HPLT/HPLT2.0_cleaned

NB: HPLT2.0 is now superseded by a newer release:

HPLT3.0

We recommed switching to v3.0, unless you have a compelling reason to stay on 2.0.

This is a large-scale collection of web-crawled documents in 191 world languages, produced by the HPLT project. The source of the data is mostly Internet Archive with some additions from Common Crawl.

For a detailed description of the dataset, please refer to our website and our pre-print.

The Cleaned variant of HPLT Datasets v2.0

This is the cleaned variant of the HPLT Datasets v2.0 converted to the Parquet format semi-automatically when being uploaded here. The original JSONL files (which take ~4x fewer disk space than this HF version) and the larger non-cleaned version can be found at https://hplt-project.org/datasets/v2.0.

Dataset Performance

Internal Evaluation

We conducted the FineWeb-style ablation studies within the HPLT project with the focus on one high-resource and one low-resource language: English and Norwegian.

We train 1.7B decoder-only LMs using 100B/30B tokens sampled from the English/Norwegian parts of our HPLT v2 dataset respectively. We replicate the FineWeb corpora comparison design and train the models with a fixed pretraining setup except for the pretraining corpus (English: four corpora; Norwegian: five corpora). Please find the general description of the training and evalutaion setups below and refer to more details in Section 6.2 and Appendix I in our paper.

English Results	Norwegian Results

English

• Corpora: HPLT v1.2, FineWeb and HPLT v2 (ours; deduplicated and cleaned versions).
• Pretraining framework and infrastructure: We trained our English models using Megatron-LM on LUMI with 16 nodes, each with 4 AMD MI250x GPUs with dual-GCD (graphics compute die) design, amounting to 8 logical devices. In total, we used 128 devices and a single 64-core CPU for approximately 84 hours, totalling 11,008 GPU hours per model.
• Evaluation tasks: ARC (Easy and Challenge), Hellaswag, PIQA, and OpenbookQA. We consider only the 0-shot evaluation regime.
• Evaluation framework: LightEval.
• Results: See the plot above. Our models trained on the HPLT v2 datasets reach similar performance to the models trained on FineWeb data and considerably outperform the models trained on HPLT v1.2.

Norwegian

• Corpora: HPLT v1.2, FineWeb-2, mC4, CulturaX, and HPLT v2 (ours).
• Pretraining framework and infrastructure: We trained our Norwegian models using Megatron-DeepSpeed on LUMI with 32 nodes, each with 4 AMD MI250x GPUs. The full pretraining run of each model took approximately 15 hours (wall-clock time), or 1,920 GPU-hours.
• Evaluation tasks: NorCommonsenseQA, NorOpenBookQA, NRK-Quiz-QA, NCB, NorIdiom, and NorQuAD. We discarded tasks that provided a low signal based on the monotonicity and non-random performance criteria defined in the FineWeb-2 evaluation design. The resulting tasks were NCB, NRK-Quiz-QA, NorCommonsenseQA, and NorQuAD. We aggregated the performance using the average normalized score. We consider only the 0-shot evaluation regime.
• Evaluation framework: NorEval, a Norwegian language understanding and generation evaluation benchmark based upon LM Evaluation Harness.
• Results: See the plot above. The Norwegian models trained on FineWeb, CulturaX, and mC4 perform on par with HPLT v2 and outperform those trained on HPLT v1.2. Performance gains start to level off after 16B tokens, with the FineWeb and HPLT v2 scores being more stable during pretraining. This suggests that CulturaX, FineWeb, and HPLT v2 are more effective corpora for Norwegian, and their mixtures potentially provide further benefits.

External Evaluation

The HuggingFace team has compared the utility of various multilingual corpora for training large language models in their FineWeb2 initiative.

They found that the HPLT v2 datasets are next to their FineWeb-2, on par with the CulturaX dataset as shown in this figure produced by HuggingFace:

This is a massive improvement compared to the HPLT v1 datasets, as can be seen on the plot above. In fact, it’s even better: if one looks at the language-specific results, it becomes clear that on Arabic, Hindi, Russian, Thai and Turkish (5 out of 9 languages HuggingFace evaluated on), HPLT v2 is on par or better than FineWeb 2. The average score is lower mostly because of Chinese, we expect it to improve a lot in HPLT v3. Note that the source of the FineWeb 2 (and CulturaX) data is exclusively CommonCrawl, while the HPLT datasets are to a large extent composed of Internet Archive crawls. Thus, FineWeb-2 and HPLT v2 are complementary to each other and should be used together.

Languages

The cleaned version of HPLT Datasets v2.0 consists of subsets corresponding to 191 language codes. Below we provide a list of language codes. For each language code the amount of text is shown as measured in:

• segments: the number of sequences of characters (possibly empty) separated by the newline symbol,
• wcwords: the number of words as defined by the Unix wc utility, i.e. the number of non-whitespaces with a whitespace or the beginning of document before,
• chars: the number of characters,
• docs: the number of documents, each document corresponds to an individual web page from the sourcing web crawls.

	lang	segments	wcwords	chars	docs	Language Name	ISO693-3 code	ISO693-3 code macro	ISO693-1 direct code	ISO693-1 through macro
0	TOTAL	3.00e+11	5.56e+12	3.74e+13	1.06e+10
1	ace_Arab	1.17e+02	8.36e+03	4.97e+04	1.60e+01	Achinese	ace
2	ace_Latn	2.06e+05	8.20e+06	5.08e+07	1.29e+04	Achinese	ace
3	afr_Latn	3.77e+07	1.00e+09	5.95e+09	1.46e+06	Afrikaans	afr		af	af
4	als_Latn	9.51e+07	2.71e+09	1.61e+10	5.38e+06	Tosk Albanian	als	sqi		sq
5	amh_Ethi	7.01e+06	1.96e+08	1.03e+09	2.96e+05	Amharic	amh		am	am
6	ara_Arab	2.20e+09	4.81e+10	2.80e+11	8.27e+07	Arabic	ara		ar	ar
7	asm_Beng	2.68e+06	7.34e+07	4.76e+08	1.76e+05	Assamese	asm		as	as
8	ast_Latn	7.43e+06	1.95e+08	1.24e+09	2.73e+05	Asturian	ast
9	awa_Deva	1.32e+05	6.05e+06	2.88e+07	7.28e+03	Awadhi	awa
10	ayr_Latn	1.88e+05	3.07e+06	2.51e+07	9.22e+03	Central Aymara	ayr	aym		ay
11	azb_Arab	2.39e+06	3.96e+07	2.60e+08	6.61e+04	South Azerbaijani	azb	aze		az
12	azj_Latn	1.27e+08	2.57e+09	1.96e+10	6.48e+06	North Azerbaijani	azj	aze		az
13	bak_Cyrl	3.14e+06	7.53e+07	5.58e+08	1.71e+05	Bashkir	bak		ba	ba
14	bam_Latn	9.17e+04	3.98e+06	2.07e+07	5.72e+03	Bambara	bam		bm	bm
15	ban_Latn	6.01e+05	1.13e+07	7.72e+07	1.07e+04	Balinese	ban
16	bel_Cyrl	4.88e+07	1.21e+09	8.54e+09	2.32e+06	Belarusian	bel		be	be
17	bem_Latn	1.34e+05	4.52e+06	3.23e+07	6.14e+03	Bemba (Zambia)	bem
18	ben_Beng	1.76e+08	4.64e+09	3.02e+10	1.10e+07	Bengali	ben		bn	bn
19	bho_Deva	4.58e+05	1.35e+07	6.86e+07	2.86e+04	Bhojpuri	bho
20	bjn_Arab	1.95e+04	5.48e+05	3.32e+06	1.11e+03	Banjar	bjn	msa		ms
21	bjn_Latn	3.66e+05	8.05e+06	5.60e+07	1.88e+04	Banjar	bjn	msa		ms
22	bod_Tibt	4.65e+05	5.78e+06	2.68e+08	2.74e+04	Tibetan	bod		bo	bo
23	bos_Latn	2.68e+08	7.26e+09	4.61e+10	1.46e+07	Bosnian	bos	hbs	bs	bs
24	bug_Latn	3.86e+04	2.70e+06	1.93e+07	2.02e+03	Buginese	bug
25	bul_Cyrl	6.81e+08	1.53e+10	9.69e+10	2.81e+07	Bulgarian	bul		bg	bg
26	cat_Latn	3.83e+08	1.00e+10	6.02e+10	1.86e+07	Catalan	cat		ca	ca
27	ceb_Latn	2.86e+06	8.59e+07	5.16e+08	1.39e+05	Cebuano	ceb
28	ces_Latn	1.93e+09	4.21e+10	2.74e+11	7.53e+07	Czech	ces		cs	cs
29	cjk_Latn	3.67e+04	9.65e+05	7.43e+06	1.20e+03	Chokwe	cjk
30	ckb_Arab	5.23e+06	1.43e+08	9.13e+08	2.74e+05	Central Kurdish	ckb	kur		ku
31	crh_Latn	1.38e+06	3.68e+07	2.81e+08	1.23e+05	Crimean Tatar	crh
32	cym_Latn	1.56e+07	4.09e+08	2.40e+09	7.58e+05	Welsh	cym		cy	cy
33	dan_Latn	8.73e+08	2.12e+10	1.33e+11	3.38e+07	Danish	dan		da	da
34	deu_Latn	1.11e+10	2.52e+11	1.78e+12	4.82e+08	German	deu		de	de
35	dik_Latn	3.46e+04	2.30e+06	1.15e+07	2.32e+03	Southwestern Dinka	dik	din
36	dyu_Latn	2.46e+04	1.19e+06	5.55e+06	1.39e+03	Dyula	dyu
37	dzo_Tibt	4.00e+04	4.22e+05	7.38e+06	1.63e+03	Dzongkha	dzo		dz	dz
38	ell_Grek	1.85e+09	4.27e+10	2.84e+11	7.03e+07	Modern Greek (1453-)	ell		el	el
39	eng_Latn	1.16e+11	2.86e+12	1.71e+13	4.39e+09	English	eng		en	en
40	epo_Latn	2.04e+07	4.72e+08	2.98e+09	8.19e+05	Esperanto	epo		eo	eo
41	est_Latn	2.64e+08	4.74e+09	3.60e+10	8.45e+06	Estonian	est		et	et
42	eus_Latn	3.76e+07	7.77e+08	6.05e+09	1.97e+06	Basque	eus		eu	eu
43	ewe_Latn	1.43e+05	4.31e+06	2.13e+07	3.77e+03	Ewe	ewe		ee	ee
44	fao_Latn	4.53e+06	9.34e+07	5.82e+08	2.40e+05	Faroese	fao		fo	fo
45	fij_Latn	1.79e+05	7.26e+06	3.77e+07	8.91e+03	Fijian	fij		fj	fj
46	fin_Latn	9.77e+08	1.84e+10	1.56e+11	3.48e+07	Finnish	fin		fi	fi
47	fon_Latn	1.48e+04	1.23e+06	5.34e+06	1.23e+03	Fon	fon
48	fra_Latn	1.06e+10	2.37e+11	1.46e+12	4.02e+08	French	fra		fr	fr
49	fur_Latn	7.30e+05	2.08e+07	1.15e+08	3.67e+04	Friulian	fur
50	fuv_Latn	1.34e+05	5.14e+06	2.99e+07	7.76e+03	Nigerian Fulfulde	fuv	ful		ff
51	gaz_Latn	9.74e+05	2.89e+07	2.19e+08	4.91e+04	West Central Oromo	gaz	orm		om
52	gla_Latn	3.31e+06	8.07e+07	4.84e+08	1.37e+05	Scottish Gaelic	gla		gd	gd
53	gle_Latn	1.10e+07	2.96e+08	1.75e+09	4.91e+05	Irish	gle		ga	ga
54	glg_Latn	6.12e+07	1.64e+09	1.01e+10	3.02e+06	Galician	glg		gl	gl
55	grn_Latn	1.71e+06	3.07e+07	2.19e+08	7.34e+04	Guarani	grn		gn	gn
56	guj_Gujr	2.06e+07	5.77e+08	3.39e+09	1.13e+06	Gujarati	guj		gu	gu
57	hat_Latn	4.64e+06	1.22e+08	6.39e+08	2.13e+05	Haitian	hat		ht	ht
58	hau_Latn	5.69e+06	1.53e+08	8.54e+08	3.16e+05	Hausa	hau		ha	ha
59	heb_Hebr	4.67e+08	9.97e+09	5.68e+10	1.71e+07	Hebrew	heb		he	he
60	hin_Deva	2.67e+08	8.64e+09	4.40e+10	1.36e+07	Hindi	hin		hi	hi
61	hne_Deva	5.50e+04	2.20e+06	1.06e+07	2.81e+03	Chhattisgarhi	hne
62	hrv_Latn	2.97e+08	7.31e+09	4.80e+10	1.23e+07	Croatian	hrv	hbs	hr	hr
63	hun_Latn	1.42e+09	3.05e+10	2.25e+11	5.19e+07	Hungarian	hun		hu	hu
64	hye_Armn	6.52e+07	1.40e+09	1.07e+10	3.60e+06	Armenian	hye		hy	hy
65	ibo_Latn	1.41e+06	3.83e+07	2.05e+08	5.63e+04	Igbo	ibo		ig	ig
66	ilo_Latn	1.12e+06	2.48e+07	1.57e+08	4.88e+04	Iloko	ilo
67	ind_Latn	2.39e+09	5.46e+10	3.84e+11	9.81e+07	Indonesian	ind	msa	id	id
68	isl_Latn	6.96e+07	1.54e+09	9.59e+09	2.84e+06	Icelandic	isl		is	is
69	ita_Latn	5.13e+09	1.27e+11	8.21e+11	2.22e+08	Italian	ita		it	it
70	jav_Latn	6.43e+06	1.38e+08	9.38e+08	1.96e+05	Javanese	jav		jv	jv
71	jpn_Jpan	2.33e+10	4.24e+10	9.01e+11	4.18e+08	Japanese	jpn		ja	ja
72	kab_Latn	3.45e+05	9.22e+06	5.42e+07	1.51e+04	Kabyle	kab
73	kac_Latn	1.59e+05	5.96e+06	2.84e+07	7.59e+03	Kachin	kac
74	kam_Latn	1.43e+04	6.74e+05	4.64e+06	1.18e+03	Kamba (Kenya)	kam
75	kan_Knda	2.49e+07	5.33e+08	4.30e+09	1.34e+06	Kannada	kan		kn	kn
76	kas_Arab	2.71e+04	6.78e+05	3.47e+06	9.49e+02	Kashmiri	kas		ks	ks
77	kas_Deva	1.36e+03	3.19e+04	1.85e+05	1.06e+02	Kashmiri	kas		ks	ks
78	kat_Geor	6.37e+07	1.24e+09	1.02e+10	3.34e+06	Georgian	kat		ka	ka
79	kaz_Cyrl	8.10e+07	1.41e+09	1.11e+10	2.64e+06	Kazakh	kaz		kk	kk
80	kbp_Latn	4.68e+04	4.26e+06	2.09e+07	7.08e+03	Kabiyè	kbp
81	kea_Latn	4.39e+04	1.14e+06	6.14e+06	1.96e+03	Kabuverdianu	kea
82	khk_Cyrl	5.35e+07	1.34e+09	9.33e+09	2.12e+06	Halh Mongolian	khk	mon		mn
83	khm_Khmr	9.86e+06	1.14e+08	2.12e+09	7.01e+05	Khmer	khm		km	km
84	kik_Latn	5.19e+04	1.43e+06	9.29e+06	4.00e+03	Kikuyu	kik		ki	ki
85	kin_Latn	1.92e+06	5.07e+07	3.67e+08	9.27e+04	Kinyarwanda	kin		rw	rw
86	kir_Cyrl	1.00e+07	2.47e+08	1.92e+09	6.76e+05	Kirghiz	kir		ky	ky
87	kmb_Latn	1.18e+04	3.83e+05	2.07e+06	5.31e+02	Kimbundu	kmb
88	kmr_Latn	7.15e+06	1.96e+08	1.12e+09	3.64e+05	Northern Kurdish	kmr	kur		ku
89	knc_Arab	1.08e+04	2.62e+05	1.30e+06	2.45e+02	Central Kanuri	knc	kau		kr
90	knc_Latn	1.05e+04	2.41e+06	1.20e+07	2.47e+03	Central Kanuri	knc	kau		kr
91	kon_Latn	4.75e+04	1.94e+06	1.13e+07	2.54e+03	Kongo	kon		kg	kg
92	kor_Hang	1.36e+09	1.97e+10	8.92e+10	3.89e+07	Korean	kor		ko	ko
93	lao_Laoo	3.20e+05	5.18e+06	8.47e+07	2.95e+04	Lao	lao		lo	lo
94	lij_Latn	1.58e+05	5.59e+06	3.15e+07	8.37e+03	Ligurian	lij
95	lim_Latn	7.14e+06	1.81e+08	1.12e+09	3.68e+05	Limburgan	lim		li	li
96	lin_Latn	2.00e+05	5.56e+06	3.29e+07	7.59e+03	Lingala	lin		ln	ln
97	lit_Latn	3.22e+08	6.68e+09	5.04e+10	1.33e+07	Lithuanian	lit		lt	lt
98	lmo_Latn	2.12e+06	5.96e+07	3.45e+08	1.46e+05	Lombard	lmo
99	ltg_Latn	1.51e+05	3.79e+06	2.69e+07	9.21e+03	Latgalian	ltg	lav		lv
100	ltz_Latn	5.06e+06	1.07e+08	7.10e+08	2.47e+05	Luxembourgish	ltz		lb	lb
101	lua_Latn	3.87e+04	1.37e+06	9.00e+06	1.08e+03	Luba-Lulua	lua
102	lug_Latn	4.08e+05	9.18e+06	6.80e+07	2.13e+04	Ganda	lug		lg	lg
103	luo_Latn	8.41e+04	3.73e+06	2.03e+07	4.15e+03	Luo (Kenya and Tanzania)	luo
104	lus_Latn	3.43e+06	1.25e+08	6.52e+08	1.60e+05	Lushai	lus
105	lvs_Latn	1.74e+08	3.46e+09	2.52e+10	6.77e+06	Standard Latvian	lvs	lav		lv
106	mag_Deva	1.93e+04	8.91e+05	4.28e+06	3.28e+02	Magahi	mag
107	mai_Deva	6.46e+05	1.78e+07	9.67e+07	2.50e+04	Maithili	mai
108	mal_Mlym	4.80e+07	9.74e+08	9.49e+09	3.10e+06	Malayalam	mal		ml	ml
109	mar_Deva	3.63e+07	9.81e+08	6.62e+09	2.08e+06	Marathi	mar		mr	mr
110	min_Latn	6.01e+05	1.10e+07	7.48e+07	2.50e+04	Minangkabau	min	msa		ms
111	mkd_Cyrl	5.70e+07	1.48e+09	9.44e+09	3.57e+06	Macedonian	mkd		mk	mk
112	mlt_Latn	8.68e+06	1.96e+08	1.44e+09	3.67e+05	Maltese	mlt		mt	mt
113	mni_Beng	6.58e+04	1.63e+06	1.18e+07	2.93e+03	Manipuri	mni
114	mos_Latn	1.91e+04	8.08e+05	3.86e+06	9.31e+02	Mossi	mos
115	mri_Latn	2.80e+06	8.68e+07	4.24e+08	1.08e+05	Maori	mri		mi	mi
116	mya_Mymr	3.05e+07	4.53e+08	5.82e+09	1.37e+06	Burmese	mya		my	my
117	nld_Latn	3.08e+09	7.14e+10	4.51e+11	1.39e+08	Dutch	nld		nl	nl
118	nno_Latn	3.46e+07	8.60e+08	5.40e+09	1.42e+06	Norwegian Nynorsk	nno	nor	nn	nn
119	nob_Latn	6.76e+08	2.15e+10	1.33e+11	2.70e+07	Norwegian Bokmål	nob	nor	nb	nb
120	npi_Deva	3.71e+07	1.13e+09	7.26e+09	2.78e+06	Nepali (individual language)	npi	nep		ne
121	nso_Latn	1.43e+05	5.32e+06	2.75e+07	6.07e+03	Pedi	nso
122	nus_Latn	8.51e+03	3.93e+05	1.88e+06	2.72e+02	Nuer	nus
123	nya_Latn	1.34e+06	2.71e+07	2.03e+08	5.31e+04	Nyanja	nya		ny	ny
124	oci_Latn	4.20e+06	1.03e+08	6.35e+08	1.90e+05	Occitan (post 1500)	oci		oc	oc
125	ory_Orya	3.60e+06	1.20e+08	7.82e+08	4.13e+05	Odia	ory	ori		or
126	pag_Latn	8.58e+04	5.66e+06	3.35e+07	6.90e+03	Pangasinan	pag
127	pan_Guru	1.17e+07	3.72e+08	1.90e+09	5.85e+05	Panjabi	pan		pa	pa
128	pap_Latn	1.39e+06	4.67e+07	2.54e+08	8.98e+04	Papiamento	pap
129	pbt_Arab	8.46e+06	2.79e+08	1.30e+09	4.66e+05	Southern Pashto	pbt	pus		ps
130	pes_Arab	3.96e+09	8.86e+10	4.55e+11	9.05e+07	Iranian Persian	pes	fas		fa
131	plt_Latn	4.74e+06	1.17e+08	8.10e+08	2.08e+05	Plateau Malagasy	plt	mlg		mg
132	pol_Latn	4.46e+09	8.95e+10	6.32e+11	1.75e+08	Polish	pol		pl	pl
133	por_Latn	6.12e+09	1.46e+11	8.96e+11	2.38e+08	Portuguese	por		pt	pt
134	prs_Arab	6.90e+07	1.84e+09	9.57e+09	2.84e+06	Dari	prs	fas		fa
135	quy_Latn	4.94e+05	1.73e+07	1.43e+08	3.69e+04	Ayacucho Quechua	quy	que		qu
136	ron_Latn	1.70e+09	4.00e+10	2.51e+11	6.59e+07	Romanian	ron		ro	ro
137	run_Latn	1.75e+06	4.44e+07	3.16e+08	1.37e+05	Rundi	run		rn	rn
138	rus_Cyrl	2.63e+10	5.41e+11	3.91e+12	8.85e+08	Russian	rus		ru	ru
139	sag_Latn	5.19e+04	3.61e+06	1.67e+07	3.16e+03	Sango	sag		sg	sg
140	san_Deva	3.28e+06	4.38e+07	3.59e+08	5.49e+04	Sanskrit	san		sa	sa
141	sat_Olck	4.58e+04	1.08e+06	6.27e+06	2.57e+03	Santali	sat
142	scn_Latn	1.65e+06	4.24e+07	2.52e+08	8.20e+04	Sicilian	scn
143	shn_Mymr	9.21e+04	1.65e+06	2.12e+07	6.00e+03	Shan	shn
144	sin_Sinh	3.37e+07	7.96e+08	4.98e+09	1.15e+06	Sinhala	sin		si	si
145	slk_Latn	4.94e+08	1.06e+10	7.04e+10	2.18e+07	Slovak	slk		sk	sk
146	slv_Latn	2.39e+08	5.44e+09	3.53e+10	1.03e+07	Slovenian	slv		sl	sl
147	smo_Latn	1.01e+06	3.71e+07	1.86e+08	4.59e+04	Samoan	smo		sm	sm
148	sna_Latn	1.20e+06	2.39e+07	1.93e+08	6.11e+04	Shona	sna		sn	sn
149	snd_Arab	2.83e+06	8.95e+07	4.29e+08	1.00e+05	Sindhi	snd		sd	sd
150	som_Latn	1.64e+07	3.89e+08	2.56e+09	9.66e+05	Somali	som		so	so
151	sot_Latn	1.08e+06	3.10e+07	1.72e+08	4.39e+04	Southern Sotho	sot		st	st
152	spa_Latn	1.21e+10	3.22e+11	1.95e+12	5.03e+08	Spanish	spa		es	es
153	srd_Latn	9.17e+05	2.39e+07	1.49e+08	5.38e+04	Sardinian	srd		sc	sc
154	srp_Cyrl	9.38e+07	2.52e+09	1.62e+10	4.12e+06	Serbian	srp	hbs	sr	sr
155	ssw_Latn	6.21e+04	9.94e+05	8.82e+06	2.04e+03	Swati	ssw		ss	ss
156	sun_Latn	3.24e+06	6.96e+07	4.75e+08	1.15e+05	Sundanese	sun		su	su
157	swe_Latn	1.76e+09	4.01e+10	2.51e+11	6.68e+07	Swedish	swe		sv	sv
158	swh_Latn	3.43e+07	7.18e+08	4.66e+09	1.37e+06	Swahili (individual language)	swh	swa		sw
159	szl_Latn	6.37e+05	1.47e+07	1.04e+08	4.09e+04	Silesian	szl
160	tam_Taml	1.69e+08	2.98e+09	2.62e+10	6.11e+06	Tamil	tam		ta	ta
161	taq_Latn	1.39e+04	1.54e+06	8.84e+06	1.75e+03	Tamasheq	taq	tmh
162	tat_Cyrl	1.34e+07	2.97e+08	2.16e+09	6.31e+05	Tatar	tat		tt	tt
163	tel_Telu	3.92e+07	8.35e+08	6.50e+09	2.06e+06	Telugu	tel		te	te
164	tgk_Cyrl	2.48e+07	6.25e+08	4.59e+09	1.26e+06	Tajik	tgk		tg	tg
165	tgl_Latn	5.29e+07	1.35e+09	8.13e+09	1.87e+06	Tagalog	tgl		tl	tl
166	tha_Thai	3.39e+08	3.51e+09	6.00e+10	1.77e+07	Thai	tha		th	th
167	tir_Ethi	1.13e+06	3.67e+07	1.82e+08	6.47e+04	Tigrinya	tir		ti	ti
168	tpi_Latn	2.82e+05	1.25e+07	6.45e+07	1.40e+04	Tok Pisin	tpi
169	tsn_Latn	1.32e+05	5.27e+06	2.77e+07	6.05e+03	Tswana	tsn		tn	tn
170	tso_Latn	2.21e+05	8.67e+06	4.93e+07	1.10e+04	Tsonga	tso		ts	ts
171	tuk_Latn	3.36e+06	7.07e+07	5.70e+08	1.71e+05	Turkmen	tuk		tk	tk
172	tum_Latn	9.90e+04	2.88e+06	2.11e+07	4.38e+03	Tumbuka	tum
173	tur_Latn	2.58e+09	5.17e+10	3.90e+11	1.17e+08	Turkish	tur		tr	tr
174	twi_Latn	1.26e+05	4.70e+06	2.42e+07	5.86e+03	Twi	twi	aka	tw	tw
175	uig_Arab	8.98e+06	2.24e+08	1.75e+09	4.42e+05	Uighur	uig		ug	ug
176	ukr_Cyrl	1.17e+09	2.52e+10	1.83e+11	4.74e+07	Ukrainian	ukr		uk	uk
177	umb_Latn	5.99e+04	2.43e+06	1.54e+07	2.47e+03	Umbundu	umb
178	urd_Arab	5.06e+07	2.13e+09	1.00e+10	3.19e+06	Urdu	urd		ur	ur
179	uzn_Latn	1.48e+07	3.51e+08	2.85e+09	7.07e+05	Northern Uzbek	uzn	uzb		uz
180	vec_Latn	1.58e+06	3.53e+07	2.18e+08	8.48e+04	Venetian	vec
181	vie_Latn	3.02e+09	8.32e+10	3.80e+11	1.01e+08	Vietnamese	vie		vi	vi
182	war_Latn	2.01e+05	5.89e+06	3.56e+07	1.39e+04	Waray (Philippines)	war
183	wol_Latn	1.62e+05	5.46e+06	2.75e+07	5.68e+03	Wolof	wol		wo	wo
184	xho_Latn	1.82e+06	3.03e+07	2.59e+08	6.31e+04	Xhosa	xho		xh	xh
185	ydd_Hebr	2.94e+06	7.75e+07	4.58e+08	1.28e+05	Eastern Yiddish	ydd	yid		yi
186	yor_Latn	1.47e+06	4.28e+07	2.18e+08	6.61e+04	Yoruba	yor		yo	yo
187	yue_Hant	1.24e+06	3.27e+06	7.43e+07	6.13e+04	Yue Chinese	yue	zho		zh
188	zho_Hans	4.24e+10	7.40e+10	2.35e+12	1.25e+09	Chinese	zho		zh	zh
189	zho_Hant	4.48e+09	9.51e+09	2.87e+11	1.57e+08	Chinese	zho		zh	zh
190	zsm_Latn	5.80e+08	1.15e+10	7.84e+10	1.84e+07	Standard Malay	zsm	msa		ms
191	zul_Latn	2.71e+06	4.44e+07	3.81e+08	1.14e+05	Zulu	zul		zu	zu

Terms of Use and Takedown

Terms of Use

These data are released under these Terms of Use:

• We do not own any of the text from which these text data has been extracted.*
• We license the actual packaging of these text data under the Creative Commons CC0 license ("no rights reserved").

Notice and take down policy

Notice: Should you consider that our data contains material that is owned by you and should therefore not be reproduced here, please:

• Clearly identify yourself, with detailed contact data such as an address, telephone number or email address at which you can be contacted.
• Clearly identify the work claimed to be infringed.
• Clearly identify the material that is claimed to be infringing and information reasonably sufficient to allow us to locate the material.
• You can reach us at hplt-datasets@ufal.mff.cuni.cz

Take down: We will comply to legitimate requests by removing the affected sources from the next release of the corpora.

• It is your responsibility that any use of the data complies with any applicable legal framework, such as, among others, the EU Copyright Directive 2019/790 and the General Data Protection Regulation 2018, as amended.

Cite us

@inproceedings{burchell-etal-2025-expanded,
    title = "An Expanded Massive Multilingual Dataset for High-Performance Language Technologies ({HPLT})",
    author = {Burchell, Laurie  and
      de Gibert, Ona  and
      Arefyev, Nikolay  and
      Aulamo, Mikko  and
      Ba{\~n}{\'o}n, Marta  and
      Chen, Pinzhen  and
      Fedorova, Mariia  and
      Guillou, Liane  and
      Haddow, Barry  and
      Haji{\v{c}}, Jan  and
      Helcl, Jind{\v{r}}ich  and
      Henriksson, Erik  and
      Klimaszewski, Mateusz  and
      Komulainen, Ville  and
      Kutuzov, Andrey  and
      Kyt{\"o}niemi, Joona  and
      Laippala, Veronika  and
      M{\ae}hlum, Petter  and
      Malik, Bhavitvya  and
      Mehryary, Farrokh  and
      Mikhailov, Vladislav  and
      Moghe, Nikita  and
      Myntti, Amanda  and
      O{'}Brien, Dayy{\'a}n  and
      Oepen, Stephan  and
      Pal, Proyag  and
      Piha, Jousia  and
      Pyysalo, Sampo  and
      Ram{\'i}rez-S{\'a}nchez, Gema  and
      Samuel, David  and
      Stepachev, Pavel  and
      Tiedemann, J{\"o}rg  and
      Vari{\v{s}}, Du{\v{s}}an  and
      Vojt{\v{e}}chov{\'a}, Tereza  and
      Zaragoza-Bernabeu, Jaume},
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.acl-long.854/",
    doi = "10.18653/v1/2025.acl-long.854",
    pages = "17452--17485",
    ISBN = "979-8-89176-251-0",
    abstract = "Training state-of-the-art large language models requires vast amounts of clean and diverse textual data. However, building suitable multilingual datasets remains a challenge. In this work, we present HPLT v2, a collection of high-quality multilingual monolingual and parallel corpora, extending prior work of the HPLT project. The monolingual portion of the data contains 8T tokens covering 193 languages, while the parallel data contains 380M sentence pairs covering 51 languages. We document the entire data pipeline and release the code to reproduce it. We provide extensive analysis of the quality and characteristics of our data. Finally, we evaluate the performance of language models and machine translation systems trained on HPLT v2, demonstrating its value."
}