Dataset Viewer
Auto-converted to Parquet Duplicate
data_source
large_stringclasses
114 values
prompt
listlengths
1
1
ability
large_stringclasses
1 value
reward_model
dict
extra_info
dict
polynomial_equations
[ { "role": "user", "content": "Визначте дійсне значення (значення) t, що задовольняє рівнянню: 61*t = 0.\nРозв'язуючи рівняння, будь ласка, дотримуйтесь цих інструкцій:\n1) Надавайте всі відповіді у вигляді десяткових значень, розділених комами. Наприклад: \"-0.3773, 0.4005\"\n2) Для розв'язків, які можна ви...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Визначте дійсне значення (значення) t, що задовольняє рівнянню: 61*t = 0.\\nРозв'язуючи рівняння, будь ласка, дотримуйтесь цих інструкцій:\\n1) Надавайте всі відповіді у вигляді десяткових значень, розділених комами. Наприклад: \\\"-0.3773, 0.4005\\\"\\n2) Для ро...
{ "index": 0, "split": "train", "source_name": "polynomial_equations:uk", "language": "uk", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
base_conversion
[ { "role": "user", "content": "Переведите число 20233 из системы-4 в эквивалент в систему-3. Для систем с основанием более 10 используйте строчные латинские буквы a-z для обозначения цифр больше 9.\n\nТребования к выводу:\n- Верните ровно один блок <answer>...</answer>.\n- Внутри <answer> напишите только око...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Переведите число 20233 из системы-4 в эквивалент в систему-3. Для систем с основанием более 10 используйте строчные латинские буквы a-z для обозначения цифр больше 9.\", \"answer\": \"202201\", \"metadata\": {\"source_dataset\": \"base_conversion\", \"source_inde...
{ "index": 1, "split": "train", "source_name": "base_conversion:ru", "language": "ru", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
largest_island
[ { "role": "user", "content": "Otrzymujesz następującą binarną macierz o wymiarach 10 x 5:\n0 0 0 0 0\n0 0 0 0 0\n0 0 0 0 0\n1 1 1 1 0\n0 0 1 1 0\n0 0 0 0 0\n0 0 0 0 0\n0 0 0 0 0\n0 0 0 0 0\n0 0 0 0 0\n\nWyspa to grupa jedynek (reprezentujących ląd) połączonych w czterech kierunkach (poziomo lub pionowo).\nM...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Otrzymujesz następującą binarną macierz o wymiarach 10 x 5:\\n0 0 0 0 0\\n0 0 0 0 0\\n0 0 0 0 0\\n1 1 1 1 0\\n0 0 1 1 0\\n0 0 0 0 0\\n0 0 0 0 0\\n0 0 0 0 0\\n0 0 0 0 0\\n0 0 0 0 0\\n\\nWyspa to grupa jedynek (reprezentujących ląd) połączonych w czterech kierunkac...
{ "index": 2, "split": "train", "source_name": "largest_island:pl", "language": "pl", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
futoshiki
[ { "role": "user", "content": "Complete this 5x5 Futoshiki puzzle:\n\n_ _ _ _ 1\n ∨ \n_ 5 _ _ > _\n \n3 1 _ _ _\n \n4 > _ < _ _ _\n∨ \n_ _ 1 3 _\n\nUse the same format as the puzzle above, filling each blank (_) wi...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Complete this 5x5 Futoshiki puzzle:\\n\\n_ _ _ _ 1\\n ∨ \\n_ 5 _ _ > _\\n \\n3 1 _ _ _\\n \\n4 > _ < _ _ _\\n∨ \\n_ _ 1 3 _\\n\\nUse the same format as the puzzle above...
{ "index": 3, "split": "train", "source_name": "futoshiki", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
string_splitting
[ { "role": "user", "content": "一位工程师拥有旧机器 A、B 和 C,他发现可以通过以下规则获得一批新零件 X、Y、Z:\n1) 一台机器 A 可以拆解为两个零件 X 和一个零件 Y。\n2) 两台机器 B 可以拆解为一个零件 X。\n3) 两台机器 C 可以拆解为一个零件 Y。\n4) 一台机器 B 和一台机器 C 可以组合成一台机器 A。\n5) 一个零件 X 和一个零件 Y 可以组合成一个零件 Z。\n\n给定一定数量的初始机器,你的任务是连续循环执行上述规则 1-5,每次仅执行一条规则,直到无法再应用任何规则,或者出现状态(每种机器和零件的数量)重复为止。\n在使用一条规则...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"一位工程师拥有旧机器 A、B 和 C,他发现可以通过以下规则获得一批新零件 X、Y、Z:\\n1) 一台机器 A 可以拆解为两个零件 X 和一个零件 Y。\\n2) 两台机器 B 可以拆解为一个零件 X。\\n3) 两台机器 C 可以拆解为一个零件 Y。\\n4) 一台机器 B 和一台机器 C 可以组合成一台机器 A。\\n5) 一个零件 X 和一个零件 Y 可以组合成一个零件 Z。\\n\\n给定一定数量的初始机器,你的任务是连续循环执行上述规则 1-5,每次仅执行一条规则,直到无法再应用任何规则,或者出现状态(每种机...
{ "index": 4, "split": "train", "source_name": "string_splitting:zh", "language": "zh", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
jugs
[ { "role": "user", "content": "Solve the water jug puzzle so that exactly the target amount of water is in any jug.\nYou have three move types: 'fill', 'empty' and 'pour'.\n\nTo fill Jug A, write 'fill A'.\nTo empty Jug B, write 'empty B'.\nTo pour the contents of Jug A into Jug B, write 'pour A->B'.\nAll ju...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Solve the water jug puzzle so that exactly the target amount of water is in any jug.\\nYou have three move types: 'fill', 'empty' and 'pour'.\\n\\nTo fill Jug A, write 'fill A'.\\nTo empty Jug B, write 'empty B'.\\nTo pour the contents of Jug A into Jug B, write ...
{ "index": 5, "split": "train", "source_name": "jugs", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
modulo_grid
[ { "role": "user", "content": "Identifique o padrão matemático que define esta grade e, em seguida, use esse padrão para preencher os pontos de interrogação. Retorne a grade completa como sua resposta, com dimensões e quebras de linha idênticas às mostradas aqui.\n\n❌❌❌❌❌❌❌❌❌❌❌❌✅❌❌❌❌❌❌❌\n❌❌❌❌❌❌❌❌❌❌❌✅❌❌❌❌❌❌❌❌...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Identifique o padrão matemático que define esta grade e, em seguida, use esse padrão para preencher os pontos de interrogação. Retorne a grade completa como sua resposta, com dimensões e quebras de linha idênticas às mostradas aqui.\\n\\n❌❌❌❌❌❌❌❌❌❌❌❌✅❌❌❌❌❌❌❌\\n❌❌...
{ "index": 6, "split": "train", "source_name": "modulo_grid:pt", "language": "pt", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
binary_alternation
[ { "role": "user", "content": "Dla poniższego ciągu binarnego zwróć minimalną liczbę zamian znaków potrzebną do tego, aby stał się on naprzemienny, lub -1, jeśli nie jest to możliwe.\n\nCiąg jest naprzemienny, gdy żadne dwa sąsiednie znaki nie są takie same. Na przykład „010” i „1010” są naprzemienne, podcza...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Dla poniższego ciągu binarnego zwróć minimalną liczbę zamian znaków potrzebną do tego, aby stał się on naprzemienny, lub -1, jeśli nie jest to możliwe.\\n\\nCiąg jest naprzemienny, gdy żadne dwa sąsiednie znaki nie są takie same. Na przykład „010” i „1010” są nap...
{ "index": 7, "split": "train", "source_name": "binary_alternation:pl", "language": "pl", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
maximum_flow
[ { "role": "user", "content": "용량이 주어진 유향 그래프가 주어집니다.\n\n노드: 0, 1, 2, 3, 4, 5, 6\n유향 간선 (u, v, 용량):\n(0, 1, 19)\n(0, 2, 12)\n(0, 3, 12)\n(0, 4, 7)\n(0, 5, 13)\n(0, 6, 7)\n(1, 0, 14)\n(1, 2, 10)\n(1, 3, 20)\n(1, 4, 17)\n(1, 5, 16)\n(1, 6, 12)\n(2, 0, 16)\n(2, 3, 13)\n(2, 4, 20)\n(2, 5, 14)\n(3, 1, 18)\n(3, 2,...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"용량이 주어진 유향 그래프가 주어집니다.\\n\\n노드: 0, 1, 2, 3, 4, 5, 6\\n유향 간선 (u, v, 용량):\\n(0, 1, 19)\\n(0, 2, 12)\\n(0, 3, 12)\\n(0, 4, 7)\\n(0, 5, 13)\\n(0, 6, 7)\\n(1, 0, 14)\\n(1, 2, 10)\\n(1, 3, 20)\\n(1, 4, 17)\\n(1, 5, 16)\\n(1, 6, 12)\\n(2, 0, 16)\\n(2, 3, 13)\\n(2, 4, 20...
{ "index": 8, "split": "train", "source_name": "maximum_flow:ko", "language": "ko", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
diffie_hellman
[ { "role": "user", "content": "Dans un échange de clés Diffie-Hellman, les paramètres publics sont le nombre premier p = 41 et le générateur g = 7. La valeur publique d'Alice est A = 14 et la valeur publique de Bob est B = 38. Les exposants privés cachés satisfont A = g^a mod p et B = g^b mod p, et le secret...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Dans un échange de clés Diffie-Hellman, les paramètres publics sont le nombre premier p = 41 et le générateur g = 7. La valeur publique d'Alice est A = 14 et la valeur publique de Bob est B = 38. Les exposants privés cachés satisfont A = g^a mod p et B = g^b mod ...
{ "index": 9, "split": "train", "source_name": "diffie_hellman:fr", "language": "fr", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
vm_trace
[ { "role": "user", "content": "Набір інструкцій:\nMOV X, Y : присвоїти регістру X значення Y (Y може бути цілим числом або регістром).\nADD X, Y : виконати X = X + Y.\nSUB X, Y : виконати X = X - Y.\nCMP X, Y : порівняти X і Y та зберегти результат порівняння.\nJNZ LBL : перейти до мітки LBL, якщо результат ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Набір інструкцій:\\nMOV X, Y : присвоїти регістру X значення Y (Y може бути цілим числом або регістром).\\nADD X, Y : виконати X = X + Y.\\nSUB X, Y : виконати X = X - Y.\\nCMP X, Y : порівняти X і Y та зберегти результат порівняння.\\nJNZ LBL : перейти до мітки ...
{ "index": 10, "split": "train", "source_name": "vm_trace:uk", "language": "uk", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
palindrome_generation
[ { "role": "user", "content": "당신의 과제는 주어진 철자 목록을 사용하여 유효한 회문(palindrome)을 만드는 것입니다.\n\n회문이란 앞에서 읽으나 뒤에서 읽으나 동일한 문구를 말합니다.\n\n가능한 답이 여러 개라면, 그중 하나만 응답하십시오. 제공된 모든 철자를 반드시 사용해야 합니다.\n\n출력은 공백이나 문장 부호가 없는 하나의 문자열이어야 합니다.\n\n이제 다음 철자들을 사용하여 유효한 회문을 만드십시오: ㅡ, ㄲ, ㅡ, ㅗ, ㅅ, ㄲ, ㅗ\n\n출력 요구사항:\n- 정확히 하나의 <answer>...</...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"당신의 과제는 주어진 철자 목록을 사용하여 유효한 회문(palindrome)을 만드는 것입니다.\\n\\n회문이란 앞에서 읽으나 뒤에서 읽으나 동일한 문구를 말합니다.\\n\\n가능한 답이 여러 개라면, 그중 하나만 응답하십시오. 제공된 모든 철자를 반드시 사용해야 합니다.\\n\\n출력은 공백이나 문장 부호가 없는 하나의 문자열이어야 합니다.\\n\\n이제 다음 철자들을 사용하여 유효한 회문을 만드십시오: ㅡ, ㄲ, ㅡ, ㅗ, ㅅ, ㄲ, ㅗ\", \"answer\"...
{ "index": 11, "split": "train", "source_name": "palindrome_generation:ko", "language": "ko", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
survo
[ { "role": "user", "content": "いくつかの位置に数値が入力されており、他の位置が0で示されている5x5の行列があります。行列は以下の通りです:\n3 0 0 7 19\n7 0 2 8 20\n5 5 5 2 17\n7 8 0 6 24\n22 23 12 23 80\n各行および各列の最後の数値は、その行または列に含まれる他のすべての数値の合計を表しています。これらの条件を満たすように、[2, 3, 3, 7]の数値を使用して0の位置を埋めてください。各数値は一度しか使用できません。\n\n出力要件:\n- <answer>...</answer> ブロックを必ず 1 つだけ返...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"いくつかの位置に数値が入力されており、他の位置が0で示されている5x5の行列があります。行列は以下の通りです:\\n3 0 0 7 19\\n7 0 2 8 20\\n5 5 5 2 17\\n7 8 0 6 24\\n22 23 12 23 80\\n各行および各列の最後の数値は、その行または列に含まれる他のすべての数値の合計を表しています。これらの条件を満たすように、[2, 3, 3, 7]の数値を使用して0の位置を埋めてください。各数値は一度しか使用できません。\", \"answer\": \"3 7 2 7 1...
{ "index": 12, "split": "train", "source_name": "survo:ja", "language": "ja", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
boxnet
[ { "role": "user", "content": "You are a central planner directing agents on a grid so that colored boxes reach their matching targets.\nEach Agent occupies one 1x1 square and can only interact with objects inside that square. Agents can move a box to an adjacent square or directly to a matching target. A sq...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"You are a central planner directing agents on a grid so that colored boxes reach their matching targets.\\nEach Agent occupies one 1x1 square and can only interact with objects inside that square. Agents can move a box to an adjacent square or directly to a match...
{ "index": 13, "split": "train", "source_name": "boxnet", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
uniform_acceleration
[ { "role": "user", "content": "Cząstka rozpoczyna ruch z prędkością początkową 31 m/s przy stałym przyspieszeniu -7 m/s^2 przez czas 1 sekund. Najpierw oblicz prędkość końcową, korzystając ze wzoru v = u + at. Następnie oblicz dokładne przemieszczenie, korzystając ze wzoru s = ut + (1/2)at^2. Podaj ostateczn...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Cząstka rozpoczyna ruch z prędkością początkową 31 m/s przy stałym przyspieszeniu -7 m/s^2 przez czas 1 sekund. Najpierw oblicz prędkość końcową, korzystając ze wzoru v = u + at. Następnie oblicz dokładne przemieszczenie, korzystając ze wzoru s = ut + (1/2)at^2. ...
{ "index": 14, "split": "train", "source_name": "uniform_acceleration:pl", "language": "pl", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
basic_arithmetic
[ { "role": "user", "content": "¿Qué es -20 * -106?\n\nRequisitos de salida:\n- Devuelve exactamente un bloque <answer>...</answer>.\n- Dentro de <answer>, escribe solo la respuesta final.\n- No incluyas explicación, unidades ni palabras adicionales dentro de <answer>.\n- Si la tarea pide un formato exacto, c...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"¿Qué es -20 * -106?\", \"answer\": \"2120\", \"metadata\": {\"source_dataset\": \"basic_arithmetic\", \"source_index\": 15, \"expression\": \"-20 * -106\", \"num_terms\": 2, \"num_digits\": 3, \"language\": \"es\", \"difficulty\": {\"num_terms\": [2, 6], \"num_di...
{ "index": 15, "split": "train", "source_name": "basic_arithmetic:es", "language": "es", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
number_filtering
[ { "role": "user", "content": "Keep all numbers smaller than 38 in this list: ['93.676', '-35.295', '-41', '24.821']\nReturn the new list in the same format.\n\nOutput requirements:\n- Return exactly one <answer>...</answer> block.\n- Inside <answer>, write only the final answer.\n- Do not include explanatio...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Keep all numbers smaller than 38 in this list: ['93.676', '-35.295', '-41', '24.821']\\nReturn the new list in the same format.\", \"answer\": \"['-35.295', '-41', '24.821']\", \"metadata\": {\"source_dataset\": \"number_filtering\", \"source_index\": 16, \"langu...
{ "index": 16, "split": "train", "source_name": "number_filtering", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
decimal_chain_sum
[ { "role": "user", "content": "इस दशमलव अभिव्यक्ति का मूल्यांकन कीजिए: 21.49 + 95.4 - 22.7 =\n\nआउटपुट आवश्यकताएँ:\n- ठीक एक <answer>...</answer> ब्लॉक लौटाएँ।\n- <answer> के अंदर केवल अंतिम उत्तर लिखें।\n- <answer> के अंदर कोई व्याख्या, इकाई या अतिरिक्त शब्द शामिल न करें।\n- यदि कार्य सटीक स्वरूपण मांगता है...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"इस दशमलव अभिव्यक्ति का मूल्यांकन कीजिए: 21.49 + 95.4 - 22.7 =\", \"answer\": \"94.19\", \"metadata\": {\"source_dataset\": \"decimal_chain_sum\", \"source_index\": 17, \"num_terms\": 3, \"num_digits\": 2, \"expression\": \"21.49 + 95.4 - 22.7\", \"language\": \"h...
{ "index": 17, "split": "train", "source_name": "decimal_chain_sum:hi", "language": "hi", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
graph_chromatic_number
[ { "role": "user", "content": "Otrzymujesz graf nieskierowany.\n\nWęzły: 0, 1, 2, 3, 4, 5\nKrawędzie: (0, 1), (0, 3), (1, 3), (1, 4), (1, 5), (2, 3), (2, 4), (2, 5), (3, 4), (3, 5), (4, 5)\n\nLiczba chromatyczna to minimalna liczba kolorów potrzebna do tego, aby sąsiadujące węzły miały różne kolory.\nJaka je...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Otrzymujesz graf nieskierowany.\\n\\nWęzły: 0, 1, 2, 3, 4, 5\\nKrawędzie: (0, 1), (0, 3), (1, 3), (1, 4), (1, 5), (2, 3), (2, 4), (2, 5), (3, 4), (3, 5), (4, 5)\\n\\nLiczba chromatyczna to minimalna liczba kolorów potrzebna do tego, aby sąsiadujące węzły miały ró...
{ "index": 18, "split": "train", "source_name": "graph_chromatic_number:pl", "language": "pl", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
maximum_flow
[ { "role": "user", "content": "On vous donne un graphe orienté avec des capacités. \n\nNœuds : 0, 1, 2, 3, 4, 5, 6, 7\nArêtes orientées (u, v, capacité) :\n(0, 1, 13)\n(0, 4, 5)\n(0, 5, 2)\n(0, 6, 15)\n(0, 7, 5)\n(1, 2, 14)\n(1, 4, 6)\n(1, 6, 14)\n(1, 7, 13)\n(2, 0, 13)\n(2, 1, 10)\n(2, 3, 12)\n(2, 5, 6)\n(2...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"On vous donne un graphe orienté avec des capacités. \\n\\nNœuds : 0, 1, 2, 3, 4, 5, 6, 7\\nArêtes orientées (u, v, capacité) :\\n(0, 1, 13)\\n(0, 4, 5)\\n(0, 5, 2)\\n(0, 6, 15)\\n(0, 7, 5)\\n(1, 2, 14)\\n(1, 4, 6)\\n(1, 6, 14)\\n(1, 7, 13)\\n(2, 0, 13)\\n(2, 1, 1...
{ "index": 19, "split": "train", "source_name": "maximum_flow:fr", "language": "fr", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
systems_of_linear_equations
[ { "role": "user", "content": "Розв'яжіть систему відносно (x, y, z):\n-3x + 5y - 5z = 17\n-9x + y - 2z = 83\n9x + 5y + 6z = -135\n\nНадайте вашу відповідь у вигляді кортежу (x, y, z), розділеного комами.\n\nВимоги до виводу:\n- Поверніть рівно один блок <answer>...</answer>.\n- Усередині <answer> напишіть л...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Розв'яжіть систему відносно (x, y, z):\\n-3x + 5y - 5z = 17\\n-9x + y - 2z = 83\\n9x + 5y + 6z = -135\\n\\nНадайте вашу відповідь у вигляді кортежу (x, y, z), розділеного комами.\", \"answer\": \"-9, -6, -4\", \"metadata\": {\"source_dataset\": \"systems_of_linea...
{ "index": 20, "split": "train", "source_name": "systems_of_linear_equations:uk", "language": "uk", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
list_functions
[ { "role": "user", "content": "Vous êtes un expert en raisonnement inductif. Générez une sortie correspondant à l'entrée donnée.\nLa sortie est générée en appliquant la même règle qui lie l'entrée à la sortie pour les exemples fournis.\nVotre réponse doit être une liste de style Python utilisant des crochets...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Vous êtes un expert en raisonnement inductif. Générez une sortie correspondant à l'entrée donnée.\\nLa sortie est générée en appliquant la même règle qui lie l'entrée à la sortie pour les exemples fournis.\\nVotre réponse doit être une liste de style Python utili...
{ "index": 21, "split": "train", "source_name": "list_functions:fr", "language": "fr", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
matrix_determinants
[ { "role": "user", "content": "Calcola il determinante della seguente matrice A:\n[2, -3, 2, -4]\n[-3, 4, -1, 5]\n[-5, 4, 5, 4]\n[2, -4, -2, 4]\n\nRequisiti di output:\n- Restituisci esattamente un blocco <answer>...</answer>.\n- Dentro <answer>, scrivi solo la risposta finale.\n- Non includere spiegazioni, ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Calcola il determinante della seguente matrice A:\\n[2, -3, 2, -4]\\n[-3, 4, -1, 5]\\n[-5, 4, 5, 4]\\n[2, -4, -2, 4]\", \"answer\": \"32\", \"metadata\": {\"source_dataset\": \"matrix_determinants\", \"source_index\": 22, \"dimension\": 4, \"matrix\": [[2, -3, 2,...
{ "index": 22, "split": "train", "source_name": "matrix_determinants:it", "language": "it", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
lcm
[ { "role": "user", "content": "Trouvez le plus petit commun multiple (PPCM) de ces nombres : 47, 88\n\nExigences de sortie :\n- Retourne exactement un bloc <answer>...</answer>.\n- Dans <answer>, écris uniquement la réponse finale.\n- N'inclus pas d'explication, d'unité ni de mots supplémentaires dans <answe...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Trouvez le plus petit commun multiple (PPCM) de ces nombres : 47, 88\", \"answer\": \"4136\", \"metadata\": {\"source_dataset\": \"lcm\", \"source_index\": 23, \"numbers\": [47, 88], \"result\": 4136, \"language\": \"fr\", \"difficulty\": {\"numbers\": [2, 2], \"...
{ "index": 23, "split": "train", "source_name": "lcm:fr", "language": "fr", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
binary_matrix
[ { "role": "user", "content": "Gegeben sei eine quadratische Matrix. Berechnen Sie für jede Zelle die Taxicab- (Manhattan-) Distanz zur nächsten 0. \n\nDie Ausgabe sollte eine Matrix derselben Größe wie die Eingabematrix sein, wobei jede Zelle die Distanz zur nächsten 0 enthält.\n\nBerechnen Sie die Distanz ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Gegeben sei eine quadratische Matrix. Berechnen Sie für jede Zelle die Taxicab- (Manhattan-) Distanz zur nächsten 0. \\n\\nDie Ausgabe sollte eine Matrix derselben Größe wie die Eingabematrix sein, wobei jede Zelle die Distanz zur nächsten 0 enthält.\\n\\nBerechn...
{ "index": 24, "split": "train", "source_name": "binary_matrix:de", "language": "de", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
binary_matrix
[ { "role": "user", "content": "正方行列が与えられたとき、各セルから最も近い0までのタクシー距離(マンハッタン距離)を求めてください。\n\n出力は入力行列と同じサイズの行列とし、各セルには最も近い0までの距離が含まれるものとします。\n\n以下の行列の各セルについて、最も近い0までの距離を求めてください:\n0 1 1 1 1\n1 1 0 0 0\n1 1 1 0 0\n0 1 1 1 1\n1 1 1 0 0\n\n出力要件:\n- <answer>...</answer> ブロックを必ず 1 つだけ返してください。\n- <answer> の中には最終回答だけを書いてくださ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"正方行列が与えられたとき、各セルから最も近い0までのタクシー距離(マンハッタン距離)を求めてください。\\n\\n出力は入力行列と同じサイズの行列とし、各セルには最も近い0までの距離が含まれるものとします。\\n\\n以下の行列の各セルについて、最も近い0までの距離を求めてください:\\n0 1 1 1 1\\n1 1 0 0 0\\n1 1 1 0 0\\n0 1 1 1 1\\n1 1 1 0 0\", \"answer\": \"0 1 1 1 1\\n1 1 0 0 0\\n1 2 1 0 0\\n0 1 2 1...
{ "index": 25, "split": "train", "source_name": "binary_matrix:ja", "language": "ja", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
regex_tasks
[ { "role": "user", "content": "Expresión regular: (ba|ab){3}(c)?\nCadena candidata: cbcac\n¿Coincide la cadena candidata completamente con la expresión regular? Responde solo con Sí o No.\n\nRequisitos de salida:\n- Devuelve exactamente un bloque <answer>...</answer>.\n- Dentro de <answer>, escribe solo la r...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Expresión regular: (ba|ab){3}(c)?\\nCadena candidata: cbcac\\n¿Coincide la cadena candidata completamente con la expresión regular? Responde solo con Sí o No.\", \"answer\": \"No\", \"metadata\": {\"language\": \"es\", \"source_dataset\": \"regex_tasks\", \"sourc...
{ "index": 26, "split": "train", "source_name": "regex_tasks:es", "language": "es", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
products
[ { "role": "user", "content": "Calcola il prodotto di: 22057 * 48874.\n\nRequisiti di output:\n- Restituisci esattamente un blocco <answer>...</answer>.\n- Dentro <answer>, scrivi solo la risposta finale.\n- Non includere spiegazioni, unità o parole extra dentro <answer>.\n- Se il compito richiede una format...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Calcola il prodotto di: 22057 * 48874.\", \"answer\": \"1078013818\", \"metadata\": {\"source_dataset\": \"products\", \"source_index\": 27, \"expression\": \"22057 * 48874\", \"num_terms\": 2, \"num_digits\": 5, \"language\": \"it\", \"difficulty\": {\"num_terms...
{ "index": 27, "split": "train", "source_name": "products:it", "language": "it", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
survo
[ { "role": "user", "content": "In diesem 4*4 Survo-Matrix-Rätsel:\n0 0 4 12\n0 0 7 17\n0 3 4 9\n9 14 15 38\nmüssen die 0-Zellen mit Zahlen aus der Menge [3, 5, 2, 8, 2] gefüllt werden. Das letzte Element in jeder Zeile und jeder Spalte ist die Summe aller anderen Elemente in dieser Zeile oder Spalte. Jede Za...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"In diesem 4*4 Survo-Matrix-Rätsel:\\n0 0 4 12\\n0 0 7 17\\n0 3 4 9\\n9 14 15 38\\nmüssen die 0-Zellen mit Zahlen aus der Menge [3, 5, 2, 8, 2] gefüllt werden. Das letzte Element in jeder Zeile und jeder Spalte ist die Summe aller anderen Elemente in dieser Zeile ...
{ "index": 28, "split": "train", "source_name": "survo:de", "language": "de", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
group_anagrams
[ { "role": "user", "content": "Un anagramme est un mot formé en réorganisant les lettres d'un autre mot, en utilisant toutes les lettres d'origine exactement une fois.\n\nRegroupez les anagrammes ci-dessous. Vous pouvez donner la réponse dans n'importe quel ordre.\n\nLa sortie doit être une liste de listes d...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Un anagramme est un mot formé en réorganisant les lettres d'un autre mot, en utilisant toutes les lettres d'origine exactement une fois.\\n\\nRegroupez les anagrammes ci-dessous. Vous pouvez donner la réponse dans n'importe quel ordre.\\n\\nLa sortie doit être un...
{ "index": 29, "split": "train", "source_name": "group_anagrams:fr", "language": "fr", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
binary_matrix
[ { "role": "user", "content": "नीचे दिए गए आव्यूह में प्रत्येक सेल से निकटतम 0 तक की मैनहट्टन दूरी ज्ञात करें।\n\nउसी आकार का एक आव्यूह लौटाएं, जहाँ प्रत्येक प्रविष्टि उस दूरी को संग्रहीत करे।\n\nआव्यूह:\n0 1 0 1 1 1 1\n1 1 1 1 1 1 1\n0 1 0 0 1 1 1\n0 1 1 1 0 1 1\n0 1 1 1 1 1 0\n1 0 1 1 1 0 1\n1 0 0 1 1 0 1\...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"नीचे दिए गए आव्यूह में प्रत्येक सेल से निकटतम 0 तक की मैनहट्टन दूरी ज्ञात करें।\\n\\nउसी आकार का एक आव्यूह लौटाएं, जहाँ प्रत्येक प्रविष्टि उस दूरी को संग्रहीत करे।\\n\\nआव्यूह:\\n0 1 0 1 1 1 1\\n1 1 1 1 1 1 1\\n0 1 0 0 1 1 1\\n0 1 1 1 0 1 1\\n0 1 1 1 1 1 0\\n1 0 ...
{ "index": 30, "split": "train", "source_name": "binary_matrix:hi", "language": "hi", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
futoshiki
[ { "role": "user", "content": "Решите этот пазл Футошики размером 5x5:\n\n_ _ _ _ _\n ∧ \n_ < _ 1 _ 4\n ∧ ∧ \n5 2 _ 4 _\n \n1 < _ 2 _ > _\n \n_ 3 < 5 _ _\n\nИспользуйте тот же формат, что и в пазле выше, заполняя каждое...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Решите этот пазл Футошики размером 5x5:\\n\\n_ _ _ _ _\\n ∧ \\n_ < _ 1 _ 4\\n ∧ ∧ \\n5 2 _ 4 _\\n \\n1 < _ 2 _ > _\\n \\n_ 3 < 5 _ _\\n\\nИспользуйте тот же формат, что и в ...
{ "index": 31, "split": "train", "source_name": "futoshiki:ru", "language": "ru", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
spiral_matrix
[ { "role": "user", "content": "Read the matrix below and list its elements in spiral order, starting from the top-left element.\n\nThe spiral order is clockwise, starting from the top-left corner. More precisely:\n* Start from the top-left corner and move right along the top row.\n* Move down along the right...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Read the matrix below and list its elements in spiral order, starting from the top-left element.\\n\\nThe spiral order is clockwise, starting from the top-left corner. More precisely:\\n* Start from the top-left corner and move right along the top row.\\n* Move d...
{ "index": 32, "split": "train", "source_name": "spiral_matrix", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
list_functions
[ { "role": "user", "content": "Використовуйте приклади, щоб вивести те саме правило для нашого випадку.\nВідповідайте у форматі списку у стилі Python, використовуючи квадратні дужки та коми, без додаткового тексту.\nПриклади:\nКейс 1: [16, 3, 55, 22, 5, 24, 71, 4, 80, 18]\nРезультат 1: [92]\nКейс 2: [86, 92,...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Використовуйте приклади, щоб вивести те саме правило для нашого випадку.\\nВідповідайте у форматі списку у стилі Python, використовуючи квадратні дужки та коми, без додаткового тексту.\\nПриклади:\\nКейс 1: [16, 3, 55, 22, 5, 24, 71, 4, 80, 18]\\nРезультат 1: [92...
{ "index": 33, "split": "train", "source_name": "list_functions:uk", "language": "uk", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
kakurasu
[ { "role": "user", "content": "У вас є ігрове поле головоломки Какурасу розміром 5 x 4. Ваша мета — розмістити одиниці в сітці так, щоб зважені суми відповідали заданим обмеженням: суми рядків [5, 4, 8, 4, 1] та суми стовпців [8, 1, 4, 9].\n1) Правила:\n 1) Кожна клітинка має бути заповнена або 1, або 0.\n ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"У вас є ігрове поле головоломки Какурасу розміром 5 x 4. Ваша мета — розмістити одиниці в сітці так, щоб зважені суми відповідали заданим обмеженням: суми рядків [5, 4, 8, 4, 1] та суми стовпців [8, 1, 4, 9].\\n1) Правила:\\n 1) Кожна клітинка має бути заповнена...
{ "index": 34, "split": "train", "source_name": "kakurasu:uk", "language": "uk", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
shortest_path
[ { "role": "user", "content": "Your task is to find the shortest path from the start to the destination point in a grid.\n\nThe grid is represented as a matrix with the following types of cells:\n* *: your starting point\n* #: your destination point\n* O: an open cell\n* X: a blocked cell\n\nTherefore, you n...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Your task is to find the shortest path from the start to the destination point in a grid.\\n\\nThe grid is represented as a matrix with the following types of cells:\\n* *: your starting point\\n* #: your destination point\\n* O: an open cell\\n* X: a blocked cel...
{ "index": 35, "split": "train", "source_name": "shortest_path", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
simple_equations
[ { "role": "user", "content": "6144*s = 215040을(를) 만족하는 s의 값을 구하시오. 정답만 답하시오.\n\n출력 요구사항:\n- 정확히 하나의 <answer>...</answer> 블록만 반환하세요.\n- <answer> 안에는 최종 답만 작성하세요.\n- <answer> 안에 설명, 단위, 추가 단어를 포함하지 마세요.\n- 과제가 정확한 형식을 요구하면 문자열, 격자, 기호, 공백, 줄바꿈을 그대로 보존하세요." } ]
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"6144*s = 215040을(를) 만족하는 s의 값을 구하시오. 정답만 답하시오.\", \"answer\": \"35\", \"metadata\": {\"source_dataset\": \"simple_equations\", \"source_index\": 36, \"equation\": \"6144*s = 215040\", \"variable\": \"s\", \"language\": \"ko\", \"difficulty\": {\"min_terms\": 2, \...
{ "index": 36, "split": "train", "source_name": "simple_equations:ko", "language": "ko", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
coin_flip
[ { "role": "user", "content": "Qual è la probabilità che almeno 0 di 5 lanci di una moneta equa diano testa?\n\nRequisiti di output:\n- Restituisci esattamente un blocco <answer>...</answer>.\n- Dentro <answer>, scrivi solo la risposta finale.\n- Non includere spiegazioni, unità o parole extra dentro <answer...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Qual è la probabilità che almeno 0 di 5 lanci di una moneta equa diano testa?\", \"answer\": \"1\", \"metadata\": {\"source_dataset\": \"coin_flip\", \"source_index\": 37, \"language\": \"it\", \"num_trials\": 5, \"k_heads\": 0, \"problem_type\": \"at_least\", \"...
{ "index": 37, "split": "train", "source_name": "coin_flip:it", "language": "it", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
fraction_simplification
[ { "role": "user", "content": "将分数 3783/7072 化简为最简分数。仅需给出化简后的分数作为最终答案。\n\n输出要求:\n- 只返回一个 <answer>...</answer> 块。\n- 在 <answer> 内只写最终答案。\n- 不要在 <answer> 内包含解释、单位或额外文字。\n- 如果任务要求精确格式,请严格保留字符串、网格、符号、空格和换行。" } ]
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"将分数 3783/7072 化简为最简分数。仅需给出化简后的分数作为最终答案。\", \"answer\": \"291/544\", \"metadata\": {\"source_dataset\": \"fraction_simplification\", \"source_index\": 38, \"numerator\": 3783, \"denominator\": 7072, \"simplified_numerator\": 291, \"simplified_denominator\": 544, \...
{ "index": 38, "split": "train", "source_name": "fraction_simplification:zh", "language": "zh", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
maze
[ { "role": "user", "content": "Naviga da 'k' (inizio) a 'O' (obiettivo):\n```\nSSSSSSSSS\nS##S#k#SS\nS#S#####S\nS#######S\nS###S###S\nSSSO##S#S\nS###S#SSS\nS#SS##SSS\nSSSSSSSSS\n```\nLegenda:\n'S' = Muro, '#' = Passaggio\nQual è il numero minimo di passi per raggiungere l'obiettivo?\nFornisci solo il numero ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Naviga da 'k' (inizio) a 'O' (obiettivo):\\n```\\nSSSSSSSSS\\nS##S#k#SS\\nS#S#####S\\nS#######S\\nS###S###S\\nSSSO##S#S\\nS###S#SSS\\nS#SS##SSS\\nSSSSSSSSS\\n```\\nLegenda:\\n'S' = Muro, '#' = Passaggio\\nQual è il numero minimo di passi per raggiungere l'obietti...
{ "index": 39, "split": "train", "source_name": "maze:it", "language": "it", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
coin_flip
[ { "role": "user", "content": "Qual è la probabilità che esattamente 2 su 6 lanci di una moneta equa diano testa?\n\nRequisiti di output:\n- Restituisci esattamente un blocco <answer>...</answer>.\n- Dentro <answer>, scrivi solo la risposta finale.\n- Non includere spiegazioni, unità o parole extra dentro <a...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Qual è la probabilità che esattamente 2 su 6 lanci di una moneta equa diano testa?\", \"answer\": \"0.234375\", \"metadata\": {\"source_dataset\": \"coin_flip\", \"source_index\": 40, \"language\": \"it\", \"num_trials\": 6, \"k_heads\": 2, \"problem_type\": \"ex...
{ "index": 40, "split": "train", "source_name": "coin_flip:it", "language": "it", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
maximum_flow
[ { "role": "user", "content": "Ti viene fornito un grafo orientato con capacità.\n\nNodi: 0, 1, 2, 3, 4, 5, 6, 7\nArchi orientati (u, v, capacità):\n(0, 1, 5)\n(0, 2, 18)\n(0, 3, 4)\n(0, 4, 19)\n(0, 5, 15)\n(0, 6, 1)\n(1, 2, 1)\n(1, 3, 10)\n(1, 5, 4)\n(1, 7, 12)\n(2, 1, 4)\n(2, 3, 18)\n(2, 4, 18)\n(2, 5, 20)...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Ti viene fornito un grafo orientato con capacità.\\n\\nNodi: 0, 1, 2, 3, 4, 5, 6, 7\\nArchi orientati (u, v, capacità):\\n(0, 1, 5)\\n(0, 2, 18)\\n(0, 3, 4)\\n(0, 4, 19)\\n(0, 5, 15)\\n(0, 6, 1)\\n(1, 2, 1)\\n(1, 3, 10)\\n(1, 5, 4)\\n(1, 7, 12)\\n(2, 1, 4)\\n(2, ...
{ "index": 41, "split": "train", "source_name": "maximum_flow:it", "language": "it", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
rectangle_count
[ { "role": "user", "content": "Tu tarea es contar cuántos rectángulos hay en una cuadrícula ASCII.\n\nLos rectángulos individuales están delineados con un '#', y las celdas cubiertas por exactamente dos rectángulos se muestran con '█'.\n\nTu respuesta debe ser un solo número que represente el recuento total ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Tu tarea es contar cuántos rectángulos hay en una cuadrícula ASCII.\\n\\nLos rectángulos individuales están delineados con un '#', y las celdas cubiertas por exactamente dos rectángulos se muestran con '█'.\\n\\nTu respuesta debe ser un solo número que represente...
{ "index": 42, "split": "train", "source_name": "rectangle_count:es", "language": "es", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
course_schedule
[ { "role": "user", "content": "Всього вам потрібно пройти 7 курсів, пронумерованих від 0 до 6.\n\nВам надано наступний список попередніх умов; кожна пара (a_i, b_i) означає, що ви повинні спочатку пройти курс b_i, якщо хочете пройти курс a_i:\n[(3, 0), (4, 0), (2, 0), (1, 4), (6, 4), (3, 2), (5, 4), (5, 2)]\...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Всього вам потрібно пройти 7 курсів, пронумерованих від 0 до 6.\\n\\nВам надано наступний список попередніх умов; кожна пара (a_i, b_i) означає, що ви повинні спочатку пройти курс b_i, якщо хочете пройти курс a_i:\\n[(3, 0), (4, 0), (2, 0), (1, 4), (6, 4), (3, 2)...
{ "index": 43, "split": "train", "source_name": "course_schedule:uk", "language": "uk", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
self_reference
[ { "role": "user", "content": "Combien de solutions satisfont les énoncés ci-dessous :\n Énoncé 1: 'Au moins 4 de ces 7 énoncés sont vrais.'\n Énoncé 2: 'Au plus 7 de ces 7 énoncés sont faux.'\n Énoncé 3: 'Exactement 4 de ces 7 énoncés sont vrais.'\n Énoncé 4: 'Exactement 3 de ces 7 énoncés sont faux.'\n Éno...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Combien de solutions satisfont les énoncés ci-dessous :\\n Énoncé 1: 'Au moins 4 de ces 7 énoncés sont vrais.'\\n Énoncé 2: 'Au plus 7 de ces 7 énoncés sont faux.'\\n Énoncé 3: 'Exactement 4 de ces 7 énoncés sont vrais.'\\n Énoncé 4: 'Exactement 3 de ces 7 énoncé...
{ "index": 44, "split": "train", "source_name": "self_reference:fr", "language": "fr", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
coin_flip
[ { "role": "user", "content": "公平なコインを9回投げたとき、少なくとも0回表が出る確率はいくらですか?\n\n出力要件:\n- <answer>...</answer> ブロックを必ず 1 つだけ返してください。\n- <answer> の中には最終回答だけを書いてください。\n- <answer> の中に説明、単位、余分な語句を含めないでください。\n- タスクが厳密な書式を求める場合は、文字列、グリッド、記号、空白、改行を正確に保持してください。" } ]
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"公平なコインを9回投げたとき、少なくとも0回表が出る確率はいくらですか?\", \"answer\": \"1\", \"metadata\": {\"source_dataset\": \"coin_flip\", \"source_index\": 45, \"language\": \"ja\", \"num_trials\": 9, \"k_heads\": 0, \"problem_type\": \"at_least\", \"rational\": {\"numerator\": 512, \"denomi...
{ "index": 45, "split": "train", "source_name": "coin_flip:ja", "language": "ja", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
knight_swap
[ { "role": "user", "content": "Wyzwanie Knight Swap:\n\n```\n A B C D\n ----------------\n3 | | B | . | w |\n ----------------\n2 | . | | | w |\n ----------------\n1 | . | B | . | |\n ----------------\n```\n\nLegenda:\n* 'w' = Biały skoczek\n* 'B' = Czarny skoczek\n* Puste pola są oz...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Wyzwanie Knight Swap:\\n\\n```\\n A B C D\\n ----------------\\n3 | | B | . | w |\\n ----------------\\n2 | . | | | w |\\n ----------------\\n1 | . | B | . | |\\n ----------------\\n```\\n\\nLegenda:\\n* 'w' = Biały skoczek\\n* 'B' = Czar...
{ "index": 46, "split": "train", "source_name": "knight_swap:pl", "language": "pl", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
bitwise_arithmetic
[ { "role": "user", "content": "Berechne diesen Ausdruck:\n((0xc6f6 - 0x6d6c) + (0xcdcd * 0x152d))\n\nVerwende bei negativen Ergebnissen ein Minuszeichen (wie -0x5). Antworte nur mit dem Hexadezimalwert.\n\nAusgabeanforderungen:\n- Gib genau einen <answer>...</answer>-Block zurück.\n- Schreibe innerhalb von <...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Berechne diesen Ausdruck:\\n((0xc6f6 - 0x6d6c) + (0xcdcd * 0x152d))\\n\\nVerwende bei negativen Ergebnissen ein Minuszeichen (wie -0x5). Antworte nur mit dem Hexadezimalwert.\", \"answer\": \"0x11065793\", \"metadata\": {\"source_dataset\": \"bitwise_arithmetic\"...
{ "index": 47, "split": "train", "source_name": "bitwise_arithmetic:de", "language": "de", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
cryptarithm
[ { "role": "user", "content": "Résolvez ce cryptarithme :\n\n MUP\n+ UPSS\n------\n USIS\n\nChaque lettre correspond à un chiffre unique (0-9). Aucun chiffre initial ne peut être 0.\nFournissez, dans votre réponse finale, une correspondance entre les lettres et les chiffres séparée par des virgules qui sa...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Résolvez ce cryptarithme :\\n\\n MUP\\n+ UPSS\\n------\\n USIS\\n\\nChaque lettre correspond à un chiffre unique (0-9). Aucun chiffre initial ne peut être 0.\\nFournissez, dans votre réponse finale, une correspondance entre les lettres et les chiffres séparée ...
{ "index": 48, "split": "train", "source_name": "cryptarithm:fr", "language": "fr", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
bitwise_arithmetic
[ { "role": "user", "content": "इस व्यंजक का मूल्यांकन करें:\n((0x5480 * 0xc1b0) * (0x9fe7 << 0x2))\n\nऋणात्मक उत्तरों के लिए, ऋणात्मक चिह्न का उपयोग करें (जैसे -0x5)। केवल हेक्साडेसिमल मान के साथ उत्तर दें।\n\nआउटपुट आवश्यकताएँ:\n- ठीक एक <answer>...</answer> ब्लॉक लौटाएँ।\n- <answer> के अंदर केवल अंतिम उत्त...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"इस व्यंजक का मूल्यांकन करें:\\n((0x5480 * 0xc1b0) * (0x9fe7 << 0x2))\\n\\nऋणात्मक उत्तरों के लिए, ऋणात्मक चिह्न का उपयोग करें (जैसे -0x5)। केवल हेक्साडेसिमल मान के साथ उत्तर दें।\", \"answer\": \"0x9fbb82cca000\", \"metadata\": {\"source_dataset\": \"bitwise_arit...
{ "index": 49, "split": "train", "source_name": "bitwise_arithmetic:hi", "language": "hi", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
lcm
[ { "role": "user", "content": "求这些数字的最小公倍数 (LCM):94, 68\n\n输出要求:\n- 只返回一个 <answer>...</answer> 块。\n- 在 <answer> 内只写最终答案。\n- 不要在 <answer> 内包含解释、单位或额外文字。\n- 如果任务要求精确格式,请严格保留字符串、网格、符号、空格和换行。" } ]
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"求这些数字的最小公倍数 (LCM):94, 68\", \"answer\": \"3196\", \"metadata\": {\"source_dataset\": \"lcm\", \"source_index\": 50, \"numbers\": [94, 68], \"result\": 3196, \"language\": \"zh\", \"difficulty\": {\"numbers\": [2, 2], \"value\": [1, 100]}}}" }
{ "index": 50, "split": "train", "source_name": "lcm:zh", "language": "zh", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
survo
[ { "role": "user", "content": "इस 5*5 सुरवो (Survo) पहेली को पूरा करें:\n9 3 9 8 29\n0 0 1 0 20\n4 5 0 3 17\n5 3 0 8 23\n27 20 22 20 89\n0 से चिह्नित खानों को भरना है। प्रत्येक पंक्ति और स्तंभ की अंतिम संख्या शेष संख्याओं का योग है। उपयोग करने के लिए संख्याएँ: [9, 1, 9, 7, 5]। प्रत्येक संख्या का उपयोग केवल ए...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"इस 5*5 सुरवो (Survo) पहेली को पूरा करें:\\n9 3 9 8 29\\n0 0 1 0 20\\n4 5 0 3 17\\n5 3 0 8 23\\n27 20 22 20 89\\n0 से चिह्नित खानों को भरना है। प्रत्येक पंक्ति और स्तंभ की अंतिम संख्या शेष संख्याओं का योग है। उपयोग करने के लिए संख्याएँ: [9, 1, 9, 7, 5]। प्रत्येक स...
{ "index": 51, "split": "train", "source_name": "survo:hi", "language": "hi", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
modulo_grid
[ { "role": "user", "content": "이 격자를 정의하는 수학적 패턴을 찾아낸 다음, 그 패턴을 사용하여 물음표를 채우십시오. 여기에 표시된 것과 동일한 크기와 줄 바꿈 형식을 유지하여 완성된 전체 격자를 정답으로 제출하십시오.\n\n✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅\n❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌\n❌✅❌❌✅❌✅❌❌✅❌✅❌❌✅❌✅❌❌✅\n❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌\n❌✅✅✅✅❌✅✅✅✅❌✅✅✅✅❌✅✅✅✅\n❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌\n❌✅❌❌✅❌✅❌❌✅❌✅❌❌✅❌✅❌❌✅\n❌✅❌❌❌❌✅❌❌❌❌✅❌❌...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"이 격자를 정의하는 수학적 패턴을 찾아낸 다음, 그 패턴을 사용하여 물음표를 채우십시오. 여기에 표시된 것과 동일한 크기와 줄 바꿈 형식을 유지하여 완성된 전체 격자를 정답으로 제출하십시오.\\n\\n✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅✅\\n❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌\\n❌✅❌❌✅❌✅❌❌✅❌✅❌❌✅❌✅❌❌✅\\n❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌\\n❌✅✅✅✅❌✅✅✅✅❌✅✅✅✅❌✅✅✅✅\\n❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌❌✅❌❌❌\\n❌✅❌❌✅❌✅❌❌✅❌✅❌...
{ "index": 52, "split": "train", "source_name": "modulo_grid:ko", "language": "ko", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
rotten_oranges
[ { "role": "user", "content": "An NxN grid is provided where each tile can be:\na (an empty space)\nC (a piece of fresh orange)\nW (a piece of rotten orange)\n\nEach step, any piece of fresh orange that is orthogonally adjacent to a piece of rotten orange will also rot. In other words, only left, right, up, ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"An NxN grid is provided where each tile can be:\\na (an empty space)\\nC (a piece of fresh orange)\\nW (a piece of rotten orange)\\n\\nEach step, any piece of fresh orange that is orthogonally adjacent to a piece of rotten orange will also rot. In other words, on...
{ "index": 53, "split": "train", "source_name": "rotten_oranges", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
kakurasu
[ { "role": "user", "content": "Нижче наведено сітку головоломки Какурасу розміром 4 x 5. Ваша мета — розмістити одиниці в сітці так, щоб зважені суми рядків та стовпців відповідали заданим обмеженням. Суми рядків: [2, 5, 9, 10]. Суми стовпців: [9, 1, 3, 6, 7]. \n1) Правила:\n 1) Кожна клітинка повинна місти...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Нижче наведено сітку головоломки Какурасу розміром 4 x 5. Ваша мета — розмістити одиниці в сітці так, щоб зважені суми рядків та стовпців відповідали заданим обмеженням. Суми рядків: [2, 5, 9, 10]. Суми стовпців: [9, 1, 3, 6, 7]. \\n1) Правила:\\n 1) Кожна кліти...
{ "index": 54, "split": "train", "source_name": "kakurasu:uk", "language": "uk", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
course_schedule
[ { "role": "user", "content": "कोर्सों की संख्या 0 से 4 तक है (कुल: 5)।\n\nप्रत्येक युग्म (a, b) का अर्थ है कि b को a से पहले लिया जाना चाहिए:\n[(0, 1), (4, 1), (0, 2), (2, 3), (3, 4), (2, 4), (3, 2), (1, 2)]\n\nक्या निर्भरताओं का उल्लंघन किए बिना सभी कोर्स पूरे किए जा सकते हैं?\nकेवल हाँ या नहीं के साथ उत्त...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"कोर्सों की संख्या 0 से 4 तक है (कुल: 5)।\\n\\nप्रत्येक युग्म (a, b) का अर्थ है कि b को a से पहले लिया जाना चाहिए:\\n[(0, 1), (4, 1), (0, 2), (2, 3), (3, 4), (2, 4), (3, 2), (1, 2)]\\n\\nक्या निर्भरताओं का उल्लंघन किए बिना सभी कोर्स पूरे किए जा सकते हैं?\\nकेवल हा...
{ "index": 55, "split": "train", "source_name": "course_schedule:hi", "language": "hi", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
jugs
[ { "role": "user", "content": "いずれかの水差しに、ちょうど目標量の水が入るように水差しパズルを解いてください。\n操作は「満たす」「空にする」「注ぐ」の3種類です。\n\n水差しAを満たすには、「満たす A」と入力します。\n水差しBを空にするには、「空にする B」と入力します。\n水差しAの内容を水差しBに注ぐには、「注ぐ A->B」と入力します。\n\n最初はすべての水差しが空です。\n\n空の水差しの容量は次の通りです: A:12, B:6, C:13\n目標量は: 9 リットルです。\n\nいずれかの水差しが目標量になるような操作列を、JSON形式のリストで回答してくださ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"いずれかの水差しに、ちょうど目標量の水が入るように水差しパズルを解いてください。\\n操作は「満たす」「空にする」「注ぐ」の3種類です。\\n\\n水差しAを満たすには、「満たす A」と入力します。\\n水差しBを空にするには、「空にする B」と入力します。\\n水差しAの内容を水差しBに注ぐには、「注ぐ A->B」と入力します。\\n\\n最初はすべての水差しが空です。\\n\\n空の水差しの容量は次の通りです: A:12, B:6, C:13\\n目標量は: 9 リットルです。\\n\\nいずれかの水差しが目標量にな...
{ "index": 56, "split": "train", "source_name": "jugs:ja", "language": "ja", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
largest_island
[ { "role": "user", "content": "Se te proporciona la siguiente matriz binaria de 8 x 6:\n0 0 1 1 0 1\n0 0 1 1 1 1\n0 0 0 1 1 1\n1 1 1 1 0 0\n1 1 1 1 0 0\n1 1 0 0 0 0\n1 0 0 0 0 0\n1 1 1 0 0 0\n\nUna isla es un grupo de 1s (que representan tierra) conectados en 4 direcciones (horizontal o verticalmente).\nPued...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Se te proporciona la siguiente matriz binaria de 8 x 6:\\n0 0 1 1 0 1\\n0 0 1 1 1 1\\n0 0 0 1 1 1\\n1 1 1 1 0 0\\n1 1 1 1 0 0\\n1 1 0 0 0 0\\n1 0 0 0 0 0\\n1 1 1 0 0 0\\n\\nUna isla es un grupo de 1s (que representan tierra) conectados en 4 direcciones (horizonta...
{ "index": 57, "split": "train", "source_name": "largest_island:es", "language": "es", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
coin_flip
[ { "role": "user", "content": "公平なコインを15回投げます。ちょうど9回表が出る確率はいくらですか?\n\n出力要件:\n- <answer>...</answer> ブロックを必ず 1 つだけ返してください。\n- <answer> の中には最終回答だけを書いてください。\n- <answer> の中に説明、単位、余分な語句を含めないでください。\n- タスクが厳密な書式を求める場合は、文字列、グリッド、記号、空白、改行を正確に保持してください。" } ]
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"公平なコインを15回投げます。ちょうど9回表が出る確率はいくらですか?\", \"answer\": \"0.1527404785\", \"metadata\": {\"source_dataset\": \"coin_flip\", \"source_index\": 58, \"language\": \"ja\", \"num_trials\": 15, \"k_heads\": 9, \"problem_type\": \"exact\", \"rational\": {\"numerator\": 5005,...
{ "index": 58, "split": "train", "source_name": "coin_flip:ja", "language": "ja", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
number_filtering
[ { "role": "user", "content": "Keep all numbers larger than -34.087 in this list: ['2', '-12', '-43.932', '15.03', '-60.062', '92.2', '-51.344', '29.01', '41.6']\nReturn the new list in the same format.\n\nOutput requirements:\n- Return exactly one <answer>...</answer> block.\n- Inside <answer>, write only t...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Keep all numbers larger than -34.087 in this list: ['2', '-12', '-43.932', '15.03', '-60.062', '92.2', '-51.344', '29.01', '41.6']\\nReturn the new list in the same format.\", \"answer\": \"['2', '-12', '15.03', '92.2', '29.01', '41.6']\", \"metadata\": {\"source...
{ "index": 59, "split": "train", "source_name": "number_filtering", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
coin_flip
[ { "role": "user", "content": "Ao lançar uma moeda justa 11 vezes, qual é a probabilidade de obter pelo menos 11 caras?\n\nRequisitos de saída:\n- Retorne exatamente um bloco <answer>...</answer>.\n- Dentro de <answer>, escreva apenas a resposta final.\n- Não inclua explicação, unidades ou palavras extras de...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Ao lançar uma moeda justa 11 vezes, qual é a probabilidade de obter pelo menos 11 caras?\", \"answer\": \"0.00048828125\", \"metadata\": {\"source_dataset\": \"coin_flip\", \"source_index\": 60, \"language\": \"pt\", \"num_trials\": 11, \"k_heads\": 11, \"problem...
{ "index": 60, "split": "train", "source_name": "coin_flip:pt", "language": "pt", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
vm_trace
[ { "role": "user", "content": "指令集:\n* MOV X, Y : 将寄存器 X 设置为值 Y(Y 可以是整数或寄存器)。\n* ADD X, Y : 设置 X = X + Y。\n* SUB X, Y : 设置 X = X - Y。\n* CMP X, Y : 比较 X 和 Y 并存储比较结果。\n* JNZ LBL : 如果最后一次 CMP 的结果不相等,则跳转到标签 LBL。\n* HALT : 停止执行。\n寄存器为有符号整数,除非通过 MOV 赋值,否则初始值均为 0。\n一个周期意味着执行一条非标签指令。\n\n程序:\n1) MOV R0, 3\n2) MOV R1...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"指令集:\\n* MOV X, Y : 将寄存器 X 设置为值 Y(Y 可以是整数或寄存器)。\\n* ADD X, Y : 设置 X = X + Y。\\n* SUB X, Y : 设置 X = X - Y。\\n* CMP X, Y : 比较 X 和 Y 并存储比较结果。\\n* JNZ LBL : 如果最后一次 CMP 的结果不相等,则跳转到标签 LBL。\\n* HALT : 停止执行。\\n寄存器为有符号整数,除非通过 MOV 赋值,否则初始值均为 0。\\n一个周期意味着执行一条非标签指令。\\n\\n程序:...
{ "index": 61, "split": "train", "source_name": "vm_trace:zh", "language": "zh", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
uniform_acceleration
[ { "role": "user", "content": "A particle starts with initial velocity 33 m/s under constant acceleration -3 m/s^2 for 9 seconds. First, find the final velocity using v = u + at. Then, find the exact displacement traveled using s = ut + (1/2)at^2. Give your final answer (displacement) as a simplified fractio...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"A particle starts with initial velocity 33 m/s under constant acceleration -3 m/s^2 for 9 seconds. First, find the final velocity using v = u + at. Then, find the exact displacement traveled using s = ut + (1/2)at^2. Give your final answer (displacement) as a sim...
{ "index": 62, "split": "train", "source_name": "uniform_acceleration", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
string_manipulation
[ { "role": "user", "content": "Продовжуйте переписувати рядок за наведеними нижче правилами, доки процес не стабілізується або не повернеться до попереднього стану.\n\nПравила:\n1. Якщо рядок містить парну кількість 'b' (і хоча б один 'b'), додайте 'ab' у кінець.\n2. Якщо рядок починається з 'ab', замініть ц...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Продовжуйте переписувати рядок за наведеними нижче правилами, доки процес не стабілізується або не повернеться до попереднього стану.\\n\\nПравила:\\n1. Якщо рядок містить парну кількість 'b' (і хоча б один 'b'), додайте 'ab' у кінець.\\n2. Якщо рядок починається...
{ "index": 63, "split": "train", "source_name": "string_manipulation:uk", "language": "uk", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
ransom_note
[ { "role": "user", "content": "Can you spell out the word `lahvi` by using letters found in `ujvaukiufh`?\nYou cannot reuse any letter. Answer only with True or False.\n\n\nOutput requirements:\n- Return exactly one <answer>...</answer> block.\n- Inside <answer>, write only the final answer.\n- Do not includ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Can you spell out the word `lahvi` by using letters found in `ujvaukiufh`?\\nYou cannot reuse any letter. Answer only with True or False.\\n\", \"answer\": \"False\", \"metadata\": {\"source_dataset\": \"ransom_note\", \"source_index\": 64, \"language\": \"en\", ...
{ "index": 64, "split": "train", "source_name": "ransom_note", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
diffie_hellman
[ { "role": "user", "content": "Diffie-Hellman 키 교환에서 공개 파라미터는 소수 p = 41와 생성자 g = 7입니다. Alice의 공개값은 A = 15이고 Bob의 공개값은 B = 8입니다. 숨겨진 개인 지수는 A = g^a mod p 및 B = g^b mod p를 만족하며, 공유 비밀값은 s = B^a mod p (또는 s = A^b mod p)입니다. 이 숫자들은 작으므로, p에 대한 나머지 연산을 통해 지수를 테스트하여 a 또는 b를 찾아낸 뒤 s를 계산할 수 있습니다. s의 정수 값만 반환하세요.\n\n...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Diffie-Hellman 키 교환에서 공개 파라미터는 소수 p = 41와 생성자 g = 7입니다. Alice의 공개값은 A = 15이고 Bob의 공개값은 B = 8입니다. 숨겨진 개인 지수는 A = g^a mod p 및 B = g^b mod p를 만족하며, 공유 비밀값은 s = B^a mod p (또는 s = A^b mod p)입니다. 이 숫자들은 작으므로, p에 대한 나머지 연산을 통해 지수를 테스트하여 a 또는 b를 찾아낸 뒤 s를 계산할 수 있습니다. s의 정...
{ "index": 65, "split": "train", "source_name": "diffie_hellman:ko", "language": "ko", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
checkers_capture
[ { "role": "user", "content": "给你一个 8x8 的跳棋盘。\n\n棋子:\n+ b = 黑棋\n+ w = 白棋\n+ . = 空格\n\n本任务规则:\n+ 黑方向下移动(向较大的行号方向移动)。\n+ 白方向上移动(向较小的行号方向移动)。\n+ 黑棋 通过斜向跳过一个相邻的 白棋 到达一个空格来进行吃子。\n+ 仅考虑单步跳跃;忽略任何连续的多步跳跃。\n+ 恰好存在一个 黑方 的吃子机会,且吃子是强制性的。\n\n坐标采用从 1 开始的索引,格式为 (行, 列)。\n\n棋盘:\n 1 2 3 4 5 6 7 8\n1 | . . . . . . . .\n2 | ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"给你一个 8x8 的跳棋盘。\\n\\n棋子:\\n+ b = 黑棋\\n+ w = 白棋\\n+ . = 空格\\n\\n本任务规则:\\n+ 黑方向下移动(向较大的行号方向移动)。\\n+ 白方向上移动(向较小的行号方向移动)。\\n+ 黑棋 通过斜向跳过一个相邻的 白棋 到达一个空格来进行吃子。\\n+ 仅考虑单步跳跃;忽略任何连续的多步跳跃。\\n+ 恰好存在一个 黑方 的吃子机会,且吃子是强制性的。\\n\\n坐标采用从 1 开始的索引,格式为 (行, 列)。\\n\\n棋盘:\\n 1 2 3 4 5 ...
{ "index": 66, "split": "train", "source_name": "checkers_capture:zh", "language": "zh", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
futoshiki
[ { "role": "user", "content": "Решите этот пазл Футошики размером 8x8:\n\n_ < _ _ > 4 > _ < _ _ _\n ∧ \n_ _ 8 _ _ > _ 1 6\n ∨ ∨ ∧ ∧ \n7 _ 4 8 _ 5 < 6 2\n ∨ ∨ ∧ ∨ ∧ ∧\n_ _ _ 2 _ > 4 < _ _\n∧ ...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Решите этот пазл Футошики размером 8x8:\\n\\n_ < _ _ > 4 > _ < _ _ _\\n ∧ \\n_ _ 8 _ _ > _ 1 6\\n ∨ ∨ ∧ ∧ \\n7 _ 4 8 _ 5 < 6 2\\n ∨ ∨ ∧ ∨ ∧ ∧\\n_ _ _ 2 _ > 4 < _ ...
{ "index": 67, "split": "train", "source_name": "futoshiki:ru", "language": "ru", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
tic_tac_toe
[ { "role": "user", "content": "三目並べの盤面状態が与えられます。\n\nプレイヤー1は「X」を使用します。\nプレイヤー2は「O」を使用します。\n現在は プレイヤー1 (X) のターンです。\n\n盤面:\n_ _ _\nX O X\nO O X\n\n両プレイヤーが最善を尽くした場合、ゲームの結果はどうなりますか?\n以下のいずれか1つのみで答えてください:\n- プレイヤー1\n- プレイヤー2\n- 引き分け\n\n\n出力要件:\n- <answer>...</answer> ブロックを必ず 1 つだけ返してください。\n- <answer> の中には最終回答だけを書い...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"三目並べの盤面状態が与えられます。\\n\\nプレイヤー1は「X」を使用します。\\nプレイヤー2は「O」を使用します。\\n現在は プレイヤー1 (X) のターンです。\\n\\n盤面:\\n_ _ _\\nX O X\\nO O X\\n\\n両プレイヤーが最善を尽くした場合、ゲームの結果はどうなりますか?\\n以下のいずれか1つのみで答えてください:\\n- プレイヤー1\\n- プレイヤー2\\n- 引き分け\\n\", \"answer\": \"プレイヤー1\", \"metadata\": {\"langu...
{ "index": 68, "split": "train", "source_name": "tic_tac_toe:ja", "language": "ja", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
simple_geometry
[ { "role": "user", "content": "Рассмотрим выпуклый 5-угольник, первые 4 внутренних углов которого равны: 166.0°, 106.0°, 50.0°, 150.0°. Найдите величину оставшегося угла. Выведите только одно число; не используйте символы и слова.\n\nТребования к выводу:\n- Верните ровно один блок <answer>...</answer>.\n- Вн...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Рассмотрим выпуклый 5-угольник, первые 4 внутренних углов которого равны: 166.0°, 106.0°, 50.0°, 150.0°. Найдите величину оставшегося угла. Выведите только одно число; не используйте символы и слова.\", \"answer\": \"68\", \"metadata\": {\"source_dataset\": \"sim...
{ "index": 69, "split": "train", "source_name": "simple_geometry:ru", "language": "ru", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
matrix_determinants
[ { "role": "user", "content": "计算下列矩阵 A 的行列式:\n[5, 3, 0]\n[-3, -5, 1]\n[3, 1, 1]\n\n输出要求:\n- 只返回一个 <answer>...</answer> 块。\n- 在 <answer> 内只写最终答案。\n- 不要在 <answer> 内包含解释、单位或额外文字。\n- 如果任务要求精确格式,请严格保留字符串、网格、符号、空格和换行。" } ]
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"计算下列矩阵 A 的行列式:\\n[5, 3, 0]\\n[-3, -5, 1]\\n[3, 1, 1]\", \"answer\": \"-12\", \"metadata\": {\"source_dataset\": \"matrix_determinants\", \"source_index\": 70, \"dimension\": 3, \"matrix\": [[5, 3, 0], [-3, -5, 1], [3, 1, 1]], \"language\": \"zh\", \"difficulty\":...
{ "index": 70, "split": "train", "source_name": "matrix_determinants:zh", "language": "zh", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
base_conversion
[ { "role": "user", "content": "Convert the base-15 number dd to its equivalent in hexadecimal.\nFor bases greater than 10, represent digits above 9 with the lowercase letters a-z.\n\n\nOutput requirements:\n- Return exactly one <answer>...</answer> block.\n- Inside <answer>, write only the final answer.\n- D...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Convert the base-15 number dd to its equivalent in hexadecimal.\\nFor bases greater than 10, represent digits above 9 with the lowercase letters a-z.\\n\", \"answer\": \"d0\", \"metadata\": {\"source_dataset\": \"base_conversion\", \"source_index\": 71, \"languag...
{ "index": 71, "split": "train", "source_name": "base_conversion", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
diffie_hellman
[ { "role": "user", "content": "In a Diffie-Hellman key exchange, public parameters are prime p = 41 and generator g = 7. Alice's public value is A = 15 and Bob's public value is B = 8. The hidden private exponents satisfy A = g^a mod p and B = g^b mod p, and the shared secret is s = B^a mod p (equivalently s...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"In a Diffie-Hellman key exchange, public parameters are prime p = 41 and generator g = 7. Alice's public value is A = 15 and Bob's public value is B = 8. The hidden private exponents satisfy A = g^a mod p and B = g^b mod p, and the shared secret is s = B^a mod p ...
{ "index": 72, "split": "train", "source_name": "diffie_hellman", "language": "en", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
string_synthesis
[ { "role": "user", "content": "Existem nove blocos diferentes {R} {O} {P} <R> <O> <P> (R) (O) (P) \n1) Um {R}, um {O} e um {P} podem ser combinados para formar um <R>. \n2) Um {R} e um {O} podem ser combinados para formar um <P>. \n3) Um {O} e um {P} podem ser combinados para formar um <O>. \n4) Dois {P} pod...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Existem nove blocos diferentes {R} {O} {P} <R> <O> <P> (R) (O) (P) \\n1) Um {R}, um {O} e um {P} podem ser combinados para formar um <R>. \\n2) Um {R} e um {O} podem ser combinados para formar um <P>. \\n3) Um {O} e um {P} podem ser combinados para formar um <O>....
{ "index": 73, "split": "train", "source_name": "string_synthesis:pt", "language": "pt", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
simple_geometry
[ { "role": "user", "content": "Étant donné un polygone convexe à 4 côtés, ses 3 premiers angles intérieurs sont : 145.0°, 101.0°, 69.0°. Quelle est la mesure de l'angle intérieur restant (en degrés) ? Répondez par la valeur de l'angle manquant sous la forme d'un seul nombre uniquement.\n\nExigences de sortie...
rgym
{ "style": "rule", "ground_truth": "{\"question\": \"Étant donné un polygone convexe à 4 côtés, ses 3 premiers angles intérieurs sont : 145.0°, 101.0°, 69.0°. Quelle est la mesure de l'angle intérieur restant (en degrés) ? Répondez par la valeur de l'angle manquant sous la forme d'un seul nombre uniquement.\", \"an...
{ "index": 74, "split": "train", "source_name": "simple_geometry:fr", "language": "fr", "developer_prompt": "none", "system_prompt_used": false, "prompt_suffix_used": true }
End of preview. Expand in Data Studio

reasoning-gym-verl-datasets

This dataset contains procedurally generated reasoning tasks from the Reasoning Gym (r-gym) framework, structured and pre-processed in parquet format for training models with veRL.

These datasets were used to train MauroPello/Qwen3-1.7B-RL-final using GRPO (Group Relative Policy Optimization).

Dataset Splits & Structure

Split Name Path Size (Examples) Description
train train.parquet 100,000 Raw training set containing randomly sampled tasks across English and translation languages.
train_interleaved train_interleaved.parquet 100,000 Stratified training set where rows are interleaved across tasks and languages to stabilize learning and gradient updates in RL.
validation val.parquet 10,000 Main validation dataset.
validation_mini val_mini.parquet ~1,000 Mini validation dataset containing one example per task-language specification.
held_out_languages held_out_languages_validation.parquet 2,000 Generalization test set covering languages not included in SFT or training (Dutch nl, Turkish tr).
held_out_tasks held_out_tasks.parquet 1,040 Generalization test set covering tasks held out during training (e.g. circuit_logic, codeio, emoji_mystery, game_of_life_halting, propositional_logic, rush_hour, string_insertion, tsumego).

Usage

You can load these splits directly using the Hugging Face datasets library:

from datasets import load_dataset

# Load train_interleaved (used for RL training)
dataset = load_dataset("MauroPello/reasoning-gym-verl-datasets", split="train_interleaved")

# Load generalization tasks validation split
held_out_tasks = load_dataset("MauroPello/reasoning-gym-verl-datasets", split="held_out_tasks")

Each example contains:

  • data_source: The identifier of the Reasoning Gym task.
  • prompt: Conversational turn list containing the system instructions and user request.
  • ability: Capability tag (always rgym).
  • reward_model: Dictionary containing the expected ground_truth and matching rule style.
  • extra_info: Metadata specifying the target language, original indices, task name, and developer prompt options.
Downloads last month
58

Models trained or fine-tuned on MauroPello/reasoning-gym-verl-datasets