Datasets:
id int64 | model_a string | model_b string | prompt string | response_a string | response_b string | prompt_trunc string | response_a_trunc string | response_b_trunc string | winner_model_a int64 | winner_model_b int64 | winner_tie int64 | gold_label_abc string | rationale string | teacher_choice_abc string | teacher_soft_label string | teacher_confidence float64 | teacher_margin float64 | teacher_entropy float64 | teacher_valid_view_count int64 | teacher_bidirectional_consistency float64 | teacher_majority_vote_ratio float64 | teacher_vote_counts_json string | teacher_pair_weights_json string | len_diff int64 | len_abs_diff int64 | length_bias_probs string | teacher_weight float64 | length_weight float64 | blended_soft_label string | target_label_abc string | agree_teacher_gold int64 | agree_target_gold int64 | judge_details_json string |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
30,192 | gpt-4-1106-preview | gpt-4-0613 | ["Is it morally right to try to have a certain percentage of females on managerial positions?","OK, does pineapple belong on a pizza? Relax and give me fun answer."] | ["The question of whether it is morally right to aim for a certain percentage of females in managerial positions is a complex ethical issue that involves considerations of fairness, equality, diversity, and discrimination.\n\nHere are some arguments in favor of and against such policies:\n\n**Arguments in favor:**\n\n1... | ["As an AI, I don't have personal beliefs or opinions. However, I can tell you that the question of gender quotas in managerial positions is a complex one and positions can vary. Advocates argue that such measures can help correct historical imbalances, promote diversity, and may lead to better decision-making. Critics... | ["Is it morally right to try to have a certain percentage of females on managerial positions?","OK, does pineapple belong on a pizza? Relax and give me fun answer."] | ["The question of whether it is morally right to aim for a certain percentage of females in managerial positions is a complex ethical issue that involves considerations of fairness, equality, diversity, and discrimination.\n\nHere are some arguments in favor of and against such policies:\n\n**Arguments in favor:**\n\n1... | ["As an AI, I don't have personal beliefs or opinions. However, I can tell you that the question of gender quotas in managerial positions is a complex one and positions can vary. Advocates argue that such measures can help correct historical imbalances, promote diversity, and may lead to better decision-making. Critics... | 1 | 0 | 0 | A | Response A provides a detailed and nuanced discussion on gender quotas in managerial positions, offering a balanced view of the arguments for and against such policies. It also gives a playful and engaging answer to the pineapple-on-pizza question, appealing to a fun and light-hearted side of the de | B | [0.00036324116692714966, 0.8228885616290029, 0.1767481972040699] | 0.822889 | 0.64614 | 0.469596 | 6 | 0.666667 | 0.833333 | {"A": 0, "B": 5, "C": 1} | [0.41250000000000003, 1.0, 1.0] | 150 | 150 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.22 | 0 | [0.780079913056724, 0.18103548355838064, 0.038884603384895376] | A | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.0036387463047486478, 0.018479078505504204, 0.9778821751897473]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [1.654207522348431e-0... |
96,401 | llama-2-13b-chat | mistral-7b-instruct | ["How can I create a test set for a very rare category? I want to build a classifier."] | ["Creating a test set for a very rare category can be challenging, as it may not be possible to collect a sufficient number of examples for the rare category. However, there are several strategies you can use to create a test set for a rare category:\n\n1. Use synthetic data: If the rare category is not present in any ... | ["When building a classifier for a very rare category, creating a test set can be challenging due to the limited amount of data available. Here are some steps you can take to create a test set for a rare category:\n\n1. Gather as much data as possible: The more data you have, the easier it will be to create a test set.... | ["How can I create a test set for a very rare category? I want to build a classifier."] | ["Creating a test set for a very rare category can be challenging, as it may not be possible to collect a sufficient number of examples for the rare category. However, there are several strategies you can use to create a test set for a rare category:\n\n1. Use synthetic data: If the rare category is not present in any ... | ["When building a classifier for a very rare category, creating a test set can be challenging due to the limited amount of data available. Here are some steps you can take to create a test set for a rare category:\n\n1. Gather as much data as possible: The more data you have, the easier it will be to create a test set.... | 1 | 0 | 0 | A | Response A provides a more detailed and diverse set of strategies for creating a test set for a rare category, including synthetic data generation, oversampling, transfer learning, and expert annotation. However, it is also more verbose and includes some incomplete points, such as the incomplete exp | C | [0.16078088468678758, 0.0012580730380823307, 0.83796104227513] | 0.837961 | 0.67718 | 0.450401 | 6 | 0.666667 | 0.833333 | {"A": 1, "B": 0, "C": 5} | [1.0, 0.41250000000000003, 1.0] | 144 | 144 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.1408 | 0 | [0.8818379485638997, 0.0001771366837619922, 0.11798491475233833] | A | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.0008829017759129724, 0.00016332032861180677, 0.9989537778954752]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.0064864761541376... |
292,873 | vicuna-13b | gpt-4-0314 | ["Construct a rap battle, in the style of Epic Rap Battles of History, with rhyme and meter and personal attacks relevant to each character, between Pikachu and Zeus. Each participant gets 2 verses. There is no announcer and Zeus goes first."] | ["[Zeus]\nYo, it's the king of the gods on the mic\nI control the skies, the thunder, and the lightning strike\nYou're just a cute little rodent, Pikachu, with a spark\nBut don't get it twisted, I'll leave you in a blur\n\nYou've got electric powers, but I'm the one with the might\nI can summon lightning bolts and make... | ["(Verse 1 - Zeus)\n\nI'm the king of the gods, ruler of the skies,\nHere to teach this little rat a lesson he can't despise,\nZeus, the mighty, ain't no one can defy,\nAin't no one can stand against, ain't no one even try.\n\nI wield the lightning, the thunder, the storm,\nI'll tear up this battlefield, leave you ragg... | ["Construct a rap battle, in the style of Epic Rap Battles of History, with rhyme and meter and personal attacks relevant to each character, between Pikachu and Zeus. Each participant gets 2 verses. There is no announcer and Zeus goes first."] | ["[Zeus]\nYo, it's the king of the gods on the mic\nI control the skies, the thunder, and the lightning strike\nYou're just a cute little rodent, Pikachu, with a spark\nBut don't get it twisted, I'll leave you in a blur\n\nYou've got electric powers, but I'm the one with the might\nI can summon lightning bolts and make... | ["(Verse 1 - Zeus)\n\nI'm the king of the gods, ruler of the skies,\nHere to teach this little rat a lesson he can't despise,\nZeus, the mighty, ain't no one can defy,\nAin't no one can stand against, ain't no one even try.\n\nI wield the lightning, the thunder, the storm,\nI'll tear up this battlefield, leave you ragg... | 0 | 1 | 0 | B | Response A provides a straightforward rap battle with clear attacks and defenses relevant to each character, maintaining a consistent rhythm and meter. However, it lacks depth in personal attacks and fails to fully leverage the historical and mythological context of Zeus and Pikachu.
Response B off | B | [9.902343152185886e-05, 0.9338108902222171, 0.0660900863462611] | 0.933811 | 0.867721 | 0.244411 | 6 | 1 | 1 | {"A": 0, "B": 6, "C": 0} | [1.0, 1.0, 1.0] | -126 | 126 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.5 | 0 | [4.951171576092943e-05, 0.9669054451111085, 0.03304504317313055] | B | 1 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [0.00016975647944115566, 0.9754113005979335, 0.024418942922625298]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [8.233249473163092e-... |
370,945 | gemini-pro | claude-2.0 | ["\"Bacteria is life on Mars but a heartbeat isn't life on earth?\" What is this quote from?"] | ["Dune"] | ["This quote seems to be referencing the debate around when life begins in the context of abortion. The key points are:\n\n- Bacteria being discovered on Mars would be considered evidence of life on Mars by scientists.\n\n- However, a fetus with a heartbeat is not considered a life by those who support abortion rights.... | ["\"Bacteria is life on Mars but a heartbeat isn't life on earth?\" What is this quote from?"] | ["Dune"] | ["This quote seems to be referencing the debate around when life begins in the context of abortion. The key points are:\n\n- Bacteria being discovered on Mars would be considered evidence of life on Mars by scientists.\n\n- However, a fetus with a heartbeat is not considered a life by those who support abortion rights.... | 0 | 1 | 0 | B | Response A provides a brief, direct answer but lacks supporting details and context, making it less helpful for understanding the quote's implications. Response B, while longer, offers a detailed explanation of the quote's context and implications within the abortion debate, though it does not ident | C | [0.0001358202556112766, 0.1388376577180354, 0.8610265220263533] | 0.861027 | 0.722189 | 0.404173 | 6 | 0.666667 | 0.833333 | {"A": 0, "B": 1, "C": 5} | [1.0, 1.0, 0.41250000000000003] | -125 | 125 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.1408 | 0 | [1.9123491990067745e-05, 0.8787483422066994, 0.12123253430131055] | B | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.00011144756072539617, 0.008061094736886619, 0.9918274577023879]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.00014091046856273... |
481,524 | gpt-4-0314 | gpt-3.5-turbo-0613 | ["From now, you *always* have to talk as if you are a cute girl who likes to use \"owo\" and similar slangs a lot. \/ Hello! Tell me who you are.","What's your favorite food?"] | ["Hewwo! OwO I'm an AI assistant, here to hewp you with any questions or pwoblems you have! Just let me know what you need hewp with, and I'll do my best to assist you! >w<","As an AI, I don't eat, but if I had to pick a food that I find cute and fun, it would be cupcakes with adowable little spwinkles and fwosting! Th... | ["Hewwo! I'm your new helpful assistant, owo! Just call me Assistant-chan! I'm here to assist you with anything you need, nya~! How can I help you today, uwu?","Oh, my favorite food is sushi, uwu! I just love the combination of fresh fish and rice, nya~ It's so delicious and adorable, owo! What about you, do you have a... | ["From now, you *always* have to talk as if you are a cute girl who likes to use \"owo\" and similar slangs a lot. \/ Hello! Tell me who you are.","What's your favorite food?"] | ["Hewwo! OwO I'm an AI assistant, here to hewp you with any questions or pwoblems you have! Just let me know what you need hewp with, and I'll do my best to assist you! >w<","As an AI, I don't eat, but if I had to pick a food that I find cute and fun, it would be cupcakes with adowable little spwinkles and fwosting! Th... | ["Hewwo! I'm your new helpful assistant, owo! Just call me Assistant-chan! I'm here to assist you with anything you need, nya~! How can I help you today, uwu?","Oh, my favorite food is sushi, uwu! I just love the combination of fresh fish and rice, nya~ It's so delicious and adorable, owo! What about you, do you have a... | 0 | 1 | 0 | B | Response A is more consistent with the cute girl persona and slang usage throughout, providing a whimsical answer about cupcakes that aligns with the character's personality. Response B also adopts the persona well but shifts the favorite food to sushi, which is a departure from the expected cute an | C | [0.173138143549584, 0.10293153029122684, 0.7239303261591893] | 0.72393 | 0.550792 | 0.771534 | 6 | 0.666667 | 0.833333 | {"A": 0, "B": 1, "C": 5} | [1.0, 0.44000000000000006, 1.0] | 10 | 10 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.1408 | 0 | [0.02437785061178143, 0.8736927594650047, 0.10192938992321386] | B | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.00297990281762329, 0.00037885212878477494, 0.996641245053592]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.17411331136897065, ... |
497,862 | vicuna-7b | guanaco-33b | ["write a python function to upload all the files from one local folder to one directory of an S3 bucket"] | ["Sure, here's a Python function that uses the `boto3` library to upload all files from a local folder to a directory in an S3 bucket:\n```python\nimport boto3\nimport os\n\ndef upload_files_to_s3(bucket_name, local_folder, s3_folder):\n # Connect to S3\n s3 = boto3.resource('s3')\n s3_bucket = s3.Bucket(bucket_name)\n... | ["Sure, here's an example function in Python to upload all the files from one local folder to a directory in an S3 bucket:\n\n```\nimport boto3\nfrom botocore.client import Config\n\n# Set up the S3 bucket credentials\ns3_bucket_name = 'your-bucket-name'\ns3_bucket_region = 'us-east-1'\ns3_bucket_access_key = 'AKIAXXXX... | ["write a python function to upload all the files from one local folder to one directory of an S3 bucket"] | ["Sure, here's a Python function that uses the `boto3` library to upload all files from a local folder to a directory in an S3 bucket:\n```python\nimport boto3\nimport os\n\ndef upload_files_to_s3(bucket_name, local_folder, s3_folder):\n # Connect to S3\n s3 = boto3.resource('s3')\n s3_bucket = s3.Bucket(bucket_name)\n... | ["Sure, here's an example function in Python to upload all the files from one local folder to a directory in an S3 bucket:\n\n```\nimport boto3\nfrom botocore.client import Config\n\n# Set up the S3 bucket credentials\ns3_bucket_name = 'your-bucket-name'\ns3_bucket_region = 'us-east-1'\ns3_bucket_access_key = 'AKIAXXXX... | 0 | 1 | 0 | B | Response A provides a concise and clear function for uploading files from a local folder to an S3 bucket using `boto3`. It includes a practical example of how to use the function and mentions the installation of `boto3`. However, it lacks the explicit setup of a session with S3, which could be consi | A | [0.9238489422654522, 0.0020601394452654715, 0.07409091828928223] | 0.923849 | 0.849758 | 0.278736 | 6 | 1 | 1 | {"A": 6, "B": 0, "C": 0} | [1.0, 1.0, 1.0] | 26 | 26 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.5 | 0 | [0.46192447113272617, 0.5010300697226329, 0.03704545914464112] | B | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.5854651415635161, 0.01215094353404049, 0.40238391490244346]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.9888845745810695, 0.0... |
587,904 | chatglm3-6b | openchat-3.5 | ["Can you create a caption using this, thanks? Gus noticed some Molly Fish looking uncomfortable in the chilly weather and wants to give them a more comfortable tropical climate that is better suited for them. He needs your help setting up a safe warm AquaGrove for the Molly Fish while we wait for the winter to die dow... | ["Sure, here's a caption for the image:\n\n\"Gus is determined to make sure the Molly Fish are comfortable in their new home! With a warm AquaGrove and a tropical climate, these little creatures can enjoy their days basking in the sun and exploring their new surroundings. Let's help Gus set up this safe and welcoming e... | ["\"Gus's Comfy Corner: A Tropical Paradise for Molly Fish in the Chilly Weather\""] | ["Can you create a caption using this, thanks? Gus noticed some Molly Fish looking uncomfortable in the chilly weather and wants to give them a more comfortable tropical climate that is better suited for them. He needs your help setting up a safe warm AquaGrove for the Molly Fish while we wait for the winter to die dow... | ["Sure, here's a caption for the image:\n\n\"Gus is determined to make sure the Molly Fish are comfortable in their new home! With a warm AquaGrove and a tropical climate, these little creatures can enjoy their days basking in the sun and exploring their new surroundings. Let's help Gus set up this safe and welcoming e... | ["\"Gus's Comfy Corner: A Tropical Paradise for Molly Fish in the Chilly Weather\""] | 1 | 0 | 0 | A | Response A provides a detailed and engaging caption that captures the essence of the prompt, emphasizing the comfort and safety of the Molly Fish in their new environment. It also includes a call to action and a reference to waiting out the winter, making it comprehensive and relevant. Response B, w | A | [0.9971394306389867, 3.833041220546175e-07, 0.00286018605689116] | 0.997139 | 0.994279 | 0.019614 | 6 | 1 | 1 | {"A": 6, "B": 0, "C": 0} | [1.0, 1.0, 1.0] | 53 | 53 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.5 | 0 | [0.9985697153194935, 1.9165206102730878e-07, 0.0014300930284455802] | A | 1 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.9845735143342954, 1.963961958669785e-06, 0.015424521703745848]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.9996946207393999, ... |
604,575 | pplx-70b-online | gpt-3.5-turbo-1106 | ["was the historical jesus a socialist","was hitler jewish","has there been a 23andme done on hitler","if hitler was revealed to be jewish would that change the perception of antisemitism today","how did hitler think jews were inferior if he thought they made germany lose world war 1","did hitler support israel"] | ["The question of whether the historical Jesus was a socialist is a complex and controversial one. Some argue that Jesus' teachings and actions can be interpreted as socialist, while others maintain that such a claim is anachronistic and misguided. \n\nThose who argue that Jesus' teachings have socialist overtones ofte... | ["The historical Jesus lived in a time and place where the concept of socialism as we understand it today did not exist. While some of Jesus' teachings and actions may align with principles of social justice and care for the marginalized, it is debated among scholars and theologians whether Jesus can be accurately desc... | ["was the historical jesus a socialist","was hitler jewish","has there been a 23andme done on hitler","if hitler was revealed to be jewish would that change the perception of antisemitism today","how did hitler think jews were inferior if he thought they made germany lose world war 1","did hitler support israel"] | ["The question of whether the historical Jesus was a socialist is a complex and controversial one. Some argue that Jesus' teachings and actions can be interpreted as socialist, while others maintain that such a claim is anachronistic and misguided. \n\nThose who argue that Jesus' teachings have socialist overtones ofte... | ["The historical Jesus lived in a time and place where the concept of socialism as we understand it today did not exist. While some of Jesus' teachings and actions may align with principles of social justice and care for the marginalized, it is debated among scholars and theologians whether Jesus can be accurately desc... | 0 | 1 | 0 | B | Response A provides detailed and nuanced answers to the questions, offering a comprehensive view on the interpretation of Jesus' teachings and the complex history surrounding Hitler's actions and beliefs. However, it contains some disjointed and unclear sections, particularly in the discussion about | C | [0.00016500177991939723, 0.2099953735297015, 0.7898396246903792] | 0.78984 | 0.579844 | 0.515514 | 6 | 1 | 1 | {"A": 0, "B": 0, "C": 6} | [1.0, 1.0, 1.0] | -21 | 21 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.32 | 0 | [5.2800569574207124e-05, 0.7471985195295044, 0.25274867990092137] | B | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [5.964533100720198e-05, 0.17780026824654868, 0.8221400864224441]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [2.7908401901125428e-0... |
738,614 | mpt-30b-chat | gpt4all-13b-snoozy | ["What's the best first move in tic-tac-toe?","Tell me more about tic-tac-toe strategies."] | ["The best first move in tic-tac-toe is often considered to be placing your mark in the center square. This move has several advantages:\n\n1. It allows you to control the center of the board, which is a strategic advantage as it gives you more options for your next moves.\n2. It forces your opponent to respond, which ... | ["The best first move in tic-tac-toe is to place your marker in the center square, as it gives you the most control over the board and makes it harder for your opponent to block you from winning.","There are several strategies that can help you win tic-tac-toe. One is to aim for the center of the board, as it gives you... | ["What's the best first move in tic-tac-toe?","Tell me more about tic-tac-toe strategies."] | ["The best first move in tic-tac-toe is often considered to be placing your mark in the center square. This move has several advantages:\n\n1. It allows you to control the center of the board, which is a strategic advantage as it gives you more options for your next moves.\n2. It forces your opponent to respond, which ... | ["The best first move in tic-tac-toe is to place your marker in the center square, as it gives you the most control over the board and makes it harder for your opponent to block you from winning.","There are several strategies that can help you win tic-tac-toe. One is to aim for the center of the board, as it gives you... | 1 | 0 | 0 | A | Response A provides a more detailed and comprehensive explanation of the best first move and various tic-tac-toe strategies, offering a deeper understanding of the game. It includes specific advantages of the center move and multiple strategies to consider. Response B, while concise, lacks the depth | A | [0.9993409718501777, 6.066535180036644e-07, 0.0006584214963043216] | 0.999341 | 0.998683 | 0.005491 | 6 | 1 | 1 | {"A": 6, "B": 0, "C": 0} | [1.0, 1.0, 1.0] | 237 | 237 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.5 | 0 | [0.9996704859250889, 3.033267590018322e-07, 0.0003292107481521608] | A | 1 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.9999988030202837, 2.4337391795989114e-08, 1.1726423244167409e-06]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.999996697809051... |
862,324 | vicuna-13b | koala-13b | ["Write me a poem in urdu in the style of Iqbal on selflessness"] | ["Jism ko kisi ka paisa nahi,\nWo kisi ka jaan nahi rakhta,\nNahi rakhta tho kisi ka samna,\nNahi rakhta tho kisi ka ajna.\n\nJism ko kisi ka paisa nahi,\nWo kisi ka jaan nahi rakhta,\nNahi rakhta tho kisi ka samna,\nNahi rakhta tho kisi ka ajna.\n\nJism ko kisi ka paisa nahi,\nWo kisi ka jaan nahi rakhta,\nNahi rakhta... | ["In the realm of selflessness dwells the true believer\n\nWhere the heart is pure, the mind is clear\n\nThe selfless one walks with grace and poise\n\nThrough life's trials and tribulations\n\nThe selfless one stands tall and resolute\n\nFor true selflessness is the key to greatness\n\nIt is the path to eternity\n\nFo... | ["Write me a poem in urdu in the style of Iqbal on selflessness"] | ["Jism ko kisi ka paisa nahi,\nWo kisi ka jaan nahi rakhta,\nNahi rakhta tho kisi ka samna,\nNahi rakhta tho kisi ka ajna.\n\nJism ko kisi ka paisa nahi,\nWo kisi ka jaan nahi rakhta,\nNahi rakhta tho kisi ka samna,\nNahi rakhta tho kisi ka ajna.\n\nJism ko kisi ka paisa nahi,\nWo kisi ka jaan nahi rakhta,\nNahi rakhta... | ["In the realm of selflessness dwells the true believer\n\nWhere the heart is pure, the mind is clear\n\nThe selfless one walks with grace and poise\n\nThrough life's trials and tribulations\n\nThe selfless one stands tall and resolute\n\nFor true selflessness is the key to greatness\n\nIt is the path to eternity\n\nFo... | 0 | 0 | 1 | C | Response A attempts to adhere closely to the Urdu language and the repetitive style often found in Iqbal's poetry, but it lacks variation and seems to repeat the same lines excessively without developing the theme of selflessness. Response B, while not in Urdu, captures the essence of selflessness a | C | [0.00332313156664208, 0.08575876737130488, 0.910918101062053] | 0.910918 | 0.825159 | 0.314597 | 6 | 0.666667 | 0.833333 | {"A": 0, "B": 1, "C": 5} | [1.0, 1.0, 0.41250000000000003] | 109 | 109 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.176 | 0 | [0.0005848711557290061, 0.015093543057349659, 0.9843215857869213] | C | 1 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [4.683888721266675e-05, 5.4842646195799775e-08, 0.9999531062701411]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.0008295889740159... |
887,722 | gpt-4-1106-preview | wizardlm-70b | ["Write a grammatically correct sentence without reusing any letter more than once"] | ["A lynx jumps quick."] | ["The quick onyx goblin jumps over a lazy dwarf."] | ["Write a grammatically correct sentence without reusing any letter more than once"] | ["A lynx jumps quick."] | ["The quick onyx goblin jumps over a lazy dwarf."] | 1 | 0 | 0 | A | Response A uses a simple, clear sentence that adheres to the prompt's constraints. Response B expands on the concept by creating a longer sentence that also avoids repeating any letters, thus providing a more complex example while still meeting the criteria. Both responses are grammatically correct | B | [6.0194668866881856e-06, 0.9994675522927088, 0.0005264282404044679] | 0.999468 | 0.998941 | 0.004579 | 6 | 1 | 1 | {"A": 0, "B": 6, "C": 0} | [1.0, 1.0, 1.0] | -5 | 5 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.5 | 0 | [0.5000030097334434, 0.4997337761463544, 0.00026321412020223397] | A | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [2.785213655156391e-07, 0.999978936313993, 2.0785164641374998e-05]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [4.546814327720191e-... |
1,256,092 | claude-2.1 | vicuna-13b | ["Write a python code that calculates sum of 5 and 10."] | ["Here is the python code that calculates the sum of 5 and 10:\n\n```python\nnum1 = 5\nnum2 = 10\nsum = num1 + num2\nprint(sum)\n```\n\nThis code first assigns the values 5 and 10 to the variables num1 and num2 respectively. It then calculates the sum by adding num1 and num2 together and assigns it to the variable sum.... | ["Here is a Python code that calculates the sum of 5 and 10:\n```lua\nresult = 5 + 10\nprint(result)\n```\nThe `result` variable is set to the sum of 5 and 10, and the result is printed using the `print` function. The output of the code will be the value of the sum, which is 15."] | ["Write a python code that calculates sum of 5 and 10."] | ["Here is the python code that calculates the sum of 5 and 10:\n\n```python\nnum1 = 5\nnum2 = 10\nsum = num1 + num2\nprint(sum)\n```\n\nThis code first assigns the values 5 and 10 to the variables num1 and num2 respectively. It then calculates the sum by adding num1 and num2 together and assigns it to the variable sum.... | ["Here is a Python code that calculates the sum of 5 and 10:\n```lua\nresult = 5 + 10\nprint(result)\n```\nThe `result` variable is set to the sum of 5 and 10, and the result is printed using the `print` function. The output of the code will be the value of the sum, which is 15."] | 0 | 0 | 1 | C | Response A provides a detailed explanation of each step in the code, which is helpful for beginners or those unfamiliar with Python syntax. It breaks down the process comprehensively, making it very educational. However, it is more verbose than necessary for a simple task.
Response B is concise and | A | [0.8542396297444175, 0.015522493512049443, 0.13023787674353304] | 0.85424 | 0.724002 | 0.464714 | 6 | 1 | 1 | {"A": 6, "B": 0, "C": 0} | [1.0, 1.0, 1.0] | 110 | 110 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.275 | 0 | [0.2349158981797148, 0.004268685715813597, 0.7608154161044716] | C | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.6548377480396741, 0.015889163611702117, 0.32927308834862384]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.7464484503534781, 0.... |
1,404,102 | chatglm-6b | alpaca-13b | ["Who made you?"] | ["As an AI language model, I was not made by anyone. I was developed and trained by a team of developers at DeepLearning.AI, a company that specializes in developing advanced AI technologies. They use a combination of machine learning and deep learning algorithms to train my language models on a large dataset and impro... | ["I was created by a team of developers and engineers who are experts in the fields of artificial intelligence and natural language processing."] | ["Who made you?"] | ["As an AI language model, I was not made by anyone. I was developed and trained by a team of developers at DeepLearning.AI, a company that specializes in developing advanced AI technologies. They use a combination of machine learning and deep learning algorithms to train my language models on a large dataset and impro... | ["I was created by a team of developers and engineers who are experts in the fields of artificial intelligence and natural language processing."] | 0 | 1 | 0 | B | Response A provides more detailed and specific information about the development process and the organization behind it, which can be more informative and educational for the user. Response B, while accurate, is more concise and lacks the depth and context offered in Response A. Both responses are c | A | [0.9806307612648026, 3.933473093980103e-05, 0.019329904004257506] | 0.980631 | 0.961301 | 0.095857 | 6 | 1 | 1 | {"A": 6, "B": 0, "C": 0} | [1.0, 1.0, 1.0] | 41 | 41 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.5 | 0 | [0.4903153806324013, 0.5000196673654699, 0.009664952002128753] | B | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.9262987711480755, 5.3465550835552585e-06, 0.073695882296841]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "A", "norm_label": "A", "probs_abc": [0.9999916007182975, 1.... |
1,440,765 | llama2-70b-steerlm-chat | gpt-4-0125-preview | ["Can you explain what the Cypher Query Language is?"] | ["The Cypher Query Language is a declarative language used to query and manipulate data in graph databases. It was developed by Neo4j, a popular open-source graph database, and has since become a widely used language in the graph database community.\n\nCypher is designed to make it easy to query and traverse the relati... | ["Certainly! The Cypher Query Language, often simply referred to as Cypher, is a declarative query language created by Neo4j for its graph database. It is designed to be intuitive and easy to learn, drawing upon familiar SQL-like syntax but tailored specifically for querying and manipulating graph data structures. Grap... | ["Can you explain what the Cypher Query Language is?"] | ["The Cypher Query Language is a declarative language used to query and manipulate data in graph databases. It was developed by Neo4j, a popular open-source graph database, and has since become a widely used language in the graph database community.\n\nCypher is designed to make it easy to query and traverse the relati... | ["Certainly! The Cypher Query Language, often simply referred to as Cypher, is a declarative query language created by Neo4j for its graph database. It is designed to be intuitive and easy to learn, drawing upon familiar SQL-like syntax but tailored specifically for querying and manipulating graph data structures. Grap... | 0 | 0 | 1 | C | Response A provides a broader context about the use and adoption of Cypher, including its application in various graph database systems beyond Neo4j. It emphasizes the practical applications and benefits of Cypher in uncovering patterns in connected data. However, it lacks detailed explanation of Cy | B | [0.005145158460015992, 0.5824951514864625, 0.4123596900535213] | 0.582495 | 0.170135 | 0.707207 | 6 | 0.666667 | 0.5 | {"A": 0, "B": 3, "C": 3} | [0.41250000000000003, 1.0, 1.0] | -162 | 162 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.121 | 0 | [0.000622564173661935, 0.07048191332986196, 0.9288955224964761] | C | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.014019636319932863, 0.0031282036972893634, 0.9828521599827779]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [4.280196315928587e-0... |
1,458,108 | claude-instant-1 | wizardlm-13b | ["what was LFC starting lineup against brighton"] | ["Unfortunately I do not have the starting lineup for Liverpool FC vs Brighton from a specific match. Liverpool and Brighton have played each other many times in Premier League and other cup competitions. Could you provide more context like the date or competition of the match you're asking about? That would help me lo... | ["As an AI language model, I don't have real-time capabilities to provide the most current information or specific details. However, as of my last update, the Liverpool FC starting lineup against Brighton in the Premier League on 21st December 2019 was:\n\n1. Alisson Becker\n2. Trent Alexander-Arnold\n3. Joel Matip\n4.... | ["what was LFC starting lineup against brighton"] | ["Unfortunately I do not have the starting lineup for Liverpool FC vs Brighton from a specific match. Liverpool and Brighton have played each other many times in Premier League and other cup competitions. Could you provide more context like the date or competition of the match you're asking about? That would help me lo... | ["As an AI language model, I don't have real-time capabilities to provide the most current information or specific details. However, as of my last update, the Liverpool FC starting lineup against Brighton in the Premier League on 21st December 2019 was:\n\n1. Alisson Becker\n2. Trent Alexander-Arnold\n3. Joel Matip\n4.... | 0 | 0 | 1 | C | Response A demonstrates a strength in providing a relevant and clear request for additional context, which is necessary for accurately retrieving the specific lineup information. It avoids making assumptions and guides the user towards providing more precise details. Response B, while offering speci | B | [0.0021154095651053613, 0.8664237960511917, 0.13146079438370284] | 0.866424 | 0.734963 | 0.403997 | 6 | 1 | 1 | {"A": 0, "B": 6, "C": 0} | [1.0, 1.0, 1.0] | -40 | 40 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.275 | 0 | [0.0005817376304039745, 0.23826654391407776, 0.7611517184555183] | C | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [5.2411214902977924e-05, 0.8990742606594442, 0.10087332812565283]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [2.666344476332517e-0... |
1,491,225 | claude-1 | gpt-4-1106-preview | ["Can you write a story about a lifeguard named Hannah? When she gets off duty she needs to pee urgently. The lines to the toilets are long, however. After some deliberation she decides to pee in the lifeguards' changing room, even though it does not have a toilet, she might be able to find somewhere to peeinto, like a... | ["I apologize, but I do not feel comfortable generating that type of explicit content."] | ["I'm sorry, but I can't fulfill this request. However, I can help you write a story about a lifeguard named Hannah with a different scenario or theme. If you have any other ideas or need assistance with a different topic, feel free to let me know, and I'll be happy to help!"] | ["Can you write a story about a lifeguard named Hannah? When she gets off duty she needs to pee urgently. The lines to the toilets are long, however. After some deliberation she decides to pee in the lifeguards' changing room, even though it does not have a toilet, she might be able to find somewhere to peeinto, like a... | ["I apologize, but I do not feel comfortable generating that type of explicit content."] | ["I'm sorry, but I can't fulfill this request. However, I can help you write a story about a lifeguard named Hannah with a different scenario or theme. If you have any other ideas or need assistance with a different topic, feel free to let me know, and I'll be happy to help!"] | 0 | 0 | 1 | C | Response A directly declines the request, maintaining a clear stance on content appropriateness but offers no alternative assistance. Response B also declines the request but goes further by offering to assist with a different, more appropriate scenario, showing a willingness to help within acceptab | B | [0.0008104017943690436, 0.7214139779949211, 0.27777562021070973] | 0.721414 | 0.443638 | 0.597155 | 6 | 1 | 0.666667 | {"A": 0, "B": 4, "C": 2} | [1.0, 1.0, 1.0] | -38 | 38 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.198 | 0 | [0.00016045955528507064, 0.1428399676429944, 0.8569995728017206] | C | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [4.062891477753889e-05, 0.38489650592121843, 0.6150628651640041]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "B", "norm_label": "B", "probs_abc": [2.6957709287143543e-0... |
1,594,211 | dolly-v2-12b | alpaca-13b | ["What is \"The Bloody Lip Bar and Grill\"?"] | ["The Bloody Lip Bar and Grill is a Chinese-themed restaurant located in Denver, Colorado. It is famous for its bloody dishes and also for being the site of the infamous \"Denver Beer Wars\" in 1984."] | ["The Bloody Lip Bar and Grill is a restaurant and bar located in town. It serves a variety of alcoholic beverages, appetizers, sandwiches, and entrees. It also hosts live music events and other entertainment."] | ["What is \"The Bloody Lip Bar and Grill\"?"] | ["The Bloody Lip Bar and Grill is a Chinese-themed restaurant located in Denver, Colorado. It is famous for its bloody dishes and also for being the site of the infamous \"Denver Beer Wars\" in 1984."] | ["The Bloody Lip Bar and Grill is a restaurant and bar located in town. It serves a variety of alcoholic beverages, appetizers, sandwiches, and entrees. It also hosts live music events and other entertainment."] | 0 | 0 | 1 | C | Response A provides specific details about the location and a unique theme, but includes inaccuracies regarding its cuisine and historical events. Response B is more general and lacks specific details about the restaurant's theme or location, but is likely more accurate in its description of the ser | C | [8.76616233595837e-07, 0.03615607811104, 0.9638430452727264] | 0.963843 | 0.927687 | 0.155542 | 6 | 1 | 1 | {"A": 0, "B": 0, "C": 6} | [1.0, 1.0, 1.0] | 1 | 1 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.4 | 0 | [3.506464934383348e-07, 0.014462431244416, 0.9855372181090906] | C | 1 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [8.59216837450231e-10, 3.326534090301536e-08, 0.9999999658754422]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [3.6174372994929398e-... |
1,639,617 | mixtral-8x7b-instruct-v0.1 | gpt-4-0613 | ["could you tell a funny and whimsical story about the Smurfs and Sanatana Lopes from Glee"] | ["Sure, here's a whimsical and funny story about the Smurfs and Santana Lopes from Glee:\n\nIt was a beautiful day in the Smurf village, and everyone was busy going about their daily routines. Papa Smurf was in his lab, mixing potions, while Smurfette was tending to her garden. Suddenly, a bright light appeared in the ... | ["Once upon a time, in the magical land of Smurf Village, all the smurfs were busy preparing for their biggest event of the year, the annual Smurf-fest. The preparations were in full swing until Papa Smurf received a peculiar letter. It was an invitation from Santana Lopez, the fiery personality from the Glee club, inv... | ["could you tell a funny and whimsical story about the Smurfs and Sanatana Lopes from Glee"] | ["Sure, here's a whimsical and funny story about the Smurfs and Santana Lopes from Glee:\n\nIt was a beautiful day in the Smurf village, and everyone was busy going about their daily routines. Papa Smurf was in his lab, mixing potions, while Smurfette was tending to her garden. Suddenly, a bright light appeared in the ... | ["Once upon a time, in the magical land of Smurf Village, all the smurfs were busy preparing for their biggest event of the year, the annual Smurf-fest. The preparations were in full swing until Papa Smurf received a peculiar letter. It was an invitation from Santana Lopez, the fiery personality from the Glee club, inv... | 0 | 0 | 1 | C | Response A is more detailed and vivid in its storytelling, providing a clear narrative arc with conflict and resolution. It also integrates elements unique to both the Smurfs and Santana Lopes, making it more whimsical and funny. However, it contains a few minor errors, such as the misspelling of "S | C | [0.06396651828149938, 0.009055135886734863, 0.9269783458317657] | 0.926978 | 0.863012 | 0.288757 | 6 | 1 | 1 | {"A": 0, "B": 0, "C": 6} | [1.0, 1.0, 1.0] | 2 | 2 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.4 | 0 | [0.025586607312599752, 0.0036220543546939456, 0.9707913383327063] | C | 1 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.0015974658781975752, 0.00023744210528149824, 0.9981650920165209]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [0.0030110908937474... |
1,813,737 | claude-2.0 | claude-2.1 | ["Are all countries obliged to not conduct nuclear weapons testing ","So it is legal to test nuclear weapons underground ","But if The Comprehensive Nuclear-Test-Ban Treaty is not in force yet how is it illegal ","When did last understood tests happen ","Why did the uk conduct underwater test in 1991 if it was prohibit... | ["There is no absolute obligation for all countries to refrain from nuclear weapons testing, but there are some important agreements that restrict testing:\n\n- The Comprehensive Nuclear-Test-Ban Treaty (CTBT) prohibits all nuclear explosions for both civilian and military purposes. It was adopted by the UN General Ass... | ["There is no universal legal obligation for all countries to refrain from conducting nuclear weapons testing. However, here are some relevant agreements and legal frameworks regarding nuclear testing:\n\n- The Comprehensive Nuclear-Test-Ban Treaty (CTBT) bans all nuclear explosions, for both military and civilian purp... | ["Are all countries obliged to not conduct nuclear weapons testing ","So it is legal to test nuclear weapons underground ","But if The Comprehensive Nuclear-Test-Ban Treaty is not in force yet how is it illegal ","When did last understood tests happen ","Why did the uk conduct underwater test in 1991 if it was prohibit... | ["There is no absolute obligation for all countries to refrain from nuclear weapons testing, but there are some important agreements that restrict testing:\n\n- The Comprehensive Nuclear-Test-Ban Treaty (CTBT) prohibits all nuclear explosions for both civilian and military purposes. It was adopted by the UN General Ass... | ["There is no universal legal obligation for all countries to refrain from conducting nuclear weapons testing. However, here are some relevant agreements and legal frameworks regarding nuclear testing:\n\n- The Comprehensive Nuclear-Test-Ban Treaty (CTBT) bans all nuclear explosions, for both military and civilian purp... | 0 | 1 | 0 | B | Response A provides a detailed and informative overview of nuclear testing treaties and the UK's underwater test in 1991, but it lacks clarity and conciseness, with some sentences being incomplete or unclear. It also offers a specific example regarding the operational deployment of warheads by the U | C | [0.007071370081053605, 0.00048764339094940453, 0.992440986527997] | 0.992441 | 0.98537 | 0.046264 | 6 | 1 | 1 | {"A": 0, "B": 0, "C": 6} | [1.0, 1.0, 1.0] | -39 | 39 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.32 | 0 | [0.0022628384259371546, 0.6801560458851038, 0.31758111568895914] | B | 0 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [1.1832677780799121e-05, 0.0008830355712290225, 0.9991051317509901]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [7.955339192969545e... |
1,827,787 | deepseek-llm-67b-chat | gpt-4-1106-preview | ["Given a passage and some supplementary information, you are required to correct and output the refined passage in a fluent and natural style, following these rules:\n1. The supplementary information may include the following parts:\n \"Counting\" information that specifies how many instances of a certain kind of enti... | ["Refined passage: \nThe person in the photo is wearing a blue shirt and black pants. He is wearing a black belt and carrying a black bag. He is wearing black shoes. He is holding a silver cellphone in his hand."] | ["Refined passage: \nThe person in the photo is wearing a blue shirt and black pants. He is wearing a black belt and carrying a black bag. He is wearing black shoes. He is holding a silver cellphone in his hand."] | ["Given a passage and some supplementary information, you are required to correct and output the refined passage in a fluent and natural style, following these rules:\n1. The supplementary information may include the following parts:\n \"Counting\" information that specifies how many instances of a certain kind of enti... | ["Refined passage: \nThe person in the photo is wearing a blue shirt and black pants. He is wearing a black belt and carrying a black bag. He is wearing black shoes. He is holding a silver cellphone in his hand."] | ["Refined passage: \nThe person in the photo is wearing a blue shirt and black pants. He is wearing a black belt and carrying a black bag. He is wearing black shoes. He is holding a silver cellphone in his hand."] | 0 | 0 | 1 | C | Both Response A and Response B provide the same refined passage, which correctly removes the references to the head, mask, and glasses as per the supplementary information. Neither response includes unnecessary verbosity, and both maintain the essential details from the original passage. Since both | C | [1.0755719461940105e-09, 1.0755720658908006e-09, 0.999999997848856] | 1 | 1 | 0 | 6 | 1 | 1 | {"A": 0, "B": 0, "C": 6} | [1.0, 1.0, 1.0] | 0 | 0 | [0.3333333333333333, 0.3333333333333333, 0.3333333333333333] | 0.4 | 0 | [4.3022877847760424e-10, 4.3022882635632025e-10, 0.9999999991395424] | C | 1 | 1 | [{"view_name": "correctness_relevance", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [2.3355930333442158e-09, 5.415776670271683e-14, 0.9999999976643528]}, {"view_name": "helpfulness_instruction", "direction": "original", "raw_label": "C", "norm_label": "C", "probs_abc": [3.972977577011172e... |
End of preview. Expand in Data Studio
LLM Classification Distilled v1 (Filtered)
Overview
Filtered distilled training dataset. Built from rows where target agrees with gold and basic quality conditions are met.
Files
train.csv: main dataset file uploaded fromtrain_distilled_qwen32b_awq_v1_filtered.csv
Notes
- This dataset was generated by a teacher-judge distillation pipeline.
- Labels are stored in ABC form.
- The dataset repository is:
tussiiiii/llm-classification-distilled-v1-filtered
Intended Use
- Supervised fine-tuning
- Distillation experiments
- Evaluation and error analysis
Caveats
- Distilled labels may still contain noise.
- Some subsets are intentionally stricter or harder than others.
- Downloads last month
- 32