Scheduled Commit

data/clustering_individual-928d45d9-19ab-4d05-9aae-4b6574886d0d.jsonl

@@ -14,3 +14,5 @@
 {"tstamp": 1722545046.079, "task_type": "clustering", "type": "chat", "model": "nomic-ai/nomic-embed-text-v1.5", "gen_params": {}, "start": 1722545045.9992, "finish": 1722545046.079, "ip": "", "conv_id": "a7604d2a2df84e6d82bb2aef31fb739e", "model_name": "nomic-ai/nomic-embed-text-v1.5", "prompt": ["bifocal", "convex", "toric", "prismatic", "star", "planet", "comet", "asteroid", "nebula", "galaxy", "republic", "theocracy", "monarchy", "dictatorship", "yellow", "green", "blue", "orange"], "ncluster": 4, "output": "", "ndim": "3D (press for 2D)", "dim_method": "PCA", "clustering_method": "KMeans"}
 {"tstamp": 1722545572.1229, "task_type": "clustering", "type": "chat", "model": "text-embedding-004", "gen_params": {}, "start": 1722545571.8556, "finish": 1722545572.1229, "ip": "", "conv_id": "25106531000143478edcb840cdc095de", "model_name": "text-embedding-004", "prompt": ["democracy", "dictatorship", "monarchy", "republic", "theocracy", "oligarchy", "pencil", "marker", "crayon"], "ncluster": 2, "output": "", "ndim": "3D (press for 2D)", "dim_method": "PCA", "clustering_method": "KMeans"}
 {"tstamp": 1722545572.1229, "task_type": "clustering", "type": "chat", "model": "jinaai/jina-embeddings-v2-base-en", "gen_params": {}, "start": 1722545571.8556, "finish": 1722545572.1229, "ip": "", "conv_id": "91d21cc6d69341cca43d002613028371", "model_name": "jinaai/jina-embeddings-v2-base-en", "prompt": ["democracy", "dictatorship", "monarchy", "republic", "theocracy", "oligarchy", "pencil", "marker", "crayon"], "ncluster": 2, "output": "", "ndim": "3D (press for 2D)", "dim_method": "PCA", "clustering_method": "KMeans"}
+{"tstamp": 1722552941.5332, "task_type": "clustering", "type": "chat", "model": "Salesforce/SFR-Embedding-2_R", "gen_params": {}, "start": 1722552941.4861, "finish": 1722552941.5332, "ip": "", "conv_id": "296513c736b24520986d66ddc71aec59", "model_name": "Salesforce/SFR-Embedding-2_R", "prompt": ["compote"], "ncluster": 3, "output": "", "ndim": "3D (press for 2D)", "dim_method": "PCA", "clustering_method": "KMeans"}
+{"tstamp": 1722552941.5332, "task_type": "clustering", "type": "chat", "model": "sentence-transformers/all-MiniLM-L6-v2", "gen_params": {}, "start": 1722552941.4861, "finish": 1722552941.5332, "ip": "", "conv_id": "f40d1c38477b4068b4abf8f7de1b1eda", "model_name": "sentence-transformers/all-MiniLM-L6-v2", "prompt": ["compote"], "ncluster": 3, "output": "", "ndim": "3D (press for 2D)", "dim_method": "PCA", "clustering_method": "KMeans"}

data/retrieval_battle-928d45d9-19ab-4d05-9aae-4b6574886d0d.jsonl

@@ -50,3 +50,4 @@
 {"tstamp": 1722548298.1582, "task_type": "retrieval", "type": "tievote", "models": ["", ""], "ip": "", "0_conv_id": "c11837fa417b4097b7807191babfcdda", "0_model_name": "embed-english-v3.0", "0_prompt": "who died in Harry Potter 5?", "0_output": [["who died in Harry Potter 5?", "Harry Potter\n\nIn the fifth book, Harry Potter and the Order of the Phoenix, the Ministry of Magic refuses to believe that Voldemort has returned. Dumbledore re-activates the Order of the Phoenix, a secret society to counter Voldemort; meanwhile, the Ministry tightens control of Hogwarts by appointing Dolores Umbridge as High Inquisitor of Hogwarts, and she gradually increases her powers. When Umbridge bans practical teaching of Defence Against the Dark Arts, Harry, Ron and Hermione form \"Dumbledore's Army\", a secret group to continue the teachings. Harry has recurring dreams of a dark corridor in the Ministry of Magic, eventually dreaming that Sirius is being tortured there. He races to the Ministry with his friends, but it is a trap, planted in his head by Voldemort. The group are attacked by Death Eaters and saved by the Order of the Phoenix, but Sirius is killed in the battle. A duel between Dumbledore and Voldemort convinces the ministry of Voldemort's return. A prophecy concerning Harry and Voldemort is revealed: one must die at the hands of the other.\nIn the sixth book, Harry Potter and the Half-Blood Prince, Snape teaches Defence Against the Dark Arts while Horace Slughorn becomes the Potions master. Harry finds an old textbook with annotations by the Half-Blood Prince, due to which he achieves success in Potions class. Harry also takes lessons with Dumbledore, viewing memories about the early life of Voldemort in a device called a Pensieve. Harry learns from a drunken Slughorn that he used to teach Tom Riddle, and that Voldemort divided his soul into pieces, creating a series of Horcruxes. Harry and Dumbledore travel to a distant lake to destroy a Horcrux; they succeed, but Dumbledore weakens. On their return, they find Draco Malfoy and Death Eaters attacking the school. The book ends with the killing of Dumbledore by Professor Snape, the titular Half-Blood Prince."]], "0_corpus": "wikipedia", "1_conv_id": "cf5510c850da4f4ba0d7a7e7553e20a4", "1_model_name": "GritLM/GritLM-7B", "1_prompt": "who died in Harry Potter 5?", "1_output": [["who died in Harry Potter 5?", "Harry Potter\n\nIn the fifth book, Harry Potter and the Order of the Phoenix, the Ministry of Magic refuses to believe that Voldemort has returned. Dumbledore re-activates the Order of the Phoenix, a secret society to counter Voldemort; meanwhile, the Ministry tightens control of Hogwarts by appointing Dolores Umbridge as High Inquisitor of Hogwarts, and she gradually increases her powers. When Umbridge bans practical teaching of Defence Against the Dark Arts, Harry, Ron and Hermione form \"Dumbledore's Army\", a secret group to continue the teachings. Harry has recurring dreams of a dark corridor in the Ministry of Magic, eventually dreaming that Sirius is being tortured there. He races to the Ministry with his friends, but it is a trap, planted in his head by Voldemort. The group are attacked by Death Eaters and saved by the Order of the Phoenix, but Sirius is killed in the battle. A duel between Dumbledore and Voldemort convinces the ministry of Voldemort's return. A prophecy concerning Harry and Voldemort is revealed: one must die at the hands of the other.\nIn the sixth book, Harry Potter and the Half-Blood Prince, Snape teaches Defence Against the Dark Arts while Horace Slughorn becomes the Potions master. Harry finds an old textbook with annotations by the Half-Blood Prince, due to which he achieves success in Potions class. Harry also takes lessons with Dumbledore, viewing memories about the early life of Voldemort in a device called a Pensieve. Harry learns from a drunken Slughorn that he used to teach Tom Riddle, and that Voldemort divided his soul into pieces, creating a series of Horcruxes. Harry and Dumbledore travel to a distant lake to destroy a Horcrux; they succeed, but Dumbledore weakens. On their return, they find Draco Malfoy and Death Eaters attacking the school. The book ends with the killing of Dumbledore by Professor Snape, the titular Half-Blood Prince."]], "1_corpus": "wikipedia"}
 {"tstamp": 1722548346.059, "task_type": "retrieval", "type": "bothbadvote", "models": ["", ""], "ip": "", "0_conv_id": "caa409560ccd4ddd9ec01546c551c60b", "0_model_name": "embed-english-v3.0", "0_prompt": "KNRM", "0_output": [["KNRM", "Title: KaRMMa 2.0 -- Kappa Reconstruction for Mass Mapping\n\nAbstract: We present KaRMMa 2.0, an updated version of the mass map reconstruction code introduced in Fiedorowicz et al. (2022). KaRMMa is a full-sky Bayesian algorithm for reconstructing weak lensing mass maps from shear data. It forward-models the convergence field as a realization of a lognormal field. The corresponding shear map is calculated using the standard Kaiser-Squires transformation, and compared to observations at the field level. The posterior distribution of maps given the shear data is sampled using Hamiltonian Monte Carlo chains. Our work improves on the original algorithm by making it numerically efficient, enabling full-sky reconstructions at $\\approx$ 7 arcmin resolution with modest computational resources. These gains are made with no loss in accuracy or precision relative to KaRMMa 1.0. We compare the KaRMMa 2.0 posteriors against simulations across a variety of summary statistics (one-point function, two-point functions, and peak/void counts) to demonstrate our updated algorithm provides an accurate reconstruction of the convergence field at mildly non-linear scales. Unsurprisingly, the lognormal model fails as we approach non-linear scales ($\\ell \\gtrsim 200$), which in turn biases the map posteriors. These biases are at the 2% level in the recovered power spectrum, and at the 5% to 15% level for other statistics, depending on the resolution."]], "0_corpus": "arxiv", "1_conv_id": "4e87bfcb1843400c877e56d3505cca67", "1_model_name": "text-embedding-3-large", "1_prompt": "KNRM", "1_output": [["KNRM", "Title: The Next-Generation Radio Neutrino Observatory -- Multi-Messenger Neutrino Astrophysics at Extreme Energies\n\nAbstract: RNO is the mid-scale discovery instrument designed to make the first observation of neutrinos from the cosmos at extreme energies, with sensitivity well beyond current instrument capabilities. This new observatory will be the largest ground-based neutrino telescope to date, enabling the measurement of neutrinos above $10^{16}$ eV, determining the nature of the astrophysical neutrino flux that has been measured by IceCube at higher energies, similarly extending the reach of multi-messenger astrophysics to the highest energies, and enabling investigations of fundamental physics at energies unreachable by particle accelerators on Earth."]], "1_corpus": "arxiv"}
 {"tstamp": 1722548372.5618, "task_type": "retrieval", "type": "leftvote", "models": ["", ""], "ip": "", "0_conv_id": "6ce6497779424389b490d3a42385bdb8", "0_model_name": "text-embedding-3-large", "0_prompt": "papers on text embeddings and neural retriever", "0_output": [["papers on text embeddings and neural retriever", "Title: Text Embeddings for Retrieval From a Large Knowledge Base\n\nAbstract: Text embedding representing natural language documents in a semantic vector space can be used for document retrieval using nearest neighbor lookup. In order to study the feasibility of neural models specialized for retrieval in a semantically meaningful way, we suggest the use of the Stanford Question Answering Dataset (SQuAD) in an open-domain question answering context, where the first task is to find paragraphs useful for answering a given question. First, we compare the quality of various text-embedding methods on the performance of retrieval and give an extensive empirical comparison on the performance of various non-augmented base embedding with, and without IDF weighting. Our main results are that by training deep residual neural models, specifically for retrieval purposes, can yield significant gains when it is used to augment existing embeddings. We also establish that deeper models are superior to this task. The best base baseline embeddings augmented by our learned neural approach improves the top-1 paragraph recall of the system by 14%."]], "0_corpus": "arxiv", "1_conv_id": "25c491ce7e504e53904f57cd8406474a", "1_model_name": "sentence-transformers/all-MiniLM-L6-v2", "1_prompt": "papers on text embeddings and neural retriever", "1_output": [["papers on text embeddings and neural retriever", "Title: Rate increase in chemical reaction and its variance under turbulent equilibrium\n\nAbstract: As contrast to the Maxwellian distribution, equilibrium distribution of the second kind or turbulent equilibrium is shown to exist under tertiary molecular chaos hypothesis to replace the classical binary chaos by Boltzmann. It is expressed as bimodal Maxwellians each mode differing by plus/minus root-mean-squares of fluctuations in macroscopic variables. Chemical reaction rates calculated using the turbulent-equilibrium are obtained in a closed form, accounting for discrepancy between experiments and classical theory based on Arrhenius' law that underestimates the burning rate considerably. The key issue is the correct estimation of the high-energy tail of the distribution function that is minor in population, yet playing a major role for reactions with high activation energy. It is extremely sensitive to turbulence level in the temperature, causing slightly subcritical molecules to clear the potential barrier to cause reactions just like quantum particles undergoing tunnelling effect owing to the uncertainty principle. Variance of the fluctuating turbulent chemical reaction rate is also calculated, verifying that relative variance based on the turbulent equilibrium is low, whereas its classical counterpart (Arrhenius) is pathologically high. A closed set of equations governing reactive turbulent gases is presented on this sound basis."]], "1_corpus": "arxiv"}
+{"tstamp": 1722552898.5551, "task_type": "retrieval", "type": "leftvote", "models": ["", ""], "ip": "", "0_conv_id": "c34f244ea3c747ce9368287196229920", "0_model_name": "GritLM/GritLM-7B", "0_prompt": "Who is python creator?", "0_output": [["Who is python creator?", "Guido van Rossum\n\nGuido van Rossum (; born 31 January 1956) is a Dutch programmer. He is the creator of the Python programming language, for which he was the \"benevolent dictator for life\" (BDFL) until he stepped down from the position on 12 July 2018. He remained a member of the Python Steering Council through 2019, and withdrew from nominations for the 2020 election.\nLife and education\nVan Rossum was born and raised in the Netherlands, where he received a master's degree in mathematics and computer science from the University of Amsterdam in 1982. He received a bronze medal in 1974 in the International Mathematical Olympiad. He has a brother, Just van Rossum, who is a type designer and programmer who designed the typeface used in the \"Python Powered\" logo.\nVan Rossum lives in Belmont, California, with his wife, Kim Knapp, and their son.\nWork"]], "0_corpus": "wikipedia", "1_conv_id": "f5f9c44c206f41ec83dd6a6a9536219d", "1_model_name": "mixedbread-ai/mxbai-embed-large-v1", "1_prompt": "Who is python creator?", "1_output": [["Who is python creator?", "Python (programming language)\n\nHistory Python was invented in the late 1980s by Guido van Rossum at Centrum Wiskunde & Informatica (CWI) in the Netherlands as a successor to the ABC programming language, which was inspired by SETL, capable of exception handling and interfacing with the Amoeba operating system. Its implementation began in December 1989. Van Rossum shouldered sole responsibility for the project, as the lead developer, until 12 July 2018, when he announced his \"permanent vacation\" from his responsibilities as Python's \"benevolent dictator for life\" (BDFL), a title the Python community bestowed upon him to reflect his long-term commitment as the project's chief decision-maker (he's since come out of retirement and is self-titled \"BDFL-emeritus\"). In January 2019, active Python core developers elected a five-member Steering Council to lead the project.\nPython 2.0 was released on 16 October 2000, with many major new features such as list comprehensions, cycle-detecting garbage collection, reference counting, and Unicode support. Python 3.0, released on 3 December 2008, with many of its major features backported to Python 2.6.x and 2.7.x. Releases of Python 3 include the 2to3 utility, which automates the translation of Python 2 code to Python 3.\nPython 2.7's end-of-life was initially set for 2015, then postponed to 2020 out of concern that a large body of existing code could not easily be forward-ported to Python 3. No further security patches or other improvements will be released for it. Currently only 3.8 and later are supported (2023 security issues were fixed in e.g. 3.7.17, the final 3.7.x release). While Python 2.7 and older is officially unsupported, a different unofficial Python implementation, PyPy, continues to support Python 2, i.e. \"2.7.18+\" (plus 3.9 and 3.10), with the plus meaning (at least some) \"backported security updates\"."]], "1_corpus": "wikipedia"}

data/retrieval_individual-928d45d9-19ab-4d05-9aae-4b6574886d0d.jsonl

@@ -185,3 +185,5 @@
 {"tstamp": 1722548538.3092, "task_type": "retrieval", "type": "chat", "model": "text-embedding-004", "gen_params": {}, "start": 1722548538.03, "finish": 1722548538.3092, "ip": "", "conv_id": "a411458c4fde4ce29f79f83601d54568", "model_name": "text-embedding-004", "prompt": "olympic swim gold", "output": [["olympic swim gold", "Swimming at the 2012 Summer Olympics\n\nThe United States claimed a total of 31 medals (16 golds, 9 silver, and 6 bronze) in the leaderboard to maintain its supremacy as the most successful nation in swimming. Brought by an unprecedented sporting domination, Michael Phelps emerged as the most decorated Olympian of all time after winning six more medals at these Games to bring his total after the 2012 games to 22 (18 golds, 2 silver, and 2 bronze). Battling against the Americans for an overall medal count, China mounted to an unexpected second-place effort on the leaderboard with a tally of 10 medals (five golds, three silver, and bronze) after striking a superb double from Sun Yang in long-distance freestyle (both 400 and 1500 m) and Ye Shiwen in the individual medley (both 200 and 400 m). Meanwhile, France ended on a spectacular fashion in third spot with a total of seven medals (four golds, two silver, and one bronze), followed by the Netherlands with four, including two golds from Ranomi Kromowidjojo in sprint freestyle (both 50 and 100 m), and South Africa with three.\nFor the first time since 1992, Australia delivered an underwhelming performance with only a single triumph in the freestyle relay, but managed to bring home a total of ten medals. After not winning a gold in swimming since 2000, Japan produced the most medals in the post-war era to build a tally of eleven (three silver and eight bronze)."]], "corpus": "wikipedia"}
 {"tstamp": 1722548626.2977, "task_type": "retrieval", "type": "chat", "model": "jinaai/jina-embeddings-v2-base-en", "gen_params": {}, "start": 1722548624.0975, "finish": 1722548626.2977, "ip": "", "conv_id": "a53118d24ff6408f9d4659c4eaaa523a", "model_name": "jinaai/jina-embeddings-v2-base-en", "prompt": "can you zip a zip file?", "output": [["can you zip a zip file?", "Stackoverflow Stackexchange\n\nQ: How to add 'select one...' to options_from_collection_for_select Below is my select-form that works properly.\nWhen the user loads the page it shall show an initial ‘select one...’ with value null or ‘’.\nI tried to add it to the Object but wasn’t able to and would be glad to get help!\nThanks a lot!\n\nIn my view:\n= select_tag 'incident[fault_id]' , options_from_collection_for_select( Fault.all, :id, :label)\n\nI use Rails 3.2 and HAML\n\nUpdate:\nBy chance I found something really sweet:\ninclude_blank: 'select one...'\n\nor completely\n= f.collection_select :fault_id, Fault.order(:label), :id, :label, include_blank: 'select one...'\n\nIn case one likes that too...\nReference: http://api.rubyonrails.org/classes/ActionView/Helpers/FormOptionsHelper.html\n\nA: options_from_collection_for_select returns a string of option tags that have been compiled by iterating over the collection and assigning the result of a call to the value_method as the option value and the text_method as the option text. \nSo just prepend it with \"select_one\" option string without value:\n = select_tag 'incident[fault_id]', content_tag(:option,'select one...',:value=>\"\")+options_from_collection_for_select( Fault.all, :id, :label)\n\n"]], "corpus": "stackexchange"}
 {"tstamp": 1722548626.2977, "task_type": "retrieval", "type": "chat", "model": "voyage-multilingual-2", "gen_params": {}, "start": 1722548624.0975, "finish": 1722548626.2977, "ip": "", "conv_id": "ade1b400306344cc91c167999d25b585", "model_name": "voyage-multilingual-2", "prompt": "can you zip a zip file?", "output": [["can you zip a zip file?", "Superuser Stackexchange\n\nQ: Can Double Zipping Cause Problems? I have a directory that I have been storing a ton of files in. Among the 10ish types of files there are also .zips of other directories.\nIn my personal experience I vaguely remember issues when trying to zip zips (ha zip zips).\nWill compressing an already compressed file cause immediate problems, or potentially problems later on when I attempt to uncompress it?\nI read Can zipping a file break it?, but it does not really address the zipping of zips. I did notice that corruption can cause zips to break. Would zipping zips potentially corrupt files?\n\nA: You can do that, you shouldn't see any problem but compressing a compressed folder, you won't gain a lot of compression. That's it...\n"]], "corpus": "stackexchange"}
+{"tstamp": 1722552871.6956, "task_type": "retrieval", "type": "chat", "model": "GritLM/GritLM-7B", "gen_params": {}, "start": 1722552869.3505, "finish": 1722552871.6956, "ip": "", "conv_id": "c34f244ea3c747ce9368287196229920", "model_name": "GritLM/GritLM-7B", "prompt": "Who is python creator?", "output": [["Who is python creator?", "Guido van Rossum\n\nGuido van Rossum (; born 31 January 1956) is a Dutch programmer. He is the creator of the Python programming language, for which he was the \"benevolent dictator for life\" (BDFL) until he stepped down from the position on 12 July 2018. He remained a member of the Python Steering Council through 2019, and withdrew from nominations for the 2020 election.\nLife and education\nVan Rossum was born and raised in the Netherlands, where he received a master's degree in mathematics and computer science from the University of Amsterdam in 1982. He received a bronze medal in 1974 in the International Mathematical Olympiad. He has a brother, Just van Rossum, who is a type designer and programmer who designed the typeface used in the \"Python Powered\" logo.\nVan Rossum lives in Belmont, California, with his wife, Kim Knapp, and their son.\nWork"]], "corpus": "wikipedia"}
+{"tstamp": 1722552871.6956, "task_type": "retrieval", "type": "chat", "model": "mixedbread-ai/mxbai-embed-large-v1", "gen_params": {}, "start": 1722552869.3505, "finish": 1722552871.6956, "ip": "", "conv_id": "f5f9c44c206f41ec83dd6a6a9536219d", "model_name": "mixedbread-ai/mxbai-embed-large-v1", "prompt": "Who is python creator?", "output": [["Who is python creator?", "Python (programming language)\n\nHistory Python was invented in the late 1980s by Guido van Rossum at Centrum Wiskunde & Informatica (CWI) in the Netherlands as a successor to the ABC programming language, which was inspired by SETL, capable of exception handling and interfacing with the Amoeba operating system. Its implementation began in December 1989. Van Rossum shouldered sole responsibility for the project, as the lead developer, until 12 July 2018, when he announced his \"permanent vacation\" from his responsibilities as Python's \"benevolent dictator for life\" (BDFL), a title the Python community bestowed upon him to reflect his long-term commitment as the project's chief decision-maker (he's since come out of retirement and is self-titled \"BDFL-emeritus\"). In January 2019, active Python core developers elected a five-member Steering Council to lead the project.\nPython 2.0 was released on 16 October 2000, with many major new features such as list comprehensions, cycle-detecting garbage collection, reference counting, and Unicode support. Python 3.0, released on 3 December 2008, with many of its major features backported to Python 2.6.x and 2.7.x. Releases of Python 3 include the 2to3 utility, which automates the translation of Python 2 code to Python 3.\nPython 2.7's end-of-life was initially set for 2015, then postponed to 2020 out of concern that a large body of existing code could not easily be forward-ported to Python 3. No further security patches or other improvements will be released for it. Currently only 3.8 and later are supported (2023 security issues were fixed in e.g. 3.7.17, the final 3.7.x release). While Python 2.7 and older is officially unsupported, a different unofficial Python implementation, PyPy, continues to support Python 2, i.e. \"2.7.18+\" (plus 3.9 and 3.10), with the plus meaning (at least some) \"backported security updates\"."]], "corpus": "wikipedia"}