Scheduled Commit

data/retrieval_individual-af410ee3-7691-4d91-abd0-61898a8363dc.jsonl

@@ -37,3 +37,7 @@
 {"tstamp": 1739762643.8395, "task_type": "retrieval", "type": "chat", "model": "voyage-multilingual-2", "gen_params": {}, "start": 1739762643.428, "finish": 1739762643.8395, "ip": "", "conv_id": "58469564501b4bb1802182554c575a59", "model_name": "voyage-multilingual-2", "prompt": "I am looking for a paper to help me deal with data constraints as I scale up my language model.", "output": [["I am looking for a paper to help me deal with data constraints as I scale up my language model.", "Title: Scaling Data-Constrained Language Models\n\nAbstract: The current trend of scaling language models involves increasing both parameter count and training dataset size. Extrapolating this trend suggests that training dataset size may soon be limited by the amount of text data available on the internet. Motivated by this limit, we investigate scaling language models in data-constrained regimes. Specifically, we run a large set of experiments varying the extent of data repetition and compute budget, ranging up to 900 billion training tokens and 9 billion parameter models. We find that with constrained data for a fixed compute budget, training with up to 4 epochs of repeated data yields negligible changes to loss compared to having unique data. However, with more repetition, the value of adding compute eventually decays to zero. We propose and empirically validate a scaling law for compute optimality that accounts for the decreasing value of repeated tokens and excess parameters. Finally, we experiment with approaches mitigating data scarcity, including augmenting the training dataset with code data or removing commonly used filters. Models and datasets from our 400 training runs are freely available at https://github.com/huggingface/datablations."]], "corpus": "arxiv"}
 {"tstamp": 1739763545.9673, "task_type": "retrieval", "type": "chat", "model": "sentence-transformers/all-MiniLM-L6-v2", "gen_params": {}, "start": 1739763543.821, "finish": 1739763545.9673, "ip": "", "conv_id": "9b5a452e30964d2491c97f1ecce4d9ef", "model_name": "sentence-transformers/all-MiniLM-L6-v2", "prompt": "avatar07", "output": [["avatar07", "Avatar\n\nAvatar (, ; ) is a concept within Hinduism that in Sanskrit literally means . It signifies the material appearance or incarnation of a powerful deity, or spirit on Earth. The relative verb to \"alight, to make one's appearance\" is sometimes used to refer to any guru or revered human being."]], "corpus": "wikipedia"}
 {"tstamp": 1739763545.9673, "task_type": "retrieval", "type": "chat", "model": "BAAI/bge-large-en-v1.5", "gen_params": {}, "start": 1739763543.821, "finish": 1739763545.9673, "ip": "", "conv_id": "802f8d6d17d14c10b0c121f6946405f9", "model_name": "BAAI/bge-large-en-v1.5", "prompt": "avatar07", "output": [["avatar07", "Aang\n\nAvatar Aang (), or simply Aang, is the titular protagonist of Nickelodeon's animated television series Avatar: The Last Airbender (created by Michael Dante DiMartino and Bryan Konietzko), voiced by Zach Tyler Eisen. Aang was the last surviving Airbender, a monk of the Air Nomads' Southern Air Temple, and the youngest ever airbending master (for his time).\nHe is an incarnation of the \"Avatar\", the spirit of light and peace manifested in human form. As the Avatar, Aang controls all four elements (water, earth, fire, and air) and is tasked with bringing balance and keeping the Four Nations at peace. At chronologically 112 years old (biologically 12), Aang is the series' reluctant hero, spending a century in suspended animation in an iceberg before being discovered and joining new friends Katara and Sokka on a quest to master the elements and save their world from the imperialist Fire Nation."]], "corpus": "wikipedia"}
+{"tstamp": 1739763635.1518, "task_type": "retrieval", "type": "chat", "model": "voyage-multilingual-2", "gen_params": {}, "start": 1739763634.812, "finish": 1739763635.1518, "ip": "", "conv_id": "dbceb7e5436b4701b8cf905e7f100897", "model_name": "voyage-multilingual-2", "prompt": "What's the purpose of `if __name__ == '__main__'`?", "output": [["What's the purpose of `if __name__ == '__main__'`?", "Stackoverflow Stackexchange\n\nQ: What does `if name == \"__main__\"` mean in Python? \nPossible Duplicate:\nWhat does <if name==“main”:> do? \n\nI have wrote scripts in Python for quite a while now and I study more of Python as I need it. When reading other people's code I meet if name == \"__main__\": construct quite often. \nWhat is it good for?\n\nA: Basically, \nThere's a distinction between the \"main\" script file and external files which were imported or referenced in another way. If the script is the \"main\" script then the special variable __name__ will equal \"__main__\".\nYou can use this to protect the \"execution\" code from the classes and variables the script has. This enables you to import a script and use classes and variables without actually running that script's main code if it has any.\nSee also: What does if name == “main”: do?\n"]], "corpus": "stackexchange"}
+{"tstamp": 1739763635.1518, "task_type": "retrieval", "type": "chat", "model": "GritLM/GritLM-7B", "gen_params": {}, "start": 1739763634.812, "finish": 1739763635.1518, "ip": "", "conv_id": "7c58bcde40b04b0d95dda4d3a581a72c", "model_name": "GritLM/GritLM-7B", "prompt": "What's the purpose of `if __name__ == '__main__'`?", "output": [["What's the purpose of `if __name__ == '__main__'`?", "Stackoverflow Stackexchange\n\nQ: What does `if name == \"__main__\"` mean in Python? \nPossible Duplicate:\nWhat does <if name==“main”:> do? \n\nI have wrote scripts in Python for quite a while now and I study more of Python as I need it. When reading other people's code I meet if name == \"__main__\": construct quite often. \nWhat is it good for?\n\nA: Basically, \nThere's a distinction between the \"main\" script file and external files which were imported or referenced in another way. If the script is the \"main\" script then the special variable __name__ will equal \"__main__\".\nYou can use this to protect the \"execution\" code from the classes and variables the script has. This enables you to import a script and use classes and variables without actually running that script's main code if it has any.\nSee also: What does if name == “main”: do?\n"]], "corpus": "stackexchange"}
+{"tstamp": 1739763715.9699, "task_type": "retrieval", "type": "chat", "model": "GritLM/GritLM-7B", "gen_params": {}, "start": 1739763715.6716, "finish": 1739763715.9699, "ip": "", "conv_id": "e7bf48d2e837478fa7a33607fd470d06", "model_name": "GritLM/GritLM-7B", "prompt": "Which test was devised to determine whether robots can think?", "output": [["Which test was devised to determine whether robots can think?", "Turing test\n\nThe test was introduced by Turing in his 1950 paper \"Computing Machinery and Intelligence\" while working at the University of Manchester. It opens with the words: \"I propose to consider the question, 'Can machines think? Because \"thinking\" is difficult to define, Turing chooses to \"replace the question by another, which is closely related to it and is expressed in relatively unambiguous words.\" Turing describes the new form of the problem in terms of a three-person game called the \"imitation game\", in which an interrogator asks questions of a man and a woman in another room in order to determine the correct sex of the two players. Turing's new question is: \"Are there imaginable digital computers which would do well in the imitation game?\" This question, Turing believed, was one that could actually be answered. In the remainder of the paper, he argued against all the major objections to the proposition that \"machines can think\".\nSince Turing introduced his test, it has been both highly influential and widely criticized, and has become an important concept in the philosophy of artificial intelligence. Philosopher John Searle would comment on the Turing test in his Chinese room argument, a thought experiment that stipulates that a machine cannot have a \"mind\", \"understanding\", or \"consciousness\", regardless of how intelligently or human-like the program may make the computer behave. Searle criticizes Turing's test and claims it is insufficient to detect the presence of consciousness.\nHistory"]], "corpus": "wikipedia"}
+{"tstamp": 1739763715.9699, "task_type": "retrieval", "type": "chat", "model": "BAAI/bge-large-en-v1.5", "gen_params": {}, "start": 1739763715.6716, "finish": 1739763715.9699, "ip": "", "conv_id": "d937354ac0e045e5a30cb05c682b10ff", "model_name": "BAAI/bge-large-en-v1.5", "prompt": "Which test was devised to determine whether robots can think?", "output": [["Which test was devised to determine whether robots can think?", "Artificial general intelligence\n\nMany interdisciplinary approaches (e.g. cognitive science, computational intelligence, and decision making) consider additional traits such as imagination (the ability to form novel mental images and concepts) and autonomy.\nComputer-based systems that exhibit many of these capabilities exist (e.g. see computational creativity, automated reasoning, decision support system, robot, evolutionary computation, intelligent agent). However, no consensus holds that modern AI systems possess them to an adequate degree.\nPhysical traits\nOther capabilities are considered desirable in intelligent systems, as they may affect intelligence or aid in its expression. These include:\nthe ability to sense (e.g. see, hear, etc.), and\nthe ability to act (e.g. move and manipulate objects, change location to explore, etc.)\nThis includes the ability to detect and respond to hazard.\nTests for human-level AGI\nSeveral tests meant to confirm human-level AGI have been considered, including:\nThe Turing Test (Turing)\nA machine and a human both converse unseen with a second human, who must evaluate which of the two is the machine, which passes the test if it can fool the evaluator a significant fraction of the time. Note: Turing does not prescribe what should qualify as intelligence, only that knowing that it is a machine should disqualify it. The AI Eugene Goostman, imitating a 13-year-old boy, achieved Turing's estimate of convincing 33% of judges that it was human in 2014.\nThe Robot College Student Test (Goertzel)"]], "corpus": "wikipedia"}

data/retrieval_side_by_side-af410ee3-7691-4d91-abd0-61898a8363dc.jsonl

@@ -1 +1,2 @@
 {"tstamp": 1739679343.4907, "task_type": "retrieval", "type": "tievote", "models": ["GritLM/GritLM-7B", "BAAI/bge-large-en-v1.5"], "ip": "", "0_conv_id": "b38633b70a9547749dfe7776d506c37e", "0_model_name": "GritLM/GritLM-7B", "0_prompt": "england", "0_output": [["england", "England\n\nEngland is a country that is part of the United Kingdom. It is located on the island of Great Britain, of which it covers approximately 62%, and over 100 smaller adjacent islands. It has land borders with Scotland to the north and Wales to the west, and is otherwise surrounded by the North Sea to the east, the English Channel to the south, the Celtic Sea to the south-west, and the Irish Sea to the west. Continental Europe lies to the south-east, and Ireland to the west. At the 2021 census, the population was 56,490,048. London is both the largest city and the capital."]], "0_corpus": "wikipedia", "1_conv_id": "ae81188a41804203943cc5ae49e107ee", "1_model_name": "BAAI/bge-large-en-v1.5", "1_prompt": "england", "1_output": [["england", "England\n\nEngland is a country that is part of the United Kingdom. It is located on the island of Great Britain, of which it covers approximately 62%, and over 100 smaller adjacent islands. It has land borders with Scotland to the north and Wales to the west, and is otherwise surrounded by the North Sea to the east, the English Channel to the south, the Celtic Sea to the south-west, and the Irish Sea to the west. Continental Europe lies to the south-east, and Ireland to the west. At the 2021 census, the population was 56,490,048. London is both the largest city and the capital."]], "1_corpus": "wikipedia"}
+{"tstamp": 1739763768.6206, "task_type": "retrieval", "type": "leftvote", "models": ["GritLM/GritLM-7B", "BAAI/bge-large-en-v1.5"], "ip": "", "0_conv_id": "e7bf48d2e837478fa7a33607fd470d06", "0_model_name": "GritLM/GritLM-7B", "0_prompt": "Which test was devised to determine whether robots can think?", "0_output": [["Which test was devised to determine whether robots can think?", "Turing test\n\nThe test was introduced by Turing in his 1950 paper \"Computing Machinery and Intelligence\" while working at the University of Manchester. It opens with the words: \"I propose to consider the question, 'Can machines think? Because \"thinking\" is difficult to define, Turing chooses to \"replace the question by another, which is closely related to it and is expressed in relatively unambiguous words.\" Turing describes the new form of the problem in terms of a three-person game called the \"imitation game\", in which an interrogator asks questions of a man and a woman in another room in order to determine the correct sex of the two players. Turing's new question is: \"Are there imaginable digital computers which would do well in the imitation game?\" This question, Turing believed, was one that could actually be answered. In the remainder of the paper, he argued against all the major objections to the proposition that \"machines can think\".\nSince Turing introduced his test, it has been both highly influential and widely criticized, and has become an important concept in the philosophy of artificial intelligence. Philosopher John Searle would comment on the Turing test in his Chinese room argument, a thought experiment that stipulates that a machine cannot have a \"mind\", \"understanding\", or \"consciousness\", regardless of how intelligently or human-like the program may make the computer behave. Searle criticizes Turing's test and claims it is insufficient to detect the presence of consciousness.\nHistory"]], "0_corpus": "wikipedia", "1_conv_id": "d937354ac0e045e5a30cb05c682b10ff", "1_model_name": "BAAI/bge-large-en-v1.5", "1_prompt": "Which test was devised to determine whether robots can think?", "1_output": [["Which test was devised to determine whether robots can think?", "Artificial general intelligence\n\nMany interdisciplinary approaches (e.g. cognitive science, computational intelligence, and decision making) consider additional traits such as imagination (the ability to form novel mental images and concepts) and autonomy.\nComputer-based systems that exhibit many of these capabilities exist (e.g. see computational creativity, automated reasoning, decision support system, robot, evolutionary computation, intelligent agent). However, no consensus holds that modern AI systems possess them to an adequate degree.\nPhysical traits\nOther capabilities are considered desirable in intelligent systems, as they may affect intelligence or aid in its expression. These include:\nthe ability to sense (e.g. see, hear, etc.), and\nthe ability to act (e.g. move and manipulate objects, change location to explore, etc.)\nThis includes the ability to detect and respond to hazard.\nTests for human-level AGI\nSeveral tests meant to confirm human-level AGI have been considered, including:\nThe Turing Test (Turing)\nA machine and a human both converse unseen with a second human, who must evaluate which of the two is the machine, which passes the test if it can fool the evaluator a significant fraction of the time. Note: Turing does not prescribe what should qualify as intelligence, only that knowing that it is a machine should disqualify it. The AI Eugene Goostman, imitating a 13-year-old boy, achieved Turing's estimate of convincing 33% of judges that it was human in 2014.\nThe Robot College Student Test (Goertzel)"]], "1_corpus": "wikipedia"}