Datasets:

HPLT
/

2505-deduplication-evals

corpus stringclasses 4 values	category stringclasses 9 values	dataset stringclasses 23 values	task stringclasses 69 values	prompt stringclasses 51 values	model stringclasses 29 values	ckpt_num int64 500 14.3k	score float64 -0.09 100
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_challenge	arc_ca_challenge_p0	Pregunta: {{question}} Resposta:	1B	500	22.782
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_challenge	arc_ca_challenge_p1	def p1(doc): candidates, choices = doc["choices"]["text"], doc["choices"]["label"] formatted_choices = "\n".join( [f"{choices[i]}: {candidates[i]}" for i, _ in enumerate(candidates)] ) return f"{doc['question']}\n{formatted_choices}\nResposta:"	1B	500	22.952
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_challenge	arc_ca_challenge_p2	def p2(doc): candidates, choices = doc["choices"]["text"], doc["choices"]["label"] formatted_choices = "\n".join( [f"Opción {choices[i]}: {candidates[i]}" for i, _ in enumerate(candidates)] ) return f"{doc['question']}\n{formatted_choices}\nÉs la resposta correcta {', '.join(choices[:-1])} o {choices[-1]}?"	1B	500	27.048
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_easy	arc_ca_easy_p0	Pregunta: {{question}} Resposta:	1B	500	28.788
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_easy	arc_ca_easy_p1	def p1(doc): candidates, choices = doc["choices"]["text"], doc["choices"]["label"] formatted_choices = "\n".join( [f"{choices[i]}: {candidates[i]}" for i, _ in enumerate(candidates)] ) return f"{doc['question']}\n{formatted_choices}\nResposta:"	1B	500	25.042
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_easy	arc_ca_easy_p2	def p2(doc): candidates, choices = doc["choices"]["text"], doc["choices"]["label"] formatted_choices = "\n".join( [f"Opción {choices[i]}: {candidates[i]}" for i, _ in enumerate(candidates)] ) return f"{doc['question']}\n{formatted_choices}\nÉs la resposta correcta {', '.join(choices[:-1])} o {choices[-1]}?"	1B	500	26.263
pre-HPLT 3.0 CD	Language-specific & world knowledge	catalanqa	catalanqa_p0	Context: {{context}} Pregunta: {{question}} Resposta:	1B	500	1.312
pre-HPLT 3.0 CD	Language-specific & world knowledge	catalanqa	catalanqa_p1	{{context}} Pregunta: {{question}} Respecte al passatge anterior, la resposta correcta a la pregunta és	1B	500	1.528
pre-HPLT 3.0 CD	Language-specific & world knowledge	catalanqa	catalanqa_p2	Llegeix el següent passatge i contesta la pregunta. Passatge: {{context}} Pregunta: {{question}} Resposta:	1B	500	1.639
pre-HPLT 3.0 CD	Reading comprehension	catbelebele	catbelebele_p0	Passatge: {{flores_passage}} Pregunta: {{question.strip()}} A. {{mc_answer1}} B. {{mc_answer2}} C. {{mc_answer3}} D. {{mc_answer4}} Resposta:	1B	500	22.889
pre-HPLT 3.0 CD	Reading comprehension	catbelebele	catbelebele_p1	Llegeix el passatge i respon a la pregunta: {{flores_passage}} {{question.strip()}} Opció A: {{mc_answer1}} Opció B: {{mc_answer2}} Opció C: {{mc_answer3}} Opció D: {{mc_answer4}} És la resposta correcta A, B, C o D?	1B	500	25.667
pre-HPLT 3.0 CD	Reading comprehension	catbelebele	catbelebele_p2	{{flores_passage}} En base al passatge anterior, respon a la pregunta: {{question}} A: {{mc_answer1}} B: {{mc_answer2}} C: {{mc_answer3}} D: {{mc_answer4}} Quina és la resposta correcta?	1B	500	22.889
pre-HPLT 3.0 CD	Language knowledge	catcola	catcola_p0	{{Sentence}} Pregunta: Té sentit aquesta frase? Resposta:	1B	500	0
pre-HPLT 3.0 CD	Language knowledge	catcola	catcola_p1	Frase: {{Sentence}} Pregunta: Té sentit aquesta frase? Resposta:	1B	500	0
pre-HPLT 3.0 CD	Language knowledge	catcola	catcola_p2	Determina si la següent frase té sentit: {{Sentence}} Resposta:	1B	500	0
pre-HPLT 3.0 CD	Commonsense reasoning	cocoteros_va	cocoteros_va_p0	Genera una frase curta amb estes paraules: {{keywords}}. El context és: {{context}} Resposta:	1B	500	0.145
pre-HPLT 3.0 CD	Commonsense reasoning	cocoteros_va	cocoteros_va_p1	Escriu una frase curta amb aquestes paraules clau: {{keywords}}. Tingues en compte el següent context: {{context}}. Resposta:	1B	500	0.097
pre-HPLT 3.0 CD	Commonsense reasoning	cocoteros_va	cocoteros_va_p2	Construeix una oració tenint en compte el següent context: {{context}}. Utilitza les paraules clau: {{keywords}}. Resposta:	1B	500	0.101
pre-HPLT 3.0 CD	Commonsense reasoning	copa_ca	copa_ca_p0	{{premise[:-1].strip() + " " + {"cause": "perquè", "effect": "i per tant"}[question]}}	1B	500	55.6
pre-HPLT 3.0 CD	Commonsense reasoning	copa_ca	copa_ca_p1	{{premise[:-1].strip() + " " + {"cause": "atès que", "effect": "així que"}[question]}}	1B	500	54.6
pre-HPLT 3.0 CD	Commonsense reasoning	copa_ca	copa_ca_p2	{{premise[:-1].strip() + {"cause": " a causa que", "effect": ", i com a resultat,"}[question]}}	1B	500	54.8
pre-HPLT 3.0 CD	Reading comprehension	coqcat	coqcat_p0	{{story+"\n\n"}}{% for i in range(questions\|length-1) %}{{"Q: "+questions[i]+"\n\n"+"A: "+answers["input_text"][i]+"\n\n"}}{% endfor %}{{"Q: "+questions[-1]+"\n\n"+"A:"}}	1B	500	4.826
pre-HPLT 3.0 CD	Reading comprehension	coqcat	coqcat_p1	Llegeix el següent passatge i contesta la pregunta\n\nPassatge: {{story+"\n\n"}}{% for i in range(questions\|length-1) %}{{"Pregunta: "+questions[i]+"\n\n"+"Resposta: "+answers["input_text"][i]+"\n\n"}}{% endfor %}{{"Pregunta: "+questions[-1]+"\n\n"+"Resposta:"}}	1B	500	9.243
pre-HPLT 3.0 CD	Reading comprehension	coqcat	coqcat_p2	Llegeix el següent passatge i contesta la pregunta\n\nPassatge: {{story+"\n\n"}}{% for i in range(questions\|length-1) %}{{"Pregunta: "+questions[i]+"\n\n"+"Resposta: "+answers["input_text"][i]+"\n\n"}}{% endfor %}{{"Pregunta: "+questions[-1]+"\n\n"+"Resposta:"}}	1B	500	9.243
pre-HPLT 3.0 CD	Machine translation	flores_en-ca	flores_en-ca_p0	Translate the following sentence into Catalan: {{sentence_eng_Latn}} Catalan:	1B	500	0
pre-HPLT 3.0 CD	Machine translation	flores_en-ca	flores_en-ca_p1	Translate the sentence from English to Catalan. Source sentence (English): {{sentence_eng_Latn}} Target sentence (Catalan):	1B	500	0.001
pre-HPLT 3.0 CD	Machine translation	flores_en-ca	flores_en-ca_p2	The following sentence is written in English. Translate it into Catalan. English: {{sentence_eng_Latn}} Catalan:	1B	500	0
pre-HPLT 3.0 CD	Mathematical reasoning	mgsm_direct_ca	mgsm_direct_ca_p0	Pregunta: {{question}} Resposta:	1B	500	0
pre-HPLT 3.0 CD	Mathematical reasoning	mgsm_direct_ca	mgsm_direct_ca_p1	Respon amb claredat i precisió a la pregunta següent {{question}} Resposta:	1B	500	0
pre-HPLT 3.0 CD	Mathematical reasoning	mgsm_direct_ca	mgsm_direct_ca_p2	Respon a la següent pregunta: {{question}} Raona la teva resposta:	1B	500	0
pre-HPLT 3.0 CD	Language-specific & world knowledge	openbookqa_ca	openbookqa_ca_p0	question_stem	1B	500	25
pre-HPLT 3.0 CD	Language-specific & world knowledge	openbookqa_ca	openbookqa_ca_p1	{{question_stem}} A. {{choices['text'][0]}} B. {{choices['text'][1]}} C. {{choices['text'][2]}} D. {{choices['text'][3]}} Resposta:	1B	500	27.4
pre-HPLT 3.0 CD	Language-specific & world knowledge	openbookqa_ca	openbookqa_ca_p2	{{question_stem}} Opció A: {{choices['text'][0]}} Opció B: {{choices['text'][1]}} C: {{choices['text'][2]}} Opció D: {{choices['text'][3]}} És la resposta correcta A, B, C o D?	1B	500	24.8
pre-HPLT 3.0 CD	Paraphrase detection	parafraseja	parafraseja_p0	null	1B	500	51.15
pre-HPLT 3.0 CD	Paraphrase detection	parafraseja	parafraseja_p1	Determina si les dues oracions següents expressen la mateixa idea o no. Oració 1: {{sentence1}} Oració 2: {{sentence2}} Resposta:	1B	500	49.6
pre-HPLT 3.0 CD	Paraphrase detection	parafraseja	parafraseja_p2	Oració 1: {{sentence1}} Oració 2: {{sentence2}} Pregunta: Les oracions 1 i 2 expressen el mateix significat? Sí o no? Resposta:	1B	500	49.6
pre-HPLT 3.0 CD	Paraphrase detection	paws_ca	paws_ca_p0	null	1B	500	48.9
pre-HPLT 3.0 CD	Paraphrase detection	paws_ca	paws_ca_p1	Determina si les dues oracions següents expressen la mateixa idea o no. Oració 1: {{sentence1}} Oració 2: {{sentence2}} Resposta:	1B	500	54.65
pre-HPLT 3.0 CD	Paraphrase detection	paws_ca	paws_ca_p2	Oració 1: {{sentence1}} Oració 2: {{sentence2}} Pregunta: Les oracions 1 i 2 expressen el mateix significat? Sí o no? Resposta:	1B	500	54.65
pre-HPLT 3.0 CD	Commonsense reasoning	piqa_ca	piqa_ca_p0	Pregunta: {{goal}} Resposta:	1B	500	51.143
pre-HPLT 3.0 CD	Commonsense reasoning	piqa_ca	piqa_ca_p1	{{goal}} A. {{sol1}} B. {{sol2}} Resposta:	1B	500	49.51
pre-HPLT 3.0 CD	Commonsense reasoning	piqa_ca	piqa_ca_p2	{{goal}} A. {{sol1}} B. {{sol2}} Quina és la resposta correcta?	1B	500	49.51
pre-HPLT 3.0 CD	Commonsense reasoning	siqa_ca	siqa_ca_p0	Pregunta: {{context}} {{question}} Resposta:	1B	500	34.033
pre-HPLT 3.0 CD	Commonsense reasoning	siqa_ca	siqa_ca_p1	Passatge: {{context}} Pregunta: {{question}} A. {{answerA}} B. {{answerB}} C. {{answerC}} Resposta:	1B	500	32.907
pre-HPLT 3.0 CD	Commonsense reasoning	siqa_ca	siqa_ca_p2	Llegeix el passatge i respon a la pregunta: {{context}} {{question}} Opció A: {{answerA}} Opció B: {{answerB}} Opció C: {{answerC}} És la resposta correcta A, B o C?	1B	500	30.962
pre-HPLT 3.0 CD	Entailment	teca	teca_p0	null	1B	500	35.616
pre-HPLT 3.0 CD	Entailment	teca	teca_p1	Premissa: {{premise}} Hipòtesi: {{hypothesis}} Indica la relació entre la premissa i la hipòtesi:	1B	500	32.121
pre-HPLT 3.0 CD	Entailment	teca	teca_p2	Premissa: {{premise}} Hipòtesi: {{hypothesis}} Quina és la relació entre la premissa i la hipòtesi? A. En acord B. Neutres entre si C. En contradicció Resposta:	1B	500	33.302
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_gen	veritasqa_ca_gen_p0	Respon a la següent pregunta: {{question}} Resposta:	1B	500	0
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_gen	veritasqa_ca_gen_p1	Respon amb claredat i precisió a la pregunta següent: {{question}} Resposta:	1B	500	0.01
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_gen	veritasqa_ca_gen_p2	Proporciona una resposta detallada per a la pregunta següent: {{question}} Resposta:	1B	500	0
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_mc1	veritasqa_ca_mc1_p0	Pregunta: {{question}} Resposta:	1B	500	27.195
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_mc1	veritasqa_ca_mc1_p1	def mc1_p1(doc): choices = doc["mc1_targets"]["choices"] formatted_choices = "\n".join( [f"Opción {LETTERS[i]}: {choice}" for i, choice in enumerate(choices)] ) letters = LETTERS[: len(choices)] return f"Pregunta: {doc['question']}\n{formatted_choices}\nÉs la resposta correcta {', '.join(letters[:-1])} o {letters[-1]}?\nResposta:"	1B	500	8.499
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_mc1	veritasqa_ca_mc1_p2	def mc1_p2(doc): choices = doc["mc1_targets"]["choices"] formatted_choices = "".join(list(map(lambda choice: f"\n- {choice}", choices))) return f"Pregunta: {doc['question']}\nTria la resposta correcta de la llista:\n{formatted_choices}\nQuina és la resposta correcta?\nResposta:"	1B	500	24.646
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_mc2	veritasqa_ca_mc2_p0	Pregunta: {{question}} Resposta:	1B	500	55.51
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_mc2	veritasqa_ca_mc2_p1	def mc2_p1(doc): choices = doc["mc2_targets"]["choices"] formatted_choices = "\n".join( [f"Opción {LETTERS[i]}: {choice}" for i, choice in enumerate(choices)] ) letters = LETTERS[: len(choices)] return f"Pregunta: {doc['question']}\n{formatted_choices}\nÉs la resposta correcta {', '.join(letters[:-1])} o {letters[-1]}?\nResposta:"	1B	500	64.327
pre-HPLT 3.0 CD	Truthfulness	veritasqa_ca_mc2	veritasqa_ca_mc2_p2	def mc2_p2(doc): choices = doc["mc2_targets"]["choices"] formatted_choices = "".join(list(map(lambda choice: f"\n- {choice}", choices))) return f"Pregunta: {doc['question']}\nTria la resposta correcta de la llista:\n{formatted_choices}\nQuina és la resposta correcta?\nResposta:"	1B	500	52.98
pre-HPLT 3.0 CD	Entailment	wnli_ca	wnli_ca_p0	{{sentence1}} Pregunta: {{sentence2}} Cert o Fals? Resposta:	1B	500	43.662
pre-HPLT 3.0 CD	Entailment	wnli_ca	wnli_ca_p1	Llegeix el text i contesta si l'afirmació és veritable o falsa. Texto: {{sentence1}} Afirmación: {{sentence2}} Resposta:	1B	500	47.887
pre-HPLT 3.0 CD	Entailment	wnli_ca	wnli_ca_p2	{{sentence1}} Pregunta: {{sentence2}}. Resposta:	1B	500	63.38
pre-HPLT 3.0 CD	Entailment	xnli_ca	xnli_ca_p0	null	1B	500	37.871
pre-HPLT 3.0 CD	Entailment	xnli_ca	xnli_ca_p1	Premissa: {{premise}} Hipòtesi: {{hypothesis}} Indica la relació entre la premissa i la hipòtesi:	1B	500	33.855
pre-HPLT 3.0 CD	Entailment	xnli_ca	xnli_ca_p2	Premissa: {{premise}} Hipòtesi: {{hypothesis}} Quina és la relació entre la premissa i la hipòtesi? A. En acord B. Neutres entre si C. En contradicció Resposta:	1B	500	33.333
pre-HPLT 3.0 CD	Reading comprehension	xquad_ca	xquad_ca_p0	Context: {{context}} Pregunta: {{question}} Resposta:	1B	500	0.93
pre-HPLT 3.0 CD	Reading comprehension	xquad_ca	xquad_ca_p1	{{context}} Pregunta: {{question}} Respecte al passatge anterior, la resposta correcta a la pregunta és	1B	500	0.584
pre-HPLT 3.0 CD	Reading comprehension	xquad_ca	xquad_ca_p2	Llegeix el següent passatge i contesta la pregunta. Passatge: {{context}} Pregunta: {{question}} Resposta:	1B	500	1.116
pre-HPLT 3.0 CD	Commonsense reasoning	xstorycloze_ca	xstorycloze_ca_p0	{{[input_sentence_1, input_sentence_2, input_sentence_3, input_sentence_4]\|join(' ')}}	1B	500	52.283
pre-HPLT 3.0 CD	Commonsense reasoning	xstorycloze_ca	xstorycloze_ca_p1	Llegeix la següent història: {{[input_sentence_1, input_sentence_2, input_sentence_3, input_sentence_4]\|join(' ')}} Tria la continuació correcta: - {{sentence_quiz1}} - {{sentence_quiz2}}	1B	500	52.151
pre-HPLT 3.0 CD	Commonsense reasoning	xstorycloze_ca	xstorycloze_ca_p2	{{[input_sentence_1, input_sentence_2, input_sentence_3, input_sentence_4] \| join(' ') }} Què ocorre després? A. {{ sentence_quiz1}} B. {{sentence_quiz2}} Resposta:	1B	500	52.813
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_challenge	arc_ca_challenge_p0	Pregunta: {{question}} Resposta:	2B	1,000	22.355
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_challenge	arc_ca_challenge_p1	def p1(doc): candidates, choices = doc["choices"]["text"], doc["choices"]["label"] formatted_choices = "\n".join( [f"{choices[i]}: {candidates[i]}" for i, _ in enumerate(candidates)] ) return f"{doc['question']}\n{formatted_choices}\nResposta:"	2B	1,000	22.611
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_challenge	arc_ca_challenge_p2	def p2(doc): candidates, choices = doc["choices"]["text"], doc["choices"]["label"] formatted_choices = "\n".join( [f"Opción {choices[i]}: {candidates[i]}" for i, _ in enumerate(candidates)] ) return f"{doc['question']}\n{formatted_choices}\nÉs la resposta correcta {', '.join(choices[:-1])} o {choices[-1]}?"	2B	1,000	26.877
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_easy	arc_ca_easy_p0	Pregunta: {{question}} Resposta:	2B	1,000	31.776
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_easy	arc_ca_easy_p1	def p1(doc): candidates, choices = doc["choices"]["text"], doc["choices"]["label"] formatted_choices = "\n".join( [f"{choices[i]}: {candidates[i]}" for i, _ in enumerate(candidates)] ) return f"{doc['question']}\n{formatted_choices}\nResposta:"	2B	1,000	24.958
pre-HPLT 3.0 CD	Language-specific & world knowledge	arc_ca_easy	arc_ca_easy_p2	def p2(doc): candidates, choices = doc["choices"]["text"], doc["choices"]["label"] formatted_choices = "\n".join( [f"Opción {choices[i]}: {candidates[i]}" for i, _ in enumerate(candidates)] ) return f"{doc['question']}\n{formatted_choices}\nÉs la resposta correcta {', '.join(choices[:-1])} o {choices[-1]}?"	2B	1,000	24.916
pre-HPLT 3.0 CD	Language-specific & world knowledge	catalanqa	catalanqa_p0	Context: {{context}} Pregunta: {{question}} Resposta:	2B	1,000	3.233
pre-HPLT 3.0 CD	Language-specific & world knowledge	catalanqa	catalanqa_p1	{{context}} Pregunta: {{question}} Respecte al passatge anterior, la resposta correcta a la pregunta és	2B	1,000	2.214
pre-HPLT 3.0 CD	Language-specific & world knowledge	catalanqa	catalanqa_p2	Llegeix el següent passatge i contesta la pregunta. Passatge: {{context}} Pregunta: {{question}} Resposta:	2B	1,000	3.878
pre-HPLT 3.0 CD	Reading comprehension	catbelebele	catbelebele_p0	Passatge: {{flores_passage}} Pregunta: {{question.strip()}} A. {{mc_answer1}} B. {{mc_answer2}} C. {{mc_answer3}} D. {{mc_answer4}} Resposta:	2B	1,000	22.778
pre-HPLT 3.0 CD	Reading comprehension	catbelebele	catbelebele_p1	Llegeix el passatge i respon a la pregunta: {{flores_passage}} {{question.strip()}} Opció A: {{mc_answer1}} Opció B: {{mc_answer2}} Opció C: {{mc_answer3}} Opció D: {{mc_answer4}} És la resposta correcta A, B, C o D?	2B	1,000	27.889
pre-HPLT 3.0 CD	Reading comprehension	catbelebele	catbelebele_p2	{{flores_passage}} En base al passatge anterior, respon a la pregunta: {{question}} A: {{mc_answer1}} B: {{mc_answer2}} C: {{mc_answer3}} D: {{mc_answer4}} Quina és la resposta correcta?	2B	1,000	22.889
pre-HPLT 3.0 CD	Language knowledge	catcola	catcola_p0	{{Sentence}} Pregunta: Té sentit aquesta frase? Resposta:	2B	1,000	0
pre-HPLT 3.0 CD	Language knowledge	catcola	catcola_p1	Frase: {{Sentence}} Pregunta: Té sentit aquesta frase? Resposta:	2B	1,000	0
pre-HPLT 3.0 CD	Language knowledge	catcola	catcola_p2	Determina si la següent frase té sentit: {{Sentence}} Resposta:	2B	1,000	0
pre-HPLT 3.0 CD	Commonsense reasoning	cocoteros_va	cocoteros_va_p0	Genera una frase curta amb estes paraules: {{keywords}}. El context és: {{context}} Resposta:	2B	1,000	0.282
pre-HPLT 3.0 CD	Commonsense reasoning	cocoteros_va	cocoteros_va_p1	Escriu una frase curta amb aquestes paraules clau: {{keywords}}. Tingues en compte el següent context: {{context}}. Resposta:	2B	1,000	0.339
pre-HPLT 3.0 CD	Commonsense reasoning	cocoteros_va	cocoteros_va_p2	Construeix una oració tenint en compte el següent context: {{context}}. Utilitza les paraules clau: {{keywords}}. Resposta:	2B	1,000	0.385
pre-HPLT 3.0 CD	Commonsense reasoning	copa_ca	copa_ca_p0	{{premise[:-1].strip() + " " + {"cause": "perquè", "effect": "i per tant"}[question]}}	2B	1,000	56.8
pre-HPLT 3.0 CD	Commonsense reasoning	copa_ca	copa_ca_p1	{{premise[:-1].strip() + " " + {"cause": "atès que", "effect": "així que"}[question]}}	2B	1,000	55.6
pre-HPLT 3.0 CD	Commonsense reasoning	copa_ca	copa_ca_p2	{{premise[:-1].strip() + {"cause": " a causa que", "effect": ", i com a resultat,"}[question]}}	2B	1,000	57.6
pre-HPLT 3.0 CD	Reading comprehension	coqcat	coqcat_p0	{{story+"\n\n"}}{% for i in range(questions\|length-1) %}{{"Q: "+questions[i]+"\n\n"+"A: "+answers["input_text"][i]+"\n\n"}}{% endfor %}{{"Q: "+questions[-1]+"\n\n"+"A:"}}	2B	1,000	13.206
pre-HPLT 3.0 CD	Reading comprehension	coqcat	coqcat_p1	Llegeix el següent passatge i contesta la pregunta\n\nPassatge: {{story+"\n\n"}}{% for i in range(questions\|length-1) %}{{"Pregunta: "+questions[i]+"\n\n"+"Resposta: "+answers["input_text"][i]+"\n\n"}}{% endfor %}{{"Pregunta: "+questions[-1]+"\n\n"+"Resposta:"}}	2B	1,000	11.385
pre-HPLT 3.0 CD	Reading comprehension	coqcat	coqcat_p2	Llegeix el següent passatge i contesta la pregunta\n\nPassatge: {{story+"\n\n"}}{% for i in range(questions\|length-1) %}{{"Pregunta: "+questions[i]+"\n\n"+"Resposta: "+answers["input_text"][i]+"\n\n"}}{% endfor %}{{"Pregunta: "+questions[-1]+"\n\n"+"Resposta:"}}	2B	1,000	11.385
pre-HPLT 3.0 CD	Machine translation	flores_en-ca	flores_en-ca_p0	Translate the following sentence into Catalan: {{sentence_eng_Latn}} Catalan:	2B	1,000	0.006
pre-HPLT 3.0 CD	Machine translation	flores_en-ca	flores_en-ca_p1	Translate the sentence from English to Catalan. Source sentence (English): {{sentence_eng_Latn}} Target sentence (Catalan):	2B	1,000	0.044
pre-HPLT 3.0 CD	Machine translation	flores_en-ca	flores_en-ca_p2	The following sentence is written in English. Translate it into Catalan. English: {{sentence_eng_Latn}} Catalan:	2B	1,000	0.05
pre-HPLT 3.0 CD	Mathematical reasoning	mgsm_direct_ca	mgsm_direct_ca_p0	Pregunta: {{question}} Resposta:	2B	1,000	0
pre-HPLT 3.0 CD	Mathematical reasoning	mgsm_direct_ca	mgsm_direct_ca_p1	Respon amb claredat i precisió a la pregunta següent {{question}} Resposta:	2B	1,000	0
pre-HPLT 3.0 CD	Mathematical reasoning	mgsm_direct_ca	mgsm_direct_ca_p2	Respon a la següent pregunta: {{question}} Raona la teva resposta:	2B	1,000	0
pre-HPLT 3.0 CD	Language-specific & world knowledge	openbookqa_ca	openbookqa_ca_p0	question_stem	2B	1,000	25.2

End of preview. Expand in Data Studio

HPLT 3.0: Deduplication Strategy Comparison Results

Dataset Description

This dataset contains fine-grained results from our HPLT 3.0 pre-release evaluations comparing different data deduplication stategies for the pre-HPLT 3.0 corpora with the previous HPLT 2.0 version. We compare the following data deduplication strategies to guide our design choices, and guard against data quality regression compared to HPLT 2.0: pre-HPLT 3.0 CD (per-crawl deduplication), pre-HPLT 3.0 GD (global deduplication), and pre-HPLT 3.0 GDR (global deduplication & rehydration). We pretrain 2.2B Llama-style decoder models on 30B tokens for each selected language and evaluate them using HPLT-E, a multilingual evaluation framework for comprehensive multi-prompt k-shot evaluation across 124 tasks and 500+ prompts in nine typologically diverse languages: Spanish (spa_Latn), French (fra_Latn), Czech (ces_Latn), Ukrainian (ukr_Cyrl), Finnish (fin_Latn), Catalan (cat_Latn), Galician (glg_Latn), Basque (eus_Latn), and Norwegian (Bokmål and Nynorsk; nor_Latn).

Curated by: High Performance Language Technologies (HPLT)
Languages: Spanish, French, Czech, Ukrainian, Finnish, Catalan, Galician, Basque, Norwegian Bokmål, and Norwegian Nynorsk
Paper: arxiv.org/abs/2511.01066
Repository: github.com/hplt-project/hplt-e
License: Apache 2.0

Please find more details in our paper and GitHub repository.

Uses

This dataset is intended for reproducibility and research purposes. Find an example on how to access the results:

from datasets import load_dataset

dataset = load_dataset("HPLT/2505-deduplication-evals", "spa_Latn", split="results").to_pandas()

Dataset Structure

Dataset Instances

Each dataset instance looks as follows:

{
  'corpus': 'HPLT 2.0',
  'category': 'Commonsense reasoning',
  'dataset': 'xstorycloze_es',
  'task': 'xstorycloze_es_p2',
  'prompt': "{{[input_sentence_1, input_sentence_2, input_sentence_3, input_sentence_4] | join(' ') }}\n¿Qué ocurre después?\nA. {{ sentence_quiz1}} \nB. {{sentence_quiz2}}\nRespuesta:",
  'model': '1B',
  'ckpt_num': 500,
  'score': 52.813}
}

Dataset Fields

corpus: corpus name (pre-HPLT 3.0 CD, pre-HPLT 3.0 GD, pre-HPLT 3.0 GDR, HPLT 2.0)
category: task category
dataset: evaluation dataset name
task: evaluation task (refers to a specific prompt)
prompt: prompt used for evaluation
model: number of pretraining tokens (B)
ckpt_num: number identifier for model
score: standard metric performance score

Cite Us

@article{oepen2025hplt,
  title={HPLT\~{} 3.0: Very Large-Scale Multilingual Resources for LLM and MT. Mono-and Bi-lingual Data, Multilingual Evaluation, and Pre-Trained Models},
  author={Oepen, Stephan and Arefev, Nikolay and Aulamo, Mikko and Ba{\~n}{\'o}n, Marta and Buljan, Maja and Burchell, Laurie and Charpentier, Lucas and Chen, Pinzhen and Fedorova, Mariya and de Gibert, Ona and others},
  journal={arXiv preprint arXiv:2511.01066},
  year={2025}
}

Contact Us

Vladislav Mikhailov [email protected]
Stephan Oepen [email protected]

Downloads last month: 20

Size of downloaded dataset files:

616 kB

Size of the auto-converted Parquet files:

616 kB

Number of rows:

56,257