	23	models	model	NOUN
bracis-19080	385	24	for	for	ADP
bracis-19080	385	25	brazilian	brazilian	ADJ
bracis-19080	385	26	portuguese	portuguese	NOUN
bracis-19080	385	27	.	.	PUNCT
bracis-19080	386	1	in	in	ADP
bracis-19080	386	2	:	:	PUNCT
bracis-19080	386	3	cerri	cerri	PROPN
bracis-19080	386	4	,	,	PUNCT
bracis-19080	386	5	r.	r.	PROPN
bracis-19080	386	6	,	,	PUNCT
bracis-19080	386	7	prati	prati	PROPN
bracis-19080	386	8	,	,	PUNCT
bracis-19080	386	9	r.c	r.c	PROPN
bracis-19080	386	10	.	.	PROPN
bracis-19080	386	11	(	(	PUNCT
bracis-19080	386	12	eds	ed	NOUN
bracis-19080	386	13	.	.	PUNCT
bracis-19080	386	14	)	)	PUNCT
bracis-19080	386	15	bracis	bracis	PROPN
bracis-19080	386	16	2020	2020	NUM
bracis-19080	386	17	.	.	PUNCT
bracis-19080	387	1	lncs	lncs	PROPN
bracis-19080	387	2	(	(	PUNCT
bracis-19080	387	3	lnai	lnai	ADJ
bracis-19080	387	4	)	)	PUNCT
bracis-19080	387	5	,	,	PUNCT
bracis-19080	387	6	vol	vol	NOUN
bracis-19080	387	7	.	.	PUNCT
bracis-19080	387	8	12319	12319	NUM
bracis-19080	387	9	,	,	PUNCT
bracis-19080	387	10	pp	pp	ADP
bracis-19080	387	11	.	.	PUNCT
bracis-19080	388	1	403–417	403–417	NUM
bracis-19080	388	2	.	.	PUNCT
bracis-19080	388	3	springer	springer	NOUN
bracis-19080	388	4	,	,	PUNCT
bracis-19080	388	5	cham	cham	PROPN
bracis-19080	388	6	(	(	PUNCT
bracis-19080	388	7	2020	2020	NUM
bracis-19080	388	8	)	)	PUNCT
bracis-19080	388	9	.	.	PUNCT
bracis-19080	389	1	https://doi.org/10.1007/978-3-030-61377-8_28	https://doi.org/10.1007/978-3-030-61377-8_28	PROPN
bracis-19080	389	2	chapter	chapter	NOUN
bracis-19080	389	3	  	  	SPACE
bracis-19080	389	4	google	google	PROPN
bracis-19080	389	5	scholar	scholar	NOUN
bracis-19080	389	6	  	  	SPACE
bracis-19080	389	7	turney	turney	PROPN
bracis-19080	389	8	,	,	PUNCT
bracis-19080	389	9	p.d	p.d	PROPN
bracis-19080	389	10	.	.	PROPN
bracis-19080	389	11	,	,	PUNCT
bracis-19080	389	12	pantel	pantel	NOUN
bracis-19080	389	13	,	,	PUNCT
bracis-19080	389	14	p.	p.	NOUN
bracis-19080	389	15	:	:	PUNCT
bracis-19080	389	16	from	from	ADP
bracis-19080	389	17	frequency	frequency	NOUN
bracis-19080	389	18	to	to	ADP
bracis-19080	389	19	meaning	mean	VERB
bracis-19080	389	20	:	:	PUNCT
bracis-19080	389	21	vector	vector	NOUN
bracis-19080	389	22	space	space	NOUN
bracis-19080	389	23	models	model	NOUN
bracis-19080	389	24	of	of	ADP
bracis-19080	389	25	semantics	semantic	NOUN
bracis-19080	389	26	.	.	PUNCT
bracis-19080	390	1	j.	j.	PROPN
bracis-19080	390	2	artif	artif	PROPN
bracis-19080	390	3	.	.	PUNCT
bracis-19080	391	1	int	int	NOUN
bracis-19080	391	2	.	.	PUNCT
bracis-19080	392	1	res	re	NOUN
bracis-19080	392	2	.	.	PUNCT
bracis-19080	392	3	37(1	37(1	NUM
bracis-19080	392	4	)	)	PUNCT
bracis-19080	392	5	,	,	PUNCT
bracis-19080	392	6	141–188	141–188	NUM
bracis-19080	392	7	(	(	PUNCT
bracis-19080	392	8	2010	2010	NUM
bracis-19080	392	9	)	)	PUNCT
bracis-19080	392	10	mathscinet	mathscinet	NOUN
bracis-19080	392	11	  	  	SPACE
bracis-19080	392	12	math	math	NOUN
bracis-19080	392	13	  	  	SPACE
bracis-19080	392	14	google	google	PROPN
bracis-19080	392	15	scholar	scholar	NOUN
bracis-19080	392	16	  	  	SPACE
bracis-19080	392	17	wagner	wagner	NOUN
bracis-19080	392	18	filho	filho	PROPN
bracis-19080	392	19	,	,	PUNCT
bracis-19080	392	20	j.a	j.a	PROPN
bracis-19080	392	21	.	.	PROPN
bracis-19080	392	22	,	,	PUNCT
bracis-19080	392	23	wilkens	wilkens	PROPN
bracis-19080	392	24	,	,	PUNCT
bracis-19080	392	25	r.	r.	PROPN
bracis-19080	392	26	,	,	PUNCT
bracis-19080	392	27	idiart	idiart	NOUN
bracis-19080	392	28	,	,	PUNCT
bracis-19080	392	29	m.	m.	NOUN
bracis-19080	392	30	,	,	PUNCT
bracis-19080	392	31	villavicencio	villavicencio	NOUN
bracis-19080	392	32	,	,	PUNCT
bracis-19080	392	33	a.	a.	NOUN
bracis-19080	392	34	:	:	PUNCT
bracis-19080	392	35	the	the	DET
bracis-19080	392	36	brwac	brwac	PROPN
bracis-19080	392	37	corpus	corpus	PROPN
bracis-19080	392	38	:	:	PUNCT
bracis-19080	392	39	a	a	DET
bracis-19080	392	40	new	new	ADJ
bracis-19080	392	41	open	open	ADJ
bracis-19080	392	42	resource	resource	NOUN
bracis-19080	392	43	for	for	ADP
bracis-19080	392	44	brazilian	brazilian	ADJ
bracis-19080	392	45	portuguese	portuguese	NOUN
bracis-19080	392	46	.	.	PUNCT
bracis-19080	393	1	in	in	ADP
bracis-19080	393	2	:	:	PUNCT
bracis-19080	393	3	lrec	lrec	PROPN
bracis-19080	393	4	.	.	PUNCT
bracis-19080	394	1	elra	elra	ADV
bracis-19080	394	2	,	,	PUNCT
bracis-19080	394	3	miyazaki	miyazaki	PROPN
bracis-19080	394	4	,	,	PUNCT
bracis-19080	394	5	japan	japan	PROPN
bracis-19080	394	6	(	(	PUNCT
bracis-19080	394	7	2018	2018	NUM
bracis-19080	394	8	)	)	PUNCT
bracis-19080	394	9	google	google	PROPN
bracis-19080	394	10	scholar	scholar	NOUN
bracis-19080	394	11	  	  	SPACE
bracis-19080	394	12	yang	yang	PROPN
bracis-19080	394	13	,	,	PUNCT
bracis-19080	394	14	y.	y.	PROPN
bracis-19080	394	15	,	,	PUNCT
bracis-19080	394	16	et	et	PROPN
bracis-19080	394	17	al	al	PROPN
bracis-19080	394	18	.	.	PUNCT
bracis-19080	394	19	:	:	PUNCT
bracis-19080	394	20	multilingual	multilingual	ADJ
bracis-19080	394	21	universal	universal	ADJ
bracis-19080	394	22	sentence	sentence	NOUN
bracis-19080	394	23	encoder	encoder	NOUN
bracis-19080	394	24	for	for	ADP
bracis-19080	394	25	semantic	semantic	ADJ
bracis-19080	394	26	retrieval	retrieval	NOUN
bracis-19080	394	27	.	.	PUNCT
bracis-19080	395	1	in	in	ADP
bracis-19080	395	2	:	:	PUNCT
bracis-19080	395	3	acl	acl	NOUN
bracis-19080	395	4	:	:	PUNCT
bracis-19080	395	5	system	system	NOUN
bracis-19080	395	6	demonstrations	demonstration	NOUN
bracis-19080	395	7	,	,	PUNCT
bracis-19080	395	8	pp	pp	ADV
bracis-19080	395	9	.	.	PUNCT
bracis-19080	396	1	87–94	87–94	X
bracis-19080	396	2	.	.	PUNCT
bracis-19080	397	1	acl	acl	PROPN
bracis-19080	397	2	,	,	PUNCT
bracis-19080	397	3	online	online	ADJ
bracis-19080	397	4	,	,	PUNCT
bracis-19080	397	5	july	july	PROPN
bracis-19080	397	6	2020	2020	NUM
bracis-19080	397	7	google	google	PROPN
bracis-19080	397	8	scholar	scholar	NOUN
bracis-19080	397	9	  	  	SPACE
bracis-19080	397	10	yang	yang	PROPN
bracis-19080	397	11	,	,	PUNCT
bracis-19080	397	12	z.	z.	PROPN
bracis-19080	397	13	,	,	PUNCT
bracis-19080	397	14	dai	dai	PROPN
bracis-19080	397	15	,	,	PUNCT
bracis-19080	397	16	z.	z.	PROPN
bracis-19080	397	17	,	,	PUNCT
bracis-19080	397	18	yang	yang	PROPN
bracis-19080	397	19	,	,	PUNCT
bracis-19080	397	20	y.	y.	PROPN
bracis-19080	397	21	,	,	PUNCT
bracis-19080	397	22	carbonell	carbonell	PROPN
bracis-19080	397	23	,	,	PUNCT
bracis-19080	397	24	j.	j.	PROPN
bracis-19080	397	25	,	,	PUNCT
bracis-19080	397	26	salakhutdinov	salakhutdinov	PROPN
bracis-19080	397	27	,	,	PUNCT
bracis-19080	397	28	r.r	r.r	PROPN
bracis-19080	397	29	.	.	PROPN
bracis-19080	397	30	,	,	PUNCT
bracis-19080	397	31	le	le	PROPN
bracis-19080	397	32	,	,	PUNCT
bracis-19080	397	33	q.v	q.v	PROPN
bracis-19080	397	34	.	.	PROPN
bracis-19080	397	35	:	:	PUNCT
bracis-19080	398	1	xlnet	xlnet	PROPN
bracis-19080	398	2	:	:	PUNCT
bracis-19080	398	3	generalized	generalize	VERB
bracis-19080	398	4	autoregressive	autoregressive	ADJ
bracis-19080	398	5	pretraining	pretraine	VERB
bracis-19080	398	6	for	for	ADP
bracis-19080	398	7	language	language	NOUN
bracis-19080	398	8	understanding	understanding	NOUN
bracis-19080	398	9	.	.	PUNCT
bracis-19080	399	1	in	in	ADP
bracis-19080	399	2	:	:	PUNCT
bracis-19080	399	3	wallach	wallach	PROPN
bracis-19080	399	4	,	,	PUNCT
bracis-19080	399	5	h.	h.	PROPN
bracis-19080	399	6	,	,	PUNCT
bracis-19080	399	7	larochelle	larochelle	PROPN
bracis-19080	399	8	,	,	PUNCT
bracis-19080	399	9	h.	h.	PROPN
bracis-19080	399	10	,	,	PUNCT
bracis-19080	399	11	beygelzimer	beygelzimer	NOUN
bracis-19080	399	12	,	,	PUNCT
bracis-19080	399	13	a.	a.	NOUN
bracis-19080	399	14	,	,	PUNCT
bracis-19080	399	15	d	d	NOUN
bracis-19080	399	16	’	'	PUNCT
bracis-19080	399	17	alché	alché	ADJ
bracis-19080	399	18	-	-	PUNCT
bracis-19080	399	19	buc	buc	NOUN
bracis-19080	399	20	,	,	PUNCT
bracis-19080	399	21	f.	f.	PROPN
bracis-19080	399	22	,	,	PUNCT
bracis-19080	399	23	fox	fox	PROPN
bracis-19080	399	24	,	,	PUNCT
bracis-19080	399	25	e.	e.	PROPN
bracis-19080	399	26	,	,	PUNCT
bracis-19080	399	27	garnett	garnett	PROPN
bracis-19080	399	28	,	,	PUNCT
bracis-19080	399	29	r.	r.	PROPN
bracis-19080	399	30	(	(	PUNCT
bracis-19080	399	31	eds	eds	PROPN
bracis-19080	399	32	.	.	PUNCT
bracis-19080	399	33	)	)	PUNCT
bracis-19080	399	34	neurips	neurip	NOUN
bracis-19080	399	35	,	,	PUNCT
bracis-19080	399	36	vol	vol	NOUN
bracis-19080	399	37	.	.	PUNCT
bracis-19080	399	38	32	32	NUM
bracis-19080	399	39	.	.	PUNCT
bracis-19080	400	1	curran	curran	PROPN
bracis-19080	400	2	associates	associates	PROPN
bracis-19080	400	3	,	,	PUNCT
bracis-19080	400	4	inc	inc	PROPN
bracis-19080	400	5	.	.	PROPN
bracis-19080	400	6	(	(	PUNCT
bracis-19080	400	7	2019	2019	NUM
bracis-19080	400	8	)	)	PUNCT
bracis-19080	400	9	google	google	NOUN
bracis-19080	400	10	scholar	scholar	NOUN
bracis-19080	400	11	  	  	SPACE
bracis-19080	400	12	yosinski	yosinski	PROPN
bracis-19080	400	13	,	,	PUNCT
bracis-19080	400	14	j.	j.	PROPN
bracis-19080	400	15	,	,	PUNCT
bracis-19080	400	16	clune	clune	PROPN
bracis-19080	400	17	,	,	PUNCT
bracis-19080	400	18	j.	j.	PROPN
bracis-19080	400	19	,	,	PUNCT
bracis-19080	400	20	bengio	bengio	PROPN
bracis-19080	400	21	,	,	PUNCT
bracis-19080	400	22	y.	y.	PROPN
bracis-19080	400	23	,	,	PUNCT
bracis-19080	400	24	lipson	lipson	PROPN
bracis-19080	400	25	,	,	PUNCT
bracis-19080	400	26	h.	h.	PROPN
bracis-19080	400	27	:	:	PUNCT
bracis-19080	400	28	how	how	SCONJ
bracis-19080	400	29	transferable	transferable	ADJ
bracis-19080	400	30	are	be	AUX
bracis-19080	400	31	features	feature	NOUN
bracis-19080	400	32	in	in	ADP
bracis-19080	400	33	deep	deep	ADJ
bracis-19080	400	34	neural	neural	ADJ
bracis-19080	400	35	networks	network	NOUN
bracis-19080	400	36	?	?	PUNCT
bracis-19080	401	1	in	in	ADP
bracis-19080	401	2	:	:	PUNCT
bracis-19080	401	3	ghahramani	ghahramani	PROPN
bracis-19080	401	4	,	,	PUNCT
bracis-19080	401	5	z.	z.	PROPN
bracis-19080	401	6	,	,	PUNCT
bracis-19080	401	7	welling	well	VERB
bracis-19080	401	8	,	,	PUNCT
bracis-19080	401	9	m.	m.	NOUN
bracis-19080	401	10	,	,	PUNCT
bracis-19080	401	11	cortes	corte	NOUN
bracis-19080	401	12	,	,	PUNCT
bracis-19080	401	13	c.	c.	PROPN
bracis-19080	401	14	,	,	PUNCT
bracis-19080	401	15	lawrence	lawrence	PROPN
bracis-19080	401	16	,	,	PUNCT
bracis-19080	401	17	n.	n.	PROPN
bracis-19080	401	18	,	,	PUNCT
bracis-19080	401	19	weinberger	weinberger	PROPN
bracis-19080	401	20	,	,	PUNCT
bracis-19080	401	21	k.q	k.q	PROPN
bracis-19080	401	22	.	.	PROPN
bracis-19080	401	23	(	(	PUNCT
bracis-19080	401	24	eds	ed	NOUN
bracis-19080	401	25	.	.	PUNCT
bracis-19080	401	26	)	)	PUNCT
bracis-19080	401	27	nips	nip	NOUN
bracis-19080	401	28	,	,	PUNCT
bracis-19080	401	29	vol	vol	NOUN
bracis-19080	401	30	.	.	PUNCT
bracis-19080	401	31	27	27	NUM
bracis-19080	401	32	.	.	PUNCT
bracis-19080	402	1	curran	curran	PROPN
bracis-19080	402	2	associates	associates	PROPN
bracis-19080	402	3	,	,	PUNCT
bracis-19080	402	4	inc	inc	PROPN
bracis-19080	402	5	.	.	PROPN
bracis-19080	402	6	(	(	PUNCT
bracis-19080	402	7	2014	2014	NUM
bracis-19080	402	8	)	)	PUNCT
bracis-19080	402	9	google	google	NOUN
bracis-19080	402	10	scholar	scholar	NOUN
bracis-19080	402	11	  	  	SPACE
bracis-19080	402	12	download	download	NOUN
bracis-19080	402	13	references	reference	NOUN
bracis-19080	402	14	author	author	NOUN
bracis-19080	402	15	information	information	NOUN
bracis-19080	402	16	authors	author	NOUN
bracis-19080	402	17	and	and	CCONJ
bracis-19080	402	18	affiliations	affiliation	NOUN
bracis-19080	402	19	imd	imd	PROPN
bracis-19080	402	20	,	,	PUNCT
bracis-19080	402	21	universidade	universidade	PROPN
bracis-19080	402	22	federal	federal	PROPN
bracis-19080	402	23	do	do	AUX
bracis-19080	402	24	rio	rio	PROPN
bracis-19080	402	25	grande	grande	PROPN
bracis-19080	402	26	do	do	VERB
bracis-19080	402	27	norte	norte	NOUN
bracis-19080	402	28	,	,	PUNCT
bracis-19080	402	29	natal	natal	NOUN
bracis-19080	402	30	,	,	PUNCT
bracis-19080	402	31	rn	rn	PROPN
bracis-19080	402	32	,	,	PUNCT
bracis-19080	402	33	brazil	brazil	PROPN
bracis-19080	402	34	josé	josé	PROPN
bracis-19080	402	35	e.	e.	PROPN
bracis-19080	402	36	andrade	andrade	PROPN
bracis-19080	402	37	junior	junior	PROPN
bracis-19080	402	38	 	 	SPACE
bracis-19080	402	39	&	&	CCONJ
bracis-19080	402	40	 	 	SPACE
bracis-19080	402	41	leonardo	leonardo	PROPN
bracis-19080	402	42	c.	c.	PROPN
bracis-19080	402	43	t.	t.	PROPN
bracis-19080	402	44	bezerra	bezerra	PROPN
bracis-19080	402	45	ifood	ifood	PROPN
bracis-19080	402	46	,	,	PUNCT
bracis-19080	402	47	osasco	osasco	PROPN
bracis-19080	402	48	,	,	PUNCT
bracis-19080	402	49	sp	sp	NOUN
bracis-19080	402	50	,	,	PUNCT
bracis-19080	402	51	brazil	brazil	PROPN
bracis-19080	402	52	josé	josé	PROPN
bracis-19080	402	53	e.	e.	PROPN
bracis-19080	402	54	andrade	andrade	PROPN
bracis-19080	402	55	junior	junior	PROPN
bracis-19080	402	56	data	data	PROPN
bracis-19080	402	57	science	science	NOUN
bracis-19080	402	58	brigade	brigade	NOUN
bracis-19080	402	59	,	,	PUNCT
bracis-19080	402	60	porto	porto	PROPN
bracis-19080	402	61	alegre	alegre	PROPN
bracis-19080	402	62	,	,	PUNCT
bracis-19080	402	63	rs	rs	NOUN
bracis-19080	402	64	,	,	PUNCT
bracis-19080	402	65	brazil	brazil	PROPN
bracis-19080	402	66	jonathan	jonathan	PROPN
bracis-19080	402	67	cardoso	cardoso	PROPN
bracis-19080	402	68	-	-	PUNCT
bracis-19080	402	69	silva	silva	PROPN
bracis-19080	402	70	london	london	PROPN
bracis-19080	402	71	school	school	NOUN
bracis-19080	402	72	of	of	ADP
bracis-19080	402	73	economics	economic	NOUN
bracis-19080	402	74	and	and	CCONJ
bracis-19080	402	75	political	political	ADJ
bracis-19080	402	76	science	science	NOUN
bracis-19080	402	77	,	,	PUNCT
bracis-19080	402	78	london	london	PROPN
bracis-19080	402	79	,	,	PUNCT
bracis-19080	402	80	uk	uk	PROPN
bracis-19080	402	81	jonathan	jonathan	PROPN
bracis-19080	402	82	cardoso	cardoso	PROPN
bracis-19080	402	83	-	-	PUNCT
bracis-19080	402	84	silva	silva	NOUN
bracis-19080	402	85	authors	author	NOUN
bracis-19080	402	86	josé	josé	PROPN
bracis-19080	402	87	e.	e.	PROPN
bracis-19080	402	88	andrade	andrade	PROPN
bracis-19080	402	89	juniorview	juniorview	PROPN
bracis-19080	402	90	author	author	NOUN
bracis-19080	402	91	publications	publication	NOUN
bracis-19080	402	92	search	search	NOUN
bracis-19080	402	93	author	author	NOUN
bracis-19080	402	94	on	on	ADP
bracis-19080	402	95	:	:	PUNCT
bracis-19080	402	96	pubmed	pubmed	PROPN
bracis-19080	402	97	 	 	SPACE
bracis-19080	402	98	google	google	PROPN
bracis-19080	402	99	scholar	scholar	PROPN
bracis-19080	402	100	jonathan	jonathan	PROPN
bracis-19080	402	101	cardoso	cardoso	PROPN
bracis-19080	402	102	-	-	PUNCT
bracis-19080	402	103	silvaview	silvaview	NOUN
bracis-19080	402	104	author	author	NOUN
bracis-19080	402	105	publications	publication	NOUN
bracis-19080	402	106	search	search	NOUN
bracis-19080	402	107	author	author	NOUN
bracis-19080	402	108	on	on	ADP
bracis-19080	402	109	:	:	PUNCT
bracis-19080	402	110	pubmed	pubmed	PROPN
bracis-19080	402	111	 	 	SPACE
bracis-19080	402	112	google	google	PROPN
bracis-19080	402	113	scholar	scholar	PROPN
bracis-19080	402	114	leonardo	leonardo	PROPN
bracis-19080	402	115	c.	c.	PROPN
bracis-19080	402	116	t.	t.	PROPN
bracis-19080	402	117	bezerraview	bezerraview	PROPN
bracis-19080	402	118	author	author	NOUN
bracis-19080	402	119	publications	publication	NOUN
bracis-19080	402	120	search	search	NOUN
bracis-19080	402	121	author	author	NOUN
bracis-19080	402	122	on	on	ADP
bracis-19080	402	123	:	:	PUNCT
bracis-19080	402	124	pubmed	pubmed	PROPN
bracis-19080	402	125	 	 	SPACE
bracis-19080	402	126	google	google	PROPN
bracis-19080	402	127	scholar	scholar	NOUN
bracis-19080	402	128	corresponding	correspond	VERB
bracis-19080	402	129	author	author	NOUN
bracis-19080	402	130	correspondence	correspondence	NOUN
bracis-19080	402	131	to	to	ADP
bracis-19080	402	132	josé	josé	PROPN
bracis-19080	402	133	e.	e.	PROPN
bracis-19080	402	134	andrade	andrade	PROPN
bracis-19080	402	135	junior	junior	PROPN
bracis-19080	402	136	.	.	PUNCT
bracis-19080	403	1	editor	editor	NOUN
bracis-19080	403	2	information	information	NOUN
bracis-19080	403	3	editors	editor	NOUN
bracis-19080	403	4	and	and	CCONJ
bracis-19080	403	5	affiliations	affiliation	NOUN
bracis-19080	403	6	universidade	universidade	PROPN
bracis-19080	403	7	federal	federal	PROPN
bracis-19080	403	8	de	de	X
bracis-19080	403	9	sergipe	sergipe	PROPN
bracis-19080	403	10	,	,	PUNCT
bracis-19080	403	11	são	são	NOUN
bracis-19080	403	12	cristóvão	cristóvão	PROPN
bracis-19080	403	13	,	,	PUNCT
bracis-19080	403	14	brazil	brazil	PROPN
bracis-19080	403	15	andré	andré	PROPN
bracis-19080	403	16	britto	britto	PROPN
bracis-19080	403	17	universidade	universidade	PROPN
bracis-19080	403	18	de	de	PROPN
bracis-19080	403	19	são	são	PROPN
bracis-19080	403	20	paulo	paulo	PROPN
bracis-19080	403	21	,	,	PUNCT
bracis-19080	403	22	são	são	PROPN
bracis-19080	403	23	paulo	paulo	PROPN
bracis-19080	403	24	,	,	PUNCT
bracis-19080	403	25	brazil	brazil	PROPN
bracis-19080	403	26	karina	karina	PROPN
bracis-19080	403	27	valdivia	valdivia	PROPN
bracis-19080	403	28	delgado	delgado	VERB
bracis-19080	403	29	1	1	NUM
bracis-19080	403	30	electronic	electronic	ADJ
bracis-19080	403	31	supplementary	supplementary	ADJ
bracis-19080	403	32	material	material	NOUN
bracis-19080	403	33	below	below	ADV
bracis-19080	403	34	is	be	AUX
bracis-19080	403	35	the	the	DET
bracis-19080	403	36	link	link	NOUN
bracis-19080	403	37	to	to	ADP
bracis-19080	403	38	the	the	DET
bracis-19080	403	39	electronic	electronic	ADJ
bracis-19080	403	40	supplementary	supplementary	ADJ
bracis-19080	403	41	material	material	NOUN
bracis-19080	403	42	.	.	PUNCT
bracis-19080	404	1	supplementary	supplementary	ADJ
bracis-19080	404	2	material	material	NOUN
bracis-19080	404	3	1	1	NUM
bracis-19080	404	4	(	(	PUNCT
bracis-19080	404	5	pdf	pdf	NOUN
bracis-19080	404	6	22	22	NUM
bracis-19080	404	7	kb	kb	PROPN
bracis-19080	404	8	)	)	PUNCT
bracis-19080	404	9	rights	right	NOUN
bracis-19080	404	10	and	and	CCONJ
bracis-19080	404	11	permissions	permission	NOUN
bracis-19080	404	12	reprints	reprint	NOUN
bracis-19080	404	13	and	and	CCONJ
bracis-19080	404	14	permissions	permission	VERB
bracis-19080	404	15	copyright	copyright	NOUN
bracis-19080	404	16	information	information	NOUN
bracis-19080	404	17	©	©	PROPN
bracis-19080	404	18	2021	2021	NUM
bracis-19080	404	19	springer	springer	NOUN
bracis-19080	404	20	nature	nature	NOUN
bracis-19080	404	21	switzerland	switzerland	PROPN
bracis-19080	404	22	ag	ag	PROPN
bracis-19080	404	23	about	about	ADP
bracis-19080	404	24	this	this	DET
bracis-19080	404	25	paper	paper	NOUN
bracis-19080	404	26	cite	cite	VERB
bracis-19080	404	27	this	this	DET
bracis-19080	404	28	paper	paper	NOUN
bracis-19080	404	29	andrade	andrade	PROPN
bracis-19080	404	30	junior	junior	PROPN
bracis-19080	404	31	,	,	PUNCT
bracis-19080	404	32	j.e	j.e	PROPN
bracis-19080	404	33	.	.	PROPN
bracis-19080	404	34	,	,	PUNCT
bracis-19080	404	35	cardoso	cardoso	PROPN
bracis-19080	404	36	-	-	PUNCT
bracis-19080	404	37	silva	silva	NOUN
bracis-19080	404	38	,	,	PUNCT
bracis-19080	404	39	j.	j.	PROPN
bracis-19080	404	40	,	,	PUNCT
bracis-19080	404	41	bezerra	bezerra	PROPN
bracis-19080	404	42	,	,	PUNCT
bracis-19080	404	43	l.c.t	l.c.t	NOUN
bracis-19080	404	44	.	.	PUNCT
bracis-19080	405	1	(	(	PUNCT
bracis-19080	405	2	2021	2021	NUM
bracis-19080	405	3	)	)	PUNCT
bracis-19080	405	4	.	.	PUNCT
bracis-19080	406	1	comparing	compare	VERB
bracis-19080	406	2	contextual	contextual	ADJ
bracis-19080	406	3	embeddings	embedding	NOUN
bracis-19080	406	4	for	for	ADP
bracis-19080	406	5	semantic	semantic	ADJ
bracis-19080	406	6	textual	textual	ADJ
bracis-19080	406	7	similarity	similarity	NOUN
bracis-19080	406	8	in	in	ADP
bracis-19080	406	9	portuguese	portuguese	PROPN
bracis-19080	406	10	.	.	PUNCT
bracis-19080	407	1	in	in	ADP
bracis-19080	407	2	:	:	PUNCT
bracis-19080	407	3	britto	britto	PROPN
bracis-19080	407	4	,	,	PUNCT
bracis-19080	407	5	a.	a.	PROPN
bracis-19080	407	6	,	,	PUNCT
bracis-19080	407	7	valdivia	valdivia	PROPN
bracis-19080	407	8	delgado	delgado	PROPN
bracis-19080	407	9	,	,	PUNCT
bracis-19080	407	10	k.	k.	PROPN
bracis-19080	407	11	(	(	PUNCT
bracis-19080	407	12	eds	eds	PROPN
bracis-19080	407	13	)	)	PUNCT
bracis-19080	407	14	intelligent	intelligent	ADJ
bracis-19080	407	15	systems	system	NOUN
bracis-19080	407	16	.	.	PUNCT
bracis-19080	408	1	bracis	bracis	PROPN
bracis-19080	408	2	2021	2021	NUM
bracis-19080	408	3	.	.	PUNCT
bracis-19080	409	1	lecture	lecture	NOUN
bracis-19080	409	2	notes	note	NOUN
bracis-19080	409	3	in	in	ADP
bracis-19080	409	4	computer	computer	NOUN
bracis-19080	409	5	science	science	NOUN
bracis-19080	409	6	(	(	PUNCT
bracis-19080	409	7	)	)	PUNCT
bracis-19080	409	8	,	,	PUNCT
bracis-19080	409	9	vol	vol	NOUN
bracis-19080	409	10	13074	13074	NUM
bracis-19080	409	11	.	.	PUNCT
bracis-19080	410	1	springer	springer	NOUN
bracis-19080	410	2	,	,	PUNCT
bracis-19080	410	3	cham	cham	PROPN
bracis-19080	410	4	.	.	PUNCT
bracis-19080	411	1	https://doi.org/10.1007/978-3-030-91699-2_27	https://doi.org/10.1007/978-3-030-91699-2_27	PROPN
bracis-19080	411	2	download	download	NOUN
bracis-19080	411	3	citation	citation	NOUN
bracis-19080	411	4	.ris	.ris	PUNCT
bracis-19080	411	5	.enw	.enw	PROPN
bracis-19080	412	1	.bib	.bib	PUNCT
bracis-19080	412	2	doi	doi	PROPN
bracis-19080	412	3	:	:	PUNCT
bracis-19080	412	4	https://doi.org/10.1007/978-3-030-91699-2_27	https://doi.org/10.1007/978-3-030-91699-2_27	PROPN
bracis-19080	412	5	published	publish	VERB
bracis-19080	412	6	:	:	PUNCT
bracis-19080	412	7	28	28	NUM
bracis-19080	412	8	november	november	PROPN
bracis-19080	412	9	2021	2021	NUM
bracis-19080	412	10	publisher	publisher	NOUN
bracis-19080	412	11	name	name	NOUN
bracis-19080	412	12	:	:	PUNCT
bracis-19080	412	13	springer	springer	NOUN
bracis-19080	412	14	,	,	PUNCT
bracis-19080	412	15	cham	cham	PROPN
bracis-19080	412	16	print	print	PROPN
bracis-19080	412	17	isbn	isbn	PROPN
bracis-19080	412	18	:	:	PUNCT
bracis-19080	412	19	978	978	NUM
bracis-19080	412	20	-	-	SYM
bracis-19080	412	21	3	3	NUM
bracis-19080	412	22	-	-	PUNCT
bracis-19080	412	23	030	030	NUM
bracis-19080	412	24	-	-	PUNCT
bracis-19080	412	25	91698	91698	NUM
bracis-19080	412	26	-	-	SYM
bracis-19080	412	27	5	5	NUM
bracis-19080	412	28	online	online	ADJ
bracis-19080	412	29	isbn	isbn	NOUN
bracis-19080	412	30	:	:	PUNCT
bracis-19080	412	31	978	978	NUM
bracis-19080	412	32	-	-	SYM
bracis-19080	412	33	3	3	NUM
bracis-19080	412	34	-	-	PUNCT
bracis-19080	412	35	030	030	NUM
bracis-19080	412	36	-	-	PUNCT
bracis-19080	412	37	91699	91699	NUM
bracis-19080	412	38	-	-	SYM
bracis-19080	412	39	2	2	NUM
bracis-19080	412	40	ebook	ebook	NOUN
bracis-19080	412	41	packages	package	NOUN
bracis-19080	412	42	:	:	PUNCT
bracis-19080	412	43	computer	computer	NOUN
bracis-19080	412	44	sciencecomputer	sciencecomputer	NOUN
bracis-19080	412	45	science	science	NOUN
bracis-19080	412	46	(	(	PUNCT
bracis-19080	412	47	r0	r0	NOUN
bracis-19080	412	48	)	)	PUNCT
bracis-19080	412	49	share	share	VERB
bracis-19080	412	50	this	this	DET
bracis-19080	412	51	paper	paper	NOUN
bracis-19080	412	52	anyone	anyone	PRON
bracis-19080	412	53	you	you	PRON
bracis-19080	412	54	share	share	VERB
bracis-19080	412	55	the	the	DET
bracis-19080	412	56	following	follow	VERB
bracis-19080	412	57	link	link	NOUN
bracis-19080	412	58	with	with	ADP
bracis-19080	412	59	will	will	AUX
bracis-19080	412	60	be	be	AUX
bracis-19080	412	61	able	able	ADJ
bracis-19080	412	62	to	to	PART
bracis-19080	412	63	read	read	VERB
bracis-19080	412	64	this	this	DET
bracis-19080	412	65	content	content	NOUN
bracis-19080	412	66	:	:	PUNCT
bracis-19080	412	67	get	get	VERB
bracis-19080	412	68	shareable	shareable	ADJ
bracis-19080	412	69	linksorry	linksorry	NOUN
bracis-19080	412	70	,	,	PUNCT
bracis-19080	412	71	a	a	DET
bracis-19080	412	72	shareable	shareable	ADJ
bracis-19080	412	73	link	link	NOUN
bracis-19080	412	74	is	be	AUX
bracis-19080	412	75	not	not	PART
bracis-19080	412	76	currently	currently	ADV
bracis-19080	412	77	available	available	ADJ
bracis-19080	412	78	for	for	ADP
bracis-19080	412	79	this	this	DET
bracis-19080	412	80	article	article	NOUN
bracis-19080	412	81	.	.	PUNCT
bracis-19080	413	1	copy	copy	VERB
bracis-19080	413	2	shareable	shareable	ADJ
bracis-19080	413	3	link	link	NOUN
bracis-19080	413	4	to	to	PART
bracis-19080	413	5	clipboard	clipboard	NOUN
bracis-19080	413	6	provided	provide	VERB
bracis-19080	413	7	by	by	ADP
bracis-19080	413	8	the	the	DET
bracis-19080	413	9	springer	springer	NOUN
bracis-19080	413	10	nature	nature	PROPN
bracis-19080	413	11	sharedit	sharedit	PROPN
bracis-19080	413	12	content	content	NOUN
bracis-19080	413	13	-	-	PUNCT
bracis-19080	413	14	sharing	share	VERB
bracis-19080	413	15	initiative	initiative	NOUN
bracis-19080	413	16	keywords	keyword	NOUN
bracis-19080	413	17	deep	deep	ADJ
bracis-19080	413	18	learning	learn	VERB
bracis-19080	413	19	natural	natural	ADJ
bracis-19080	413	20	language	language	NOUN
bracis-19080	413	21	processing	process	VERB
bracis-19080	413	22	semantic	semantic	ADJ
bracis-19080	413	23	textual	textual	ADJ
bracis-19080	413	24	similarity	similarity	NOUN
bracis-19080	413	25	word	word	NOUN
bracis-19080	413	26	embeddings	embedding	NOUN
bracis-19080	413	27	publish	publish	VERB
bracis-19080	413	28	with	with	ADP
bracis-19080	413	29	us	us	PROPN
bracis-19080	413	30	policies	policy	NOUN
bracis-19080	413	31	and	and	CCONJ
bracis-19080	413	32	ethics	ethic	NOUN
bracis-19080	413	33	search	search	NOUN
bracis-19080	413	34	search	search	NOUN
bracis-19080	413	35	by	by	ADP
bracis-19080	413	36	keyword	keyword	NOUN
bracis-19080	413	37	or	or	CCONJ
bracis-19080	413	38	author	author	NOUN
bracis-19080	413	39	search	search	NOUN
bracis-19080	413	40	navigation	navigation	NOUN
bracis-19080	413	41	find	find	VERB
bracis-19080	413	42	a	a	DET
bracis-19080	413	43	journal	journal	NOUN
bracis-19080	413	44	publish	publish	VERB
bracis-19080	413	45	with	with	ADP
bracis-19080	413	46	us	we	PRON
bracis-19080	413	47	track	track	VERB
bracis-19080	413	48	your	your	PRON
bracis-19080	413	49	research	research	NOUN
bracis-19080	413	50	discover	discover	VERB
bracis-19080	413	51	content	content	NOUN
bracis-19080	413	52	journals	journal	NOUN
bracis-19080	413	53	a	a	DET
bracis-19080	413	54	-	-	PUNCT
bracis-19080	413	55	z	z	NOUN
bracis-19080	413	56	books	book	NOUN
bracis-19080	413	57	a	a	DET
bracis-19080	413	58	-	-	PUNCT
bracis-19080	413	59	z	z	NOUN
bracis-19080	413	60	publish	publish	NOUN
bracis-19080	413	61	with	with	ADP
bracis-19080	413	62	us	us	PROPN
bracis-19080	413	63	journal	journal	PROPN
bracis-19080	413	64	finder	finder	PROPN
bracis-19080	413	65	publish	publish	VERB
bracis-19080	413	66	your	your	PRON
bracis-19080	413	67	research	research	NOUN
bracis-19080	413	68	language	language	NOUN
bracis-19080	413	69	editing	edit	VERB
bracis-19080	413	70	open	open	ADJ
bracis-19080	413	71	access	access	NOUN
bracis-19080	413	72	publishing	publishing	NOUN
bracis-19080	413	73	products	product	NOUN
bracis-19080	413	74	and	and	CCONJ
bracis-19080	413	75	services	service	NOUN
bracis-19080	413	76	our	our	PRON
bracis-19080	413	77	products	product	NOUN
bracis-19080	413	78	librarians	librarian	VERB
bracis-19080	413	79	societies	society	NOUN
bracis-19080	413	80	partners	partner	NOUN
bracis-19080	413	81	and	and	CCONJ
bracis-19080	413	82	advertisers	advertiser	NOUN
bracis-19080	413	83	our	our	PRON
bracis-19080	413	84	brands	brand	NOUN
bracis-19080	413	85	springer	springer	NOUN
bracis-19080	413	86	nature	nature	PROPN
bracis-19080	413	87	portfolio	portfolio	PROPN
bracis-19080	413	88	bmc	bmc	PROPN
bracis-19080	413	89	palgrave	palgrave	PROPN
bracis-19080	413	90	macmillan	macmillan	PROPN
bracis-19080	413	91	apress	apress	PROPN
bracis-19080	413	92	discover	discover	VERB
bracis-19080	413	93	your	your	PRON
bracis-19080	413	94	privacy	privacy	NOUN
bracis-19080	413	95	choices	choice	NOUN
bracis-19080	413	96	/	/	SYM
bracis-19080	413	97	manage	manage	NOUN
bracis-19080	413	98	cookies	cookie	NOUN
bracis-19080	413	99	your	your	PRON
bracis-19080	413	100	us	us	PROPN
bracis-19080	414	1	state	state	NOUN
bracis-19080	414	2	privacy	privacy	NOUN
bracis-19080	414	3	rights	right	NOUN
bracis-19080	414	4	accessibility	accessibility	NOUN
bracis-19080	414	5	statement	statement	NOUN
bracis-19080	414	6	terms	term	NOUN
bracis-19080	414	7	and	and	CCONJ
bracis-19080	414	8	conditions	condition	NOUN
bracis-19080	414	9	privacy	privacy	NOUN
bracis-19080	414	10	policy	policy	NOUN
bracis-19080	414	11	help	help	NOUN
bracis-19080	414	12	and	and	CCONJ
bracis-19080	414	13	support	support	VERB
bracis-19080	414	14	legal	legal	ADJ
bracis-19080	414	15	notice	notice	NOUN
bracis-19080	414	16	cancel	cancel	VERB
bracis-19080	414	17	contracts	contract	NOUN
bracis-19080	414	18	here	here	ADV
bracis-19080	414	19	129.74.145.123	129.74.145.123	NUM
bracis-19080	414	20	hesburgh	hesburgh	PROPN
bracis-19080	414	21	library	library	PROPN
bracis-19080	414	22	er	er	INTJ
bracis-19080	414	23	unit	unit	NOUN
bracis-19080	414	24	(	(	PUNCT
bracis-19080	414	25	3005732405	3005732405	NUM
bracis-19080	414	26	)	)	PUNCT
bracis-19080	414	27	northeast	northeast	ADJ
bracis-19080	414	28	research	research	NOUN
bracis-19080	414	29	libraries	library	NOUN
bracis-19080	414	30	(	(	PUNCT
bracis-19080	414	31	nerl	nerl	PROPN
bracis-19080	414	32	)	)	PUNCT
bracis-19080	414	33	(	(	PUNCT
bracis-19080	414	34	8200828607	8200828607	NUM
bracis-19080	414	35	)	)	PUNCT
bracis-19080	414	36	nerl	nerl	VERB
bracis-19080	414	37	ta	ta	X
bracis-19080	414	38	account	account	NOUN
bracis-19080	414	39	(	(	PUNCT
bracis-19080	414	40	3006206169	3006206169	NUM
bracis-19080	414	41	)	)	PUNCT
bracis-19080	414	42	university	university	NOUN
bracis-19080	414	43	of	of	ADP
bracis-19080	414	44	notre	notre	PROPN
bracis-19080	414	45	dame	dame	PROPN
bracis-19080	414	46	hesburgh	hesburgh	PROPN
bracis-19080	414	47	library	library	NOUN
bracis-19080	414	48	(	(	PUNCT
bracis-19080	414	49	3000184373	3000184373	NUM
bracis-19080	414	50	)	)	PUNCT
bracis-19080	415	1	©	©	ADP
bracis-19080	415	2	2025	2025	NUM
bracis-19080	415	3	springer	springer	NOUN
bracis-19080	415	4	nature	nature	NOUN