paper_id
stringlengths 10
10
⌀ | yes_no
bool 2
classes | paper_index
int64 0
519
⌀ | evidence
stringlengths 0
37.7k
| question
stringlengths 4
11.7k
| answer
stringlengths 1
26k
|
|---|---|---|---|---|---|
null | false
| null |
When does baby start crawling?
|
There is a wide range in terms of when babies reach this development milestone. At 6 months old, they may start to rock back and forth on hands and knees, and by 9 months, they start to creep and crawl. However, each baby has their own timelines, and it's not uncommon to see babies skip crawling all together.
|
|
null | false
| null |
The list below represents a national accounts derived indicator for a country or territory's gross household disposable income per capita (including social transfers in kind). According to the OECD, 'household disposable income is income available to households such as wages and salaries, income from self-employment and unincorporated enterprises, income from pensions and other social benefits, and income from financial investments (less any payments of tax, social insurance contributions and interest on financial liabilities). 'Gross’ means that depreciation costs are not subtracted.' This indicator also takes account of social transfers in kind 'such as health or education provided for free or at reduced prices by governments and not-for-profit organisations.' The data shown below is published by the OECD and is presented in purchasing power parity (PPP) in order to adjust for price differences between countries.
|
Extract the difference between Gross Disposable income from Disposable Income
|
There is a unique difference between Gross Disposable Income and Household disposable income. The first being that "gross" indicates the total about amount of income derived from all sources such as wages, investments, self-employment, pensions and social benefits before extracting depreciation costs. Household disposable income includes all sources of income less social benefits, taxes, insurance, and liabilities such as rent, electricity and other ongoing financial commitments. What is remaining is defined as household disposable income which is then indexed globally using PPP (purchasing power parity) in order to adjust for price differences in different countries.
|
null | false
| null |
Sechrest joined the faculty of Pennsylvania State University (Penn State) in 1956 as an assistant professor of psychology. He left Penn State in 1958 to become an assistant professor at Northwestern University, where he was promoted to associate professor in 1964 and to full professor in 1967. In 1973, he became a professor of psychology at Florida State University, and in 1980, he was named director of the Center for Research on the Utilization of Scientific Knowledge in the Institute for Social Research at the University of Michigan. In 1984, he joined the University of Arizona as professor and chair of the Department of Psychology, serving as department chair until 1989. He served as president of the Society of Clinical Psychology (Division 12 of the American Psychological Association) in 1985. He continued to serve as an active faculty member at the University of Arizona until 2002, whereupon he became an emeritus professor there.
|
Give a short bulleted list of the academic positions held by Sechrest. Each bullet should have the name of the university and the titles that Sechrest had in that university.
|
- Pennsylvania State University : assistant professor
- Northwestern University: assistant professor, associate professor, full professor
- Florida State University: professor of psychology
- University of Michigan: director of the Center for Research on the Utilization of Scientific Knowledge in the Institute for Social Research
- University of Arizona: professor, chair of the department of Pshychology, emeritus professor
|
null | false
| null |
Who was Catelyn Stark originally betrothed to marry?
|
Catelyn Stark (nee Tully) was supposed to marry Brandon Stark, the heir to Winterfell, but Brandon was killed by the Mad King and instead, Catelyn married the next in line to the Winterfell claim, Eddard Stark.
|
|
null | false
| null |
What is a Hummum
|
Turkish bath
|
|
null | false
| 163
|
Sarcasm often emanates from incongruity BIBREF9 , which enforces the brain to reanalyze it BIBREF10 . This, in turn, affects the way eyes move through the text. Hence, distinctive eye-movement patterns may be observed in the case of successful processing of sarcasm in text in contrast to literal texts. This hypothesis forms the crux of our method for sarcasm detection and we validate this using our previously released freely available sarcasm dataset BIBREF8 enriched with gaze information.
Hence, distinctive eye-movement patterns may be observed in the case of successful processing of sarcasm in text in contrast to literal texts. This hypothesis forms the crux of our method for sarcasm detection
|
What hypothesis forms the crux of the team's method for sarcasm detection?
|
Distinctive eye-movement patterns may be observed in the case of successful processing of sarcasm in text in contrast to literal texts.
|
1708.05873
| false
| null |
Yet surprisingly little is known about the agenda-setting process for international development in global governance institutions. This is perhaps best demonstrated by the lack of information on how the different goals and targets of the MDGs were decided, which led to much criticism and concern about the global governance of development BIBREF1 . More generally, we know little about the types of development issues that different countries prioritise, or whether country-specific factors such as wealth or democracy make countries more likely to push for specific development issues to be put on the global political agenda.
The analysis of discussion of international development in annual UN General Debate statements therefore uncovers two principle development topics: economic development and sustainable development. We find that discussion of Topic 2 is not significantly impacted by country-specific factors, such as wealth, population, democracy, levels of ODA, and conflict (although there are regional effects). However, we find that the extent to which countries discuss sustainable development (Topic 7) in their annual GD statements varies considerably according to these different structural factors. The results suggest that broadly-speaking we do not observe linear trends in the relationship between these country-specific factors and discussion of Topic 7. Instead, we find that there are significant fluctuations in the relationship between factors such as wealth, democracy, etc., and the extent to which these states discuss sustainable development in their GD statements. These relationships require further analysis and exploration.
More generally, we know little about the types of development issues that different countries prioritise, or whether country-specific factors such as wealth or democracy make countries more likely to push for specific development issues to be put on the global political agenda.
We find that discussion of Topic 2 is not significantly impacted by country-specific factors, such as wealth, population, democracy, levels of ODA, and conflict (although there are regional effects).
|
What are the country-specific drivers of international development rhetoric?
|
The answers are shown as follows:
* wealth
* democracy
* population
* levels of ODA
* conflict
|
null | false
| null |
What makes the Expanse series better than other sci-fi?
|
What really sets The Expanse apart from other popular science fiction is the science. It's common for sci-fi writers to throw the laws of physics out the window or create a fake invention that bends the rules without any scientific basis to make their concept work. The Expanse has a realism that is unparalleled. If you've ever found yourself yelling at your book or TV saying "That's not even close to possible" then this is the sci-fi series for you.
|
|
null | false
| null |
Shandao Temple (formerly transliterated as Shantao Temple Station until 2003), secondary station name Huashan, is a station on the Bannan line of Taipei Metro in Zhongzheng District, Taipei, Taiwan. It is named after the Shandao Temple.
|
List down some facts about Shandao temple using given passage as a reference
|
1. Shandao Temple, also known as Huashan, is a station on the Bannan line of the Taipei Metro in the Zhongzheng District of Taipei, Taiwan. It was formerly transliterated as Shantao Temple Station until 2003.
2. It bears the Shandao Temple's name.
|
null | false
| null |
Humans have occupied the region since the last ice age 10,000 years ago. Fremont people and Ancestral Puebloans lived in the area until about 700 years ago. Spanish missionaries encountered Ute and Paiute tribes in the area when they first came through in 1775, but the first European-Americans to attempt settlement in the area were the Mormon Elk Mountain Mission in 1855, who soon abandoned the area. Ranchers, farmers, and prospectors later settled Moab in the neighboring Riverine Valley in the late 1870s. Word of the beauty of the surrounding rock formations spread beyond the settlement as a possible tourist destination.
The Arches area was first brought to the attention of the National Park Service by Frank A. Wadleigh, passenger traffic manager of the Denver and Rio Grande Western Railroad. Wadleigh, accompanied by railroad photographer George L. Beam, visited the area in September 1923 at the invitation of Alexander Ringhoffer, a Hungarian-born prospector living in Salt Valley. Ringhoffer had written to the railroad to interest them in the tourist potential of a scenic area he had discovered the previous year with his two sons and a son-in-law, which he called the Devils Garden (known today as the Klondike Bluffs). Wadleigh was impressed by what Ringhoffer showed him, and suggested to Park Service director Stephen T. Mather that the area be made a national monument.
The following year, additional support for the monument idea came from Laurence Gould, a University of Michigan graduate student (and future polar explorer) studying the geology of the nearby La Sal Mountains, who was shown the scenic area by local physician Dr. J. W. "Doc" Williams.
A succession of government investigators examined the area, in part due to confusion as to the precise location. In the process, the name Devils Garden was transposed to an area on the opposite side of Salt Valley that includes Landscape Arch, the longest arch in the park. Ringhoffer's original discovery was omitted, while another area nearby, known locally as the Windows, was included. Designation of the area as a national monument was supported by the Park Service in 1926 but was resisted by President Calvin Coolidge's Interior Secretary, Hubert Work. Finally, in April 1929, shortly after his inauguration, President Herbert Hoover signed a presidential proclamation creating the Arches National Monument, consisting of two comparatively small, disconnected sections. The purpose of the reservation under the 1906 Antiquities Act was to protect the arches, spires, balanced rocks, and other sandstone formations for their scientific and educational value. The name Arches was suggested by Frank Pinkely, superintendent of the Park Service's southwestern national monuments, following a visit to the Windows section in 1925.
In late 1938, President Franklin D. Roosevelt signed a proclamation that enlarged the Arches to protect additional scenic features and permit the development of facilities to promote tourism. A small adjustment was made by President Dwight Eisenhower in 1960 to accommodate a new road alignment.
In early 1969, just before leaving office, President Lyndon B. Johnson signed a proclamation substantially enlarging the Arches. Two years later, President Richard Nixon signed legislation enacted by Congress, which significantly reduced the total area enclosed, but changed its status. Arches National Park was formally dedicated in May 1972.
In 1980, vandals attempted to use an abrasive kitchen cleanser to deface ancient petroglyphs in the park, prompting park officials to recruit physicist John F. Asmus, who specialized in using lasers to restore works of art, to use his technology to repair the damage. Asmus "zapped the panel with intense light pulses and succeeded in removing most of the cleanser". In 2016, there was another vandalism event on Frame Arch in the park, where a section if the rock was carved out. Due to advances in technology, in 2018 the arch was repaired through color match and modern infilling methods.
|
Summarise me a history of the Aches National Park creation
|
First Humans appeared in this region during the last ice age 10000 years ago. The first settlers began in 1775 but were abandoned before 1855 when ranchers, farmers, and prospectors settled in Moab(the closest city to the Arches). The first attention to these beautiful rock formations was broad in 1923 by Frank A. Wadleigh. Devils Garden(known as Klondike Bluffs today) got its name from a prospector from Salt Valley Alexander Ringhoffer, Devils Garden was transposed to an area on the opposite side of Salt Valley. Landscape Arch is the longest arch in the park and the fifth longest in the world!
In 1929, President Herbert Hoover signed a presidential proclamation creating the Arches National Monument to protect the arches, spires, balanced rocks, and other sandstone formations for their scientific and educational value. The name Arches was suggested by Frank Pinkely, in 1925. From 1938 multiple presidents of the United States of America signed multiple proclamations to protest scenic features and permits for road constructions in order to attract tourists into the area but only in 1972 Arches National Park was officially created.
In 1980, vandals tried to destroy ancient petroglyphs in the park, which physicist John F. Asmus restored using laser technology for restoration. In 2016, another vandalism event happened on Frame Arch a section of rocks was carved, but thanks to technology, it was restored.
|
null | false
| null |
Who was the King of Swing
|
Benny Goodman
|
|
null | false
| null |
True or false, Chesapeake Bay Crab Spice are a flavor of Lays potato chip?
|
As hard as it may be to believe, Chesapeake Bay Crab Spice is a Lays potato chip flavor.
|
|
null | false
| null |
The name Birkenhead probably means "headland overgrown with birch", from the Old English bircen meaning birch tree, of which many once grew on the headland which jutted into the river at Woodside. The name is not derived from the Birket, a stream which enters the Mersey between Birkenhead and Seacombe. The Birket is a later name which was introduced by Ordnance Survey.
|
Given this paragraph about the town of Birkenhead, what the does the name Birkenhead probably mean?
|
It probably means "headland overgrown with birch".
|
null | false
| null |
What is a traditional IRA or 401k?
|
The two primary options for retirement savings include individual retirement accounts and a defined contribution plan such as a 401k. Individuals can choose to save on their own and open an IRA, however a IRA opened by an individual won't carry the benefit of employer matching. Typically individuals contribute a percentage of their salary to their 401k, and many employers may offer matching contributions to that 401k up to a specified dollar amount.
|
|
null | false
| null |
What is the fastest car?
|
In 2017, the Koenigsegg Agera RS reached a top speed of 447.19 km/h (277.87 mph).
|
|
null | false
| null |
What team is the latest expansion in Major League Soccer?
|
The latest expansion team in Major League Soccer is St. Louis FC. They started off their first season in 2023 with five straight wins and a goal difference of plus fifteen.
|
|
2003.06279
| false
| null |
In a more practical scenario, text networks have been used in text classification tasks BIBREF8, BIBREF9, BIBREF10. The main advantage of the model is that it does not rely on deep semantical information to obtain competitive results. Another advantage of graph-based approaches is that, when combined with other approaches, it yields competitive results BIBREF11. A simple, yet recurrent text model is the well-known word co-occurrence network. After optional textual pre-processing steps, in a co-occurrence network each different word becomes a node and edges are established via co-occurrence in a desired window. A common strategy connects only adjacent words in the so called word adjacency networks.
While the co-occurrence representation yields good results in classification scenarios, some important features are not considered in the model. For example, long-range syntactical links, though less frequent than adjacent syntactical relationships, might be disregarded from a simple word adjacency approach BIBREF12. In addition, semantically similar words not sharing the same lemma are mapped into distinct nodes. In order to address these issues, here we introduce a modification of the traditional network representation by establishing additional edges, referred to as “virtual” edges. In the proposed model, in addition to the co-occurrence edges, we link two nodes (words) if the corresponding word embedding representation is similar. While this approach still does not merge similar nodes into the same concept, similar nodes are explicitly linked via virtual edges.
A simple, yet recurrent text model is the well-known word co-occurrence network. After optional textual pre-processing steps, in a co-occurrence network each different word becomes a node and edges are established via co-occurrence in a desired window. A common strategy connects only adjacent words in the so called word adjacency networks.
While the co-occurrence representation yields good results in classification scenarios, some important features are not considered in the model. For example, long-range syntactical links, though less frequent than adjacent syntactical relationships, might be disregarded from a simple word adjacency approach BIBREF12.
|
What is the reason that traditional co-occurrence networks fail in establishing links between similar words whenever they appear distant in the text?
|
The answers are shown as follows:
* long-range syntactical links, though less frequent than adjacent syntactical relationships, might be disregarded from a simple word adjacency approach
|
2004.04315
| false
| null |
We collected COVID-19 related Arabic tweets from January 1, 2020 until April 15, 2020, using Twitter streaming API and the Tweepy Python library. We have collected more than 3,934,610 million tweets so far. In our dataset, we store the full tweet object including the id of the tweet, username, hashtags, and geolocation of the tweet. We created a list of the most common Arabic keywords associated with COVID-19. Using Twitter’s streaming API, we searched for any tweet containing the keyword(s) in the text of the tweet. Table TABREF1 shows the list of keywords used along with the starting date of tracking each keyword. Furthermore, Table TABREF2 shows the list of hashtags we have been tracking along with the number of tweets collected from each hashtag. Indeed, some tweets were irrelevant, and we kept only those that were relevant to the pandemic.
We collected COVID-19 related Arabic tweets from January 1, 2020 until April 15, 2020, using Twitter streaming API and the Tweepy Python library.
|
Over what period of time were the tweets collected?
|
The answers are shown as follows:
* from January 1, 2020 until April 15, 2020
|
null | false
| null |
what are the components of modern computers?
|
a modern computer consists of at least one processing element, typically a central processing unit (CPU) in the form of a microprocessor, along with some type of computer memory, typically semiconductor memory chips. The processing element carries out arithmetic and logical operations, and a sequencing and control unit can change the order of operations in response to stored information. Peripheral devices include input devices (keyboards, mice, joystick, etc.), output devices (monitor screens, printers, etc.), and input/output devices that perform both functions (e.g., the 2000s-era touchscreen). Peripheral devices allow information to be retrieved from an external source and they enable the result of operations to be saved and retrieved.
|
|
null | false
| null |
Classify each of the following whether they help or hurt vision: glasses, binoculars, blindfold, sleep mask, telescope
|
Glasses: Help Vision
Binoculars: Help Vision
Blindfold: Hurt Vision
Sleep Mask: Hurt Vision
Telescope: Help Vision
|
|
null | false
| null |
The World Happiness Report is a publication that contains articles and rankings of national happiness, based on respondent ratings of their own lives, which the report also correlates with various (quality of) life factors. As of March 2023, Finland has been ranked the happiest country in the world six times in a row.
The report is a publication of the Sustainable Development Solutions Network, a global initiative of the United Nations. The report primarily uses data from the Gallup World Poll. Each annual report is available to the public to download on the World Happiness Report website. The Editors of the 2020 report are John F. Helliwell, Richard Layard, Jeffrey D. Sachs, and Jan-Emmanuel De Neve. Associate Editors are Lara Aknin, Shun Wang, and Haifang Huang
|
What is the happiness index report and which country tops the happiness index in the world
|
United nations every year in collaboration with sustainable development solution network perform survey and release the report global happiness index. As of March 2023, Finland has been ranked the happiest country in the world six times in a row.
The World Happiness Report is a publication that contains articles and rankings of national happiness, based on respondent ratings of their own lives, which the report also correlates with various (quality of) life factors
|
null | false
| null |
Find the bird in this list: Arctic skua, Monkey, Lily
|
Arctic skua
|
|
null | false
| null |
Give me a bullet list of lots of possible weather forecasts
|
- Sunny
- Partially cloudy
- Cloudy
- Showers
- Thunderstorms
- Hail
- Snowstorm
- Snow flurries
- Heavy fog
|
|
null | false
| 251
|
As shown in the previous section, using the right data is critical for achieving good performance on an in-domain test set, and more data is not necessarily better. However, in real-world scenarios, the availability of data labeled by domain is limited, e.g. when working with large scale, web-crawled data. In this section we focus on a data-selection scenario where only a very small number of in-domain sentences are used to select data from a larger unlabeled parallel corpus. An established method for data selection was proposed by BIBREF4, which was also used in training the winning systems in WMT 2019 BIBREF39, BIBREF40. This method compares the cross-entropy, according to domain-specific and non-domain-specific language models, for each candidate sentence for selection. The sentences are then ranked by the cross-entropy difference, and only the top sentences are selected for training.
While the method by BIBREF4 is tried-and-true, it is based on simple n-gram language models which cannot generalize beyond the n-grams that are seen in the in-domain set. In addition, it is restricted to the in-domain and general-domain datasets it is trained on, which are usually small. On the contrary, pre-trained language models are trained on massive amounts of text, and, as we showed through unsupervised clustering, learn representations with domain-relevant information. In the following sections, we investigate whether this property of pretrained language models makes them useful for domain data selection.
In this section we focus on a dataselection scenario where only a very small number of in-domain sentences are used to select data from a larger unlabeled parallel corpus.
|
What data do they select with pre-trained language models?
|
They focus on a dataselection scenario where only a very small number of in-domain sentences.
|
null | false
| 104
|
Semantic Role Labeling (SRL) has emerged as an important task in Natural Language Processing (NLP) due to its applicability in information extraction, question answering, and other NLP tasks. SRL is the problem of finding predicate-argument structure in a sentence, as illustrated below:
INLINEFORM0
Here, the predicate WRITE has two arguments: `Mike' as A0 or the writer, and `a book' as A1 or the thing written. The labels A0 and A1 correspond to the PropBank annotations BIBREF0 .
As the need for SRL arises in different domains and languages, the existing manually annotated corpora become insufficient to build supervised systems. This has motivated work on unsupervised SRL BIBREF1 , BIBREF2 , BIBREF3 . Previous work has indicated that unsupervised systems could benefit from the word alignment information in parallel text in two or more languages BIBREF4 , BIBREF5 , BIBREF6 . For example, consider the German translation of sentence INLINEFORM0 :
INLINEFORM0
If sentences INLINEFORM0 and INLINEFORM1 have the word alignments: Mike-Mike, written-geschrieben, and book-Buch, the system might be able to predict A1 for Buch, even if there is insufficient information in the monolingual German data to learn this assignment. Thus, in languages where the resources are sparse or not good enough, or the distributions are not informative, SRL systems could be made more accurate by using parallel data with resource rich or more amenable languages.
In this paper, we propose a joint Bayesian model for unsupervised semantic role induction in multiple languages. The model consists of individual Bayesian models for each language BIBREF3 , and crosslingual latent variables to incorporate soft role agreement between aligned constituents. This latent variable approach has been demonstrated to increase the performance in a multilingual unsupervised part-of-speech tagging model based on HMMs BIBREF4 . We investigate the application of this approach to unsupervised SRL, presenting the performance improvements obtained in different settings involving labeled and unlabeled data, and analyzing the annotation effort required to obtain similar gains using labeled data.
We begin by briefly describing the unsupervised SRL pipeline and the monolingual semantic role induction model we use, and then describe our multilingual model.
This latent variable approach has been demonstrated to increase the performance in a multilingual unsupervised part-of-speech tagging model based on HMMs.
|
Which model's performance does the crosslingual latent variable approach improve?
|
A multilingual unsupervised part-of-speech tagging model based on HMMs.
|
null | false
| null |
Breathing apparatus
Main article: Scuba set
Recreational diver putting on his scuba set before diving
The defining equipment used by a scuba diver is the eponymous scuba, the self-contained underwater breathing apparatus which allows the diver to breathe while diving, and is transported by the diver. It is also commonly referred to as the scuba set.
As one descends, in addition to the normal atmospheric pressure at the surface, the water exerts increasing hydrostatic pressure of approximately 1 bar (14.7 pounds per square inch) for every 10 m (33 feet) of depth. The pressure of the inhaled breath must balance the surrounding or ambient pressure to allow controlled inflation of the lungs. It becomes virtually impossible to breathe air at normal atmospheric pressure through a tube below three feet under the water.
Most recreational scuba diving is done using a half mask which covers the diver's eyes and nose, and a mouthpiece to supply the breathing gas from the demand valve or rebreather. Inhaling from a regulator's mouthpiece becomes second nature very quickly. The other common arrangement is a full face mask which covers the eyes, nose and mouth, and often allows the diver to breathe through the nose. Professional scuba divers are more likely to use full face masks, which protect the diver's airway if the diver loses consciousness.
Open-circuit
Main article: Diving regulator
Aqualung Legend second stage (demand valve) regulator
Aqualung first stage regulator
Gekko dive computer with attached pressure gauge and compass
Suunto submersible pressure gauge display
Open circuit scuba has no provision for using the breathing gas more than once for respiration. The gas inhaled from the scuba equipment is exhaled to the environment, or occasionally into another item of equipment for a special purpose, usually to increase the buoyancy of a lifting device such as a buoyancy compensator, inflatable surface marker buoy or small lifting bag. The breathing gas is generally provided from a high-pressure diving cylinder through a scuba regulator. By always providing the appropriate breathing gas at ambient pressure, demand valve regulators ensure the diver can inhale and exhale naturally and without excessive effort, regardless of depth, as and when needed.
The most commonly used scuba set uses a "single-hose" open circuit 2-stage demand regulator, connected to a single back-mounted high-pressure gas cylinder, with the first stage connected to the cylinder valve and the second stage at the mouthpiece. This arrangement differs from Émile Gagnan's and Jacques Cousteau's original 1942 "twin-hose" design, known as the Aqua-lung, in which the cylinder pressure was reduced to ambient pressure in one or two stages which were all in the housing mounted to the cylinder valve or manifold. The "single-hose" system has significant advantages over the original system for most applications.
In the "single-hose" two-stage design, the first stage regulator reduces the cylinder pressure of up to about 300 bars (4,400 psi) to an intermediate pressure (IP) of about 8 to 10 bars (120 to 150 psi) above ambient pressure. The second stage demand valve regulator, supplied by a low-pressure hose from the first stage, delivers the breathing gas at ambient pressure to the diver's mouth. The exhaled gases are exhausted directly to the environment as waste through a non-return valve on the second stage housing. The first stage typically has at least one outlet port delivering gas at full tank pressure which is connected to the diver's submersible pressure gauge or dive computer, to show how much breathing gas remains in the cylinder.
Rebreather
An Inspiration electronic fully closed circuit rebreather
Main article: Diving rebreather
Less common are closed circuit (CCR) and semi-closed (SCR) rebreathers which, unlike open-circuit sets that vent off all exhaled gases, process all or part of each exhaled breath for re-use by removing the carbon dioxide and replacing the oxygen used by the diver. Rebreathers release few or no gas bubbles into the water, and use much less stored gas volume, for an equivalent depth and time because exhaled oxygen is recovered; this has advantages for research, military, photography, and other applications. Rebreathers are more complex and more expensive than open-circuit scuba, and special training and correct maintenance are required for them to be safely used, due to the larger variety of potential failure modes.
In a closed-circuit rebreather the oxygen partial pressure in the rebreather is controlled, so it can be maintained at a safe continuous maximum, which reduces the inert gas (nitrogen and/or helium) partial pressure in the breathing loop. Minimising the inert gas loading of the diver's tissues for a given dive profile reduces the decompression obligation. This requires continuous monitoring of actual partial pressures with time and for maximum effectiveness requires real-time computer processing by the diver's decompression computer. Decompression can be much reduced compared to fixed ratio gas mixes used in other scuba systems and, as a result, divers can stay down longer or require less time to decompress. A semi-closed circuit rebreather injects a constant mass flow of a fixed breathing gas mixture into the breathing loop, or replaces a specific percentage of the respired volume, so the partial pressure of oxygen at any time during the dive depends on the diver's oxygen consumption and/or breathing rate. Planning decompression requirements requires a more conservative approach for a SCR than for a CCR, but decompression computers with a real-time oxygen partial pressure input can optimise decompression for these systems. Because rebreathers produce very few bubbles, they do not disturb marine life or make a diver's presence known at the surface; this is useful for underwater photography, and for covert work.
|
What are the primary breathing apparatuses for scuba
|
The most common breathing apparatus for Scuba Diving is an open circuit single hose 2-stage demand regulator connected to a single gas cylinder. A less common apparatus is a Closed Circuit or Semi-Closed rebreather.
|
null | false
| 203
|
We base our experiments on a large collection of Bible translations crawled from the web, coming from various sources and periods of times. Any other multilingual data collection would work as well, but with the selected corpus we have the advantage that we cover the same genre and roughly the same coverage for each language involved. It is also easy to divide the data into training and test sets by using Bible verse numbers, which allows us to control for semantic similarity between languages in a way that would have been difficult in a corpus that is not multi-parallel. Altogether we have 1,303 translations in 990 languages that we can use for our purposes. These were chosen so that the model alphabet size is below 1000 symbols, which was satisfied by choosing only translations in Latin, Cyrillic or Greek script.
Certainly, there are disadvantages as well, such as the limited size (roughly 500 million tokens in total, with most languages having only one translation of the New Testament each, with roughly 200 thousand tokens), the narrow domain and the high overlap of named entities. The latter can lead to some unexpected effects when using nonsensical language vectors, as the model will then generate a sequence of random names.
The corpus deviates in some ways from an ideal multi-parallel corpus. Most translations are of the complete New Testament, whereas around 300 also contain the Old Testament (thus several times longer), and around ten contain only portions of the New Testament. Additionally, several languages have multiple translations, which are then concatenated. These translations may vary in age and style, but historical versions of languages (with their own ISO 639-3 code) are treated as distinct languages. During training we enforce a uniform distribution between languages when selecting training examples.
Altogether we have 1,303 translations in 990 languages that we can use for our purposes. These were chosen so that the model alphabet size is below 1000 symbols, which was satisfied by choosing only translations in Latin, Cyrillic or Greek script. Certainly, there are disadvantages as well, such as the limited size (roughly 500 million tokens in total, with most languages having only one translation of the New Testament each, with roughly 200 thousand tokens), the narrow domain and the high overlap of named entities.
|
What is the limitation of the chosen data?
|
The limited size, the narrow domain, and the high overlap of named entities.
|
1909.05890
| true
| null |
The precision when labeling the first x ranked tweets as attack tweet is shown in the figure FIGREF39. The x-axis is the number of ranked tweets treated as attack tweets. And the y-axis is the corresponding precision. The straight line in figures FIGREF39, FIGREF43 and FIGREF51 is the result of a supervised LDA algorithm which is used as a baseline. Supervised LDA achieved 96.44 percent precision with 10 fold cross validation.
The straight line in figures FIGREF39, FIGREF43 and FIGREF51 is the result of a supervised LDA algorithm which is used as a baseline.
|
Was performance of the weakly-supervised model compared to the performance of a supervised model?
|
Yes.
|
null | false
| null |
On August 29, 2019, Blumhouse Productions acquired the rights to make a Merrily We Roll Along film, with Richard Linklater directing the film and producing along with Ginger Sledge, Jason Blum, and Jonathan Marc Sherman. Ben Platt, Beanie Feldstein and Blake Jenner were cast to star in the film. Production on the film was announced to commence every couple of years to reflect the characters age over 20 years, similar to how Boyhood (also written and directed by Linklater) was shot over 12 years. The film is based on the latest version of Furth's book, which was adapted by Linklater into the film's screenplay.
Jenner exited the film in 2019 (following the filming of the "Our Time" sequence) following allegations of domestic abuse made by his ex-wife Melissa Benoist. Paul Mescal took over the role of Franklin Shepard proceeding forward, and the sequence was reshot.
|
How is the film Merrily We Roll Along being filed?
|
Director Richard Linklater plans to shoot the film Merrily We Roll Along over 20 years, shooting a few scenes each year to accurately capture the actors as they age. This is similar to how Linklater filmed his previous film, Boyhood, which was shot with the same actors over 12 years.
|
null | false
| null |
Where did Father's Day originate?
|
Father's Day was founded in Spokane, Washington.
|
|
null | false
| null |
How do archaeologists study the past?
|
Archaeologists study the past by implementing a variety of scientific strategies that include survey, excavation, analysis, interpretation, and reporting. They begin by determining an area of interest to be studied. They then survey that area using a variety of methods that include laying in a grid system for testing the area of interest to locate sites. Next, they excavate, or dig, to find evidence of past habitations. This results in recovering material culture, or artifacts. The artifacts are then taken to a laboratory where they are analyzed. This analysis including cleaning the artifacts, determining dates and functions, and entering those artifacts into a database where further statistical analysis may be performed. Once all this data is gathered and analyzed, archaeologist need to interpret what they have found. This data can tell you when a site was occupied, who lived there, what they were eating, and what time of year a site was occupied. Once the archaeologists have interpreted the data the write a report detailing the work that was done and their interpretation of the site. The report is very detailed as archaeological excavation destroys the sites being studying. The report includes photographs, field notes, and an inventory of the artifacts recovered as well as an interpretation of the site.
|
|
null | false
| null |
Identify which instrument is string or percussion: Taxi horn, Sihu
|
Sihu is string, Taxi horn is percussion.
|
|
null | false
| null |
What is the best BBQ restaurant in Raleigh?
|
Raleigh has a lot of great barbecue restaurants. For traditional NC-style BBQ try Clyde Coopers located downtown. If you like Texas-style BBQ Prime BBQ in Knightdale has great brisket and beef ribs.
|
|
null | false
| null |
What is a coded application?
|
A application is an interface that lets people connect to data and insights through a front end application. An application does not have to have a graphical user interface either. An application can also be a coded program that can help with moving data or connect to an API to perform some sort of service like data cleaning, data moving, or data quality. Windows is an example of a GUI operating system, and DOS is an example of an application without a GUI that can do functionality like you can do in windows.
|
|
null | false
| null |
The children were sent mainly to the four Dominion countries, Canada 1,532 (in nine parties), Australia 577 (three parties), New Zealand 202 (two parties), and South Africa 353 (two parties), and 838 to the USA under the United States Committee for the Care of European Children programme. In the first few months over 211,000 children were registered with the scheme. A further 24,000 children had been approved for sailing in that time and over 1,000 volunteer escorts, including doctors and nurses, enrolled. It was planned as a temporary exile for the children, to return home to their families when conditions permitted.
|
In the first few months how many children were registered?
|
211,000 children were registered in the scheme in the first few months.
|
null | false
| null |
Rita Moreno (born Rosa Dolores Alverío Marcano;[nb 1] December 11, 1931) is a Puerto Rican actress, dancer, and singer. She is noted for her work on stage and screen in a career spanning over seven decades. Moreno is one of the last remaining stars from the Golden Age of Hollywood. Among her numerous accolades, she is one of a few performers to have been awarded an Emmy, a Grammy, an Oscar, and a Tony (EGOT) and the Triple Crown of Acting, with individual competitive Academy, Emmy, and Tony awards. Additional accolades include the Presidential Medal of Freedom in 2004, the National Medal of Arts in 2009, the Screen Actors Guild Life Achievement Award in 2013, the Kennedy Center Honor in 2015, and a Peabody Award in 2019.
Moreno's early work included supporting roles in the classic musical films Singin' in the Rain (1952) and The King and I (1956), before her breakout role as Anita in West Side Story (1961), which earned her the Oscar for Best Supporting Actress, becoming the first Latin American woman to win an Academy Award. Her other notable films include Popi (1969), Carnal Knowledge (1971), The Four Seasons (1981), I Like It Like That (1994) and the cult film Slums of Beverly Hills (1998). Moreno portrayed Valentina in the 2021 remake of West Side Story directed by Steven Spielberg.
In theater, she starred as Googie Gomez in the 1975 Terrence McNally musical The Ritz earning her the Tony Award for Best Featured Actress in a Musical. She reprised her role in the 1976 film directed by Richard Lester which earned her a BAFTA Award for Best Actress nomination. She also acted in Lorraine Hansberry's The Sign in Sidney Brustein's Window in 1964 and in Neil Simon's The Odd Couple in 1985.
She was a cast member on the children's television series The Electric Company (1971-1977), and played Sister Peter Marie Reimondo on the HBO series Oz (1997-2003). She received two consecutive Primetime Emmy Awards for her roles on The Muppet Show in 1977 and The Rockford Files in 1978. She gained acclaim for her roles in Where on Earth Is Carmen Sandiego? (1994-1999), The CW series Jane the Virgin (2015–2019), and the Netflix revival of One Day at a Time (2017–2020). Her life was profiled in Rita Moreno: Just a Girl Who Decided to Go for It (2021).
|
Without quoting directly from the text give me a summary of Rita Moreno's greatest achievements.
|
Rita Moreno is an actress, dancer, and singer from Puerto Rico whose career spans several decades. She is one of the few people to have accomplished EGOT status by winning an Emmy, a Grammy, an Oscar, and a Tony. She is most well known for being the first Latin American woman to receive an Oscar for her work in Westside Story in 1961.
|
null | false
| 304
|
We present an information extraction architecture that augments a candidate-generating parser with a deep neural network. The candidate-generating parser may leverage constraints. At the same time, the architecture gains the neural networks's ability to leverage large amounts of data to learn complex features that are tuned for the application at hand. Our method assumes the existence of a potentially noisy source of supervision INLINEFORM0 , e.g. via consistency checks of extracted data against existing databases, or via human interaction. This supervision is used to train the neural network.
Our extraction system has three advantages over earlier work on information extraction with deep neural networks BIBREF2 , BIBREF3 , BIBREF4 , BIBREF5 , BIBREF6 , BIBREF7 , BIBREF8 , BIBREF9 :
Our system leverages “free” data to train a deep neural network, and does not require large-scale manual annotation. The network is trained with noisy supervision provided by measures of consistency with existing databases (e.g. an extraction ts_tick_abs (US_Unemployment, 49%) would be implausible given recent US employment history). With slight modifications, our pipeline could be trained with supervision from human interaction, such as clicks on online advertisements. Learning without explicit annotations is critical in applications where large-scale manual annotation would be prohibitively expensive.
If an extractor for the given application has already been built, the neural network boosts its accuracy without the need to re-engineer or discard the existing solution. Even for new systems, the decoupling of candidate-generation and the neural network offers advantages: the candidate-generating parser can easily enforce contraints that would be difficult to support in an algorithm relying entirely on a neural network. Note that, in particular, a carefully engineered candidate-generating parser enforces constraints intelligently, and can in many instances eliminate the need to evaluate computationally expensive constraints, e.g. API calls.
We encode the candidate-generating parser's document annotations character-by-character into vectors INLINEFORM0 that also include a one-hot encoding of the character itself. We believe that this encoding makes it easier for the network to learn character-level characteristics of the entities in the semantic relation. Moreover, our encoding lends itself well to processing both by recurrent architectures (processing character-by-character input vectors INLINEFORM1 ) and convolutional architectures (performing 1D convolutions over an input matrix whose columns are vectors INLINEFORM2 ).
In a production setting, the neural architecture presented here reduced the number of false positive extractions in financial information extraction application by INLINEFORM0 relative to a mature system developed over the course of several years.
We present an information extraction architecture that augments a candidate-generating parser with a deep neural network.
|
What architecture do they propose?
|
An information extraction architecture that augments a candidate-generating parser with a deep neural network.
|
null | false
| null |
Thomas Jefferson (April 13, 1743 – July 4, 1826) was an American statesman, diplomat, lawyer, architect, philosopher, and Founding Father who served as the third president of the United States from 1801 to 1809. Among the Committee of Five charged by the Second Continental Congress with authoring the Declaration of Independence, Jefferson was the Declaration's primary author. Following the American Revolutionary War and prior to becoming the nation's third president in 1801, Jefferson was the first United States secretary of state under George Washington and then the nation's second vice president under John Adams.
|
Who is Thomas Jefferson?
|
Thomas Jefferson (April 13, 1743 – July 4, 1826) was an American statesman, diplomat, lawyer, architect, philosopher, and Founding Father who served as the third president of the United States from 1801 to 1809. Among the Committee of Five charged by the Second Continental Congress with authoring the Declaration of Independence, Jefferson was the Declaration's primary author. Following the American Revolutionary War and prior to becoming the nation's third president in 1801, Jefferson was the first United States secretary of state under George Washington and then the nation's second vice president under John Adams.
Among the nation's Founding Fathers, Jefferson is considered unmatched in his intellectual depth and breadth. His passionate writings and advocacy for human rights, including freedom of thought, speech, and religion, were a leading inspiration behind the American Revolution, which ultimately gave rise to the American Revolutionary War, American independence, and the United States Constitution. Jefferson's ideas were globally influential in shaping and inspiring the Age of Enlightenment, which proved transformational in the late 17th and 18th centuries. He was a leading proponent of democracy, republicanism, and individual rights, and produced formative documents and decisions at the state, national, and international levels.
During the American Revolution, Jefferson represented Virginia in the Second Continental Congress in Philadelphia, which adopted the Declaration of Independence on July 4, 1776. As a Virginia legislator, he drafted a state law for religious freedom. He served as the second Governor of Virginia from 1779 to 1781, during the Revolutionary War. In 1785, Jefferson was appointed the United States Minister to France, and subsequently, the nation's first secretary of state under President George Washington from 1790 to 1793. Jefferson and James Madison organized the Democratic-Republican Party to oppose the Federalist Party during the formation of the First Party System. With Madison, he anonymously wrote the Kentucky and Virginia Resolutions in 1798 and 1799, which sought to strengthen states' rights by nullifying the federal Alien and Sedition Acts.
Jefferson and Federalist John Adams became friends as well as political rivals, serving in the Continental Congress and drafting the Declaration of Independence together. In the 1796 presidential election between the two, Jefferson came in second, which according to electoral procedure at the time, made him vice president to Adams. Jefferson challenged Adams again in 1800 and won the presidency. After his term in office, Jefferson eventually reconciled with Adams and they shared a correspondence that lasted 14 years. He and Adams both died on the same day, July 4, 1826, which was also the 50th anniversary of Declaration of Independence.
As president, Jefferson pursued the nation's shipping and trade interests against Barbary pirates and aggressive British trade policies. Starting in 1803, he promoted a western expansionist policy with the Louisiana Purchase, which doubled the nation's claimed land area. To make room for settlement, Jefferson began the process of Indian tribal removal from the newly acquired territory. As a result of peace negotiations with France, his administration reduced military forces. He was re-elected in 1804, but his second term was beset with difficulties at home, including the trial of former vice president Aaron Burr. In 1807, American foreign trade was diminished when Jefferson implemented the Embargo Act in response to British threats to U.S. shipping. The same year, Jefferson signed the Act Prohibiting Importation of Slaves.
Jefferson was a plantation owner, lawyer, and politician, and mastered many disciplines including surveying, mathematics, horticulture, and mechanics. He was also an architect in the Palladian tradition. Jefferson's keen interest in religion and philosophy led to his appointment as president of the American Philosophical Society. He largely shunned organized religion but was influenced by Christianity, Epicureanism, and deism. Jefferson rejected fundamental Christianity, denying Christ's divinity. A philologist, Jefferson knew several languages. He was a prolific letter writer and corresponded with many prominent people, including Edward Carrington, John Taylor of Caroline, and James Madison. In 1785, Jefferson authored Notes on the State of Virginia, considered perhaps the most important American book published before 1800. Jefferson championed the ideals, values, and teachings of the Enlightenment.
Since the 1790s, Jefferson was rumored to have had children by his sister-in-law and slave Sally Hemings, leading to what is known as the Jefferson-Hemings controversy. A 1998 DNA test concluded that one of Sally Hemings's children, Eston Hemings, was of the Jefferson male line. According to scholarly consensus, based on documentary and statistical evaluation, as well as oral history, Jefferson probably fathered at least six children with Hemings, including four that survived to adulthood.
After retiring from public office, Jefferson founded the University of Virginia. Presidential scholars and historians generally praise Jefferson's public achievements, including his advocacy of religious freedom and tolerance in Virginia, his peaceful acquisition of the Louisiana Territory from France without war or controversy, and his ambitious and successful Lewis and Clark Expedition. Some modern historians are critical of Jefferson's personal involvement with slavery. Jefferson is consistently ranked among the top ten presidents of American history.
|
null | false
| null |
Classify each of the following as characters in the 2003 reboot of Battlestar Galactica or Pokemon: Starbuck, Admiral Adama, Snorlax, Lee Adama, Dugtrio, Gaius Baltar, Laura Roslin, Venusaur, Lapras
|
Battlestar Galactica: Starbuck, Admiral Adam, Lee Adama, Gaius Baltar, Laura Roslin
Pokemon: Snorlax, Dugtrio, Venosaur, Lapras, Venusaur
|
|
null | false
| null |
What are some good body weight exercises for the lower body?
|
Some good exercises to strengthen the lower body include box squats, lunges, side lunges, one leg Romanian dead lifts, glute marches, and curtsey squats.
|
|
null | false
| null |
Is the Vex or FIRST robotics competition a better choice for a child in high school?
|
Both VEX and FIRST have very strong robotics competitions for the high school level. If you are a parent choosing between the two, multiple factors need to be taken into consideration.
The first factor is the availability of a robotics program in your area. While you can form an independent team that is not affiliated with your school or another organization, it will require a significant financial and time investment. At the high school level teams typically design their robots with CAD software and must machine the parts themselves. Teams often consist of 20 or more members so that all of the aspects of the competition can be covered, such as creating and assembling the hardware, electrical wiring, systems design, programming and documenting the engineering process. All of the parts and equipment need to be purchased and maintained, and members put in a lot of hours during the competition season. Joining a pre-existing team that already has equipment, members and mentors familiar with a program will make it easiest for a child to participate, regardless of which competition the team is a part of.
If both VEX and FIRST programs are available, then it depends on how intense of an experience the child is interested in. The VEX competition has a lower barrier to entry because it requires smaller robots. Two pairs of robots compete against each other in a match to see which pair can get more points. While teams will need to machine some parts and must have skills in electrical and software design, a team can be smaller and still compete effectively. The FIRST competition involves much larger and more complex robots and challenges. Two alliances of three robots each compete against each other in a match, with a total of six robots on the field at once. Teams need to have more members to effectively compete, and participants need more in-depth training on hardware, electrical, system and software design. There is also a business requirement, where teams have to create a marketing banner, brochure and video for their team to submit as part of the competition.
Whether you choose VEX or FIRST, a child in high school that participates in a well-run robotics program will gain valuable engineering and organizational experience that will serve them well, both on a college application and after graduation.
|
|
null | false
| null |
The climate of the mountain varies from humid tropical at its base to alpine at its summit. The temperature at the summit of Mount Kinabalu usually stands from −4 to 8 °C (25 to 46 °F) from December to January, and 3 to 12 °C (37 to 54 °F) from June to September. Due to the coldness of the mountain from December to January, there are a few occasions where frost and ice appear at the summit of Mount Kinabalu. Snow has been recorded three times in this area; in 1975, 1993 and 2022.
|
What temperature can I expect at the summit if I climb Mount Kinabalu
|
Temperature ranges from -4°C to 12 °C .
|
null | false
| null |
After his collaboration with Ibn Jazayy ended, Ibn Batuta retired to a judicial post in a small Moroccan provincial town, where he died in 1368.
Ibn Battuta has been called the greatest of all travel writers, having traveled farther than Marco Polo. In his work, he provided priceless glimpses of the various people, courts and religious monuments around the world. His travelogue has been the source of countless research projects and historical investigations.
Even if some of the stories were borrowed, and some of the tales a bit too marvelous to be believed, Ibn Battuta's rilha remains an enlightening and influential work of travel literature to this day.
|
Summarize legacy of ibn-Batuta
|
Ibn-Batuta was considered one of the greatest traveler of his time. His travelogues provided first ever glimpse of cultures, dynasties and religions of the world. Despite its shortcomings, his work remains influential even today.
|
null | false
| null |
Is Dirty picture a movie?
|
Yes. The Dirty Picture is a 2011 Indian Hindi-language biographical musical drama film inspired by the life of Silk Smitha, an Indian actress. It is a real life story.
|
|
null | false
| null |
The Indian Institutes of Technology (IITs) are prestigious central government owned public technical institutes located across India. Known for their excellence in education, they are under the ownership of the Ministry of Education of the Government of India. They are governed by the Institutes of Technology Act, 1961, declaring them as Institutes of National Importance and laying down their powers, duties, and framework for governance as the country's premier institutions in the field of technology.The act currently lists twenty-three IITs. Each IIT has autonomy and is linked to others through a common council called the IIT Council, which oversees their administration. The Minister of Education of India is the ex officio Chairperson of the IIT Council.
|
Tell me some key points about IITs from given text.
|
1. The Indian Institutes of Technology (IITs) are renowned public technical institutions owned by the Central government and located across India.
2. IITs are known for excellence in education owned by ministry of education of government of India.
3. The Institutes of Technology Act, 1961, which established them as Institutes of National
Importance and established their authority, accountability, and governing structure as the
nation's top technological institutions.
4. The act currently list 23 IITs. Each IIT is autonomous and linked with other via IIT council.
5. The ex officio Chairperson of the IIT Council is minister of education.
|
null | false
| null |
Given these two sets of instructions, will you end up at the same place if you started at the same point? First instructions: 1) Take 5 steps forward 2) Turn around 3) Take 3 steps forward. Second instructions: 1) Turn left 2) Turn left 3) Take 6 steps forward 4) Turn around 4) Take 4 steps forward.
|
No, you will not end up at the same place. The first set of instructions ends two steps in front of the starting point, since you go forward 5 steps then backwards 3 steps. The second set of instructions ends up two steps behind the starting point, since you turn left twice, which is the same as turning around, and then go forward 6 and backwards 4.
|
|
null | false
| null |
Give me three festive holiday meals I could serve to my family for Christmas
|
Here are 3 festive meals for the holidays:
* Prime Rib and Yorkshire Pudding is very traditional
* Turkey, mashed potatoes, cranberries and stuffing are a colorful and delicious American tradition
* Leg of lamb with mint sauce is also a cheerful option
|
|
null | false
| null |
What is carnitine?
|
Carnitine is a quaternary ammonium compound involved in metabolism in most mammals, plants, and some bacteria. In support of energy metabolism, carnitine transports long-chain fatty acids into mitochondria to be oxidized for free energy production, and also participates in removing products of metabolism from cells. Given its key metabolic roles, carnitine is concentrated in tissues like skeletal and cardiac muscle that metabolize fatty acids as an energy source. Generally individuals, including strict vegetarians, synthesize enough L-carnitine in vivo.
Carnitine exists as one of two stereoisomers (the two enantiomers d-carnitine (S-(+)-) and l-carnitine (R-(−)-)). Both are biologically active, but only l-carnitine naturally occurs in animals, and d-carnitine is toxic as it inhibits the activity of the l-form. At room temperature, pure carnitine is a whiteish powder, and a water-soluble zwitterion with relatively low toxicity. Derived from amino acids, carnitine was first extracted from meat extracts in 1905, leading to its name from Latin, "caro/carnis" or flesh.
Some individuals with genetic or medical disorders (such as preterm infants) cannot make enough carnitine, requiring dietary supplementation. Despite common carnitine supplement consumption among athletes for improved exercise performance or recovery, there is insufficient high-quality clinical evidence to indicate it provides any benefit.
|
|
null | false
| null |
In 1871, at the age of eight, Narendranath enrolled at Ishwar Chandra Vidyasagar's Metropolitan Institution, where he went to school until his family moved to Raipur in 1877. In 1879, after his family's return to Calcutta, he was the only student to receive first-division marks in the Presidency College entrance examination. He was an avid reader in a wide range of subjects, including philosophy, religion, history, social science, art and literature. He was also interested in Hindu scriptures, including the Vedas, the Upanishads, the Bhagavad Gita, the Ramayana, the Mahabharata and the Puranas. Narendra was trained in Indian classical music, and regularly participated in physical exercise, sports and organised activities. Narendra studied Western logic, Western philosophy and European history at the General Assembly's Institution (now known as the Scottish Church College). In 1881, he passed the Fine Arts examination, and completed a Bachelor of Arts degree in 1884. Narendra studied the works of David Hume, Immanuel Kant, Johann Gottlieb Fichte, Baruch Spinoza, Georg W. F. Hegel, Arthur Schopenhauer, Auguste Comte, John Stuart Mill and Charles Darwin. He became fascinated with the evolutionism of Herbert Spencer and corresponded with him, translating Herbert Spencer's book Education (1861) into Bengali. While studying Western philosophers, he also learned Sanskrit scriptures and Bengali literature.
|
As per the passage, list the subjects of which Narendranath was an avid reader.
|
philosophy, religion, history, social science, art and literature
|
1705.00108
| false
| null |
Our main contribution is to show that the context sensitive representation captured in the LM embeddings is useful in the supervised sequence tagging setting. When we include the LM embeddings in our system overall performance increases from 90.87% to 91.93% INLINEFORM0 for the CoNLL 2003 NER task, a more then 1% absolute F1 increase, and a substantial improvement over the previous state of the art. We also establish a new state of the art result (96.37% INLINEFORM1 ) for the CoNLL 2000 Chunking task.
When we include the LM embeddings in our system overall performance increases from 90.87% to 91.93% INLINEFORM0 for the CoNLL 2003 NER task, a more then 1% absolute F1 increase, and a substantial improvement over the previous state of the art. We also establish a new state of the art result (96.37% INLINEFORM1 ) for the CoNLL 2000 Chunking task.
|
what results do they achieve?
|
91.93% F1 score on CoNLL 2003 NER task and 96.37% F1 score on CoNLL 2000 Chunking task
|
1705.09665
| true
| null |
Our full dataset consists of all subreddits on Reddit from January 2013 to December 2014, for which there are at least 500 words in the vocabulary used to estimate our measures, in at least 4 months of the subreddit's history. We compute our measures over the comments written by users in a community in time windows of months, for each sufficiently active month, and manually remove communities where the bulk of the contributions are in a foreign language. This results in 283 communities ( INLINEFORM0 ), for a total of 4,872 community-months ( INLINEFORM1 ).
We compute our measures over the comments written by users in a community in time windows of months, for each sufficiently active month, and manually remove communities where the bulk of the contributions are in a foreign language.
|
Do they report results only on English data?
|
No.
|
null | false
| null |
Give me a list of ways that a runner can advance a base in baseball without there being a hit.
|
- The batter walks, forcing the runner to advance to the next base
- The runner steals a base
- The pitcher balks
- There is a wild pitch that goes past the catcher
|
|
null | false
| null |
Vivek Oberoi was born in Hyderabad, Telangana, to a Punjabi father, Suresh Oberoi, an actor. In 1974, Suresh married Yashodhara, from a Punjabi business family settled in the south, in Chennai. Because of his mother’s family being assimilated in the region, Oberoi is himself fluent in Tamil.
Oberoi went on to study at Mayo College, Ajmer and Mithibai College, Mumbai. At an actors' workshop in London, he was spotted by the director of New York University who took Oberoi to New York, where he completed his master's degree in film acting.
|
Who are the parents of Vivek Oberoi?
|
Suresh Oberoi and Yashodhara.
|
null | false
| null |
Who wrote Gone with the Wind
|
Margaret Mitchell
|
|
null | false
| null |
What modern city used to be known as New Amsterdam?
|
The modern city used to be known as New Amsterdam was founded in 1626, when the Dutch West India Company established a trading post at the tip of Manhattan Island. The city was renamed New York in 1664.
|
|
null | false
| null |
Let's play a game of tic tac toe. I'll be X and you be O. Unplayed squares will be "_". I'll get first.
_ _ _
_ O _
_ _ _
|
Fun! Here's my next move:
_ _ _
_ O X
_ _ _
|
|
2001.05672
| false
| null |
convertible.pl: implementing DCG rules for 1st and 3rd steps in the three-steps conversion, as well as other rules including lexicon.
convertible.pl: implementing DCG rules for 1st and 3rd steps in the three-steps conversion, as well as other rules including lexicon.
|
What DCGs are used?
|
Author's own DCG rules are defined from scratch.
|
null | false
| null |
Tell me which of these countries has more than 1 billion populations: Japan, Canada, United States, Russia
|
None of these
|
|
null | false
| null |
What is a good city to visit in Michigan in the summer?
|
Michigan is well known for the variety of activities it offers year round. In the summer, camping, golfing, bicycling, hiking and swimming are some of the popular activities locals and visitors enjoy. While there are many great cities in the state that cater to these and other activities, a perennial favorite is Traverse City. Traverse City is located on the western side of Michigan right on Lake Michigan and Grand Traverse Bay. This side of Michigan is often referred to as the Middle or Third Coast given the abundant shoreline. Traverse City is known for ease of access to all of the activities listed above as well as offering a small town feel with larger city amenities. There are also numerous cities and destinations all within a day’s drive including the famous Sleeping Bear Sand Dunes national lakeshore. The western side of Michigan is also known for its wine production and Traverse City has over 30 wineries within a 30 minute drive. Throughout the summer, Traverse City hosts many events, the most famous being the National Cherry Festival as Michigan is the largest tart cherry producing state in the US.
|
|
null | false
| null |
How Fast Does An Average Person Run?
|
Males run at an average speed of 8 mph, and women run at an average speed of 6.5 mph.
People running for their lives, not for recreational purposes, can run at an average speed of 12 mph. Athletes and gym enthusiasts tend to be faster than the average person and may run at a speed as high as 14 to 17 mph. The numbers change when referring to endurance and long-distance trails. Sprinters running a short distance average 10 mph. However, it’s hard to maintain that momentum for a long distance. When using data on the average running speed for five kilometers, males have an average running speed of 5.9 mph, and women have an average running speed of 5 mph.
|
|
null | false
| null |
The repertoire of Bharatanatyam, like all major classical Indian dance forms, follows the three categories of performance in the Natya Shastra. These are Nritta (Nirutham), Nritya (Niruthiyam) and Natya (Natyam).
The Nritta performance is abstract, fast and rhythmic aspect of the dance. The viewer is presented with pure movement in Bharatanatyam, wherein the emphasis is the beauty in motion, form, speed, range and pattern. This part of the repertoire has no interpretative aspect, no telling of story. It is a technical performance, and aims to engage the senses (prakriti) of the audience.
The Nritya is slower and expressive aspect of the dance that attempts to communicate feelings, storyline particularly with spiritual themes in Hindu dance traditions. In a nritya, the dance-acting expands to include silent expression of words through gestures and body motion set to musical notes. The actor articulates a legend or a spiritual message. This part of a Bharatanatyam repertoire is more than sensory enjoyment, it aims to engage the emotions and mind of the viewer.
The Natyam is a play, typically a team performance, but can be acted out by a solo performer where the dancer uses certain standardized body movements to indicate a new character in the underlying story. A Natya incorporates the elements of a Nritya.
|
Please give me a short description of each of the three categories of performance that Bharatanatyam follows.
|
1. Nritta: This is the pure movement aspect of the dance of Bharatanatyam. The dancer showcases movements in varying form, speed, range and levels.
2. Nritya: This is the expressive aspect of the dance, wherein the dancer tells a story, usually of a spiritual theme of Hindu tradition, through both facial expressions and movements.
3. Natyam: This is the theatrical aspect of the dance, where the dancer incorporates the elements of Nritya, to enact different characters of a play. This can be performed either solo or as a team.
|
null | false
| 232
|
Laughter is a special, higher-order function that only humans possess. In the analysis of laughter, as Wikipedia says, “Laughter is thought to be a shift of composition (schema)", and laughter frequently occurs when there is a change from a composition of receiver. However, the viewpoint of laughter differs greatly depending on the position of the receiver. Therefore, the quantitative measurement of laughter is very difficult. Image Ogiri on web services such as "Bokete" BIBREF0 have recently appeared, where users post funny captions for thematic images and the captions are evaluated in an SNS-like environment. Users compete to obtain the greatest number of “stars”. Although quantification of laughter is considered to be a very difficult task, the correspondence between evaluations and images on Bokete allows us to treat laughter quantitatively. Image captioning is an active topic in computer vision, and we believe that humorous image captioning can be realized. The main contributions of the present paper are as follows:
BoketeDB
In the experimental section, we compare the proposed method based on Funny Score and BoketeDB pre-trained parameters with a baseline provided by MS COCO Pre-trained CNN+LSTM. We also compare the results of the NJM with funny captions provided by humans. In an evaluation by humans, the results provided by the proposed method were ranked lower than those provided by humans (22.59% vs. 67.99%) but were ranked higher than the baseline (9.41%). Finally, we show the generated funny captions for several images.
The main contributions of the present paper are as follows: BoketeDB In the experimental section, we compare the proposed method based on Funny Score and BoketeDB pre-trained parameters with a baseline provided by MS COCO Pre-trained CNN+LSTM.
|
What did their method base on?
|
Funny Score and BoketeDB pre-trained parameters。
|
null | false
| null |
What are the most airplanes to learn how to fly?
|
The most popular single engine propeller plane in the United States is Cessna 172 model, where it gives pilots the ability to complete visual flight rules (VFR) flying as well as Instrument flight rules (IFR). Cessnas tend to be reliable and safe planes with proper maintenance.
|
|
null | false
| null |
in which country could you spend a Kwanza
|
Angola
|
|
null | false
| null |
What are some tips that kids could use to learn the lines of a musical school play? Write them as if they are narrated by an elementary school student.
|
One good method is to practice your lines of the play with other people, who might not necessarily be in the play (for example, practice with your parents!). It's also good to review your lines and repeat them by heart a few minutes later. Then over time try to repeat them without looking at the play book. One thing to avoid is to only read the lines but not say them out loud. Saying the lines out loud is really important for remembering them and practicing your acting. And above all, try to have fun!
|
|
null | false
| null |
Who was General Sir Henry Clinton
|
General Sir Henry Clinton, KB (16 April 1730 – 23 December 1795) was a British Army officer and politician who sat in the House of Commons between 1772 and 1795. He is best known for his service as a general during the American War of Independence. First arriving in Boston in May 1775, from 1778 to 1782 he was the British Commander-in-Chief in North America. In addition to his military service, due to the influence of his cousin Henry Pelham-Clinton, 2nd Duke of Newcastle, he was a Member of Parliament for many years. Late in life he was named Governor of Gibraltar, but died before assuming the post.
|
|
null | false
| 438
|
x s * ← Rapid optimization of x s * IG-score using the oracle 6:
x s n ← [x s * , x s n−1 ] Combine optimized samples with existing samples 7: end for DARK (Algorithm 1) tackles the primary limitation on effectively learning models, which is having suitable sequences to train and test against. The first step is to take some initial set of samples x s 0 to train an initial sampling model p γ 1 (x s 0 ) with parameters γ 1 ∈ Γ. Here we use the aforementioned seed samples, but this could be random sequences or sequences from DARK 0 -Grad, an approach we developed that backpropagates the negative IG-score to generate samples (Appendix A.2), in a fashion similar to. We use the seed samples out of convenience as they have the highest score, and we reasoned that they would require less iterations by comparison. The sampling models are all generative sequence models that can be sampled. After training p γ 1 (x s 0 ), it is used to generate a large set of samples x s * . Instead of then training a new sampling model on x s * , each sequence in x s * is quickly optimized to improve the IG-score. This is done by a quick greedy hill-climb on the sequence for a set number of steps (Line 5 in Algorithm 1), which we refer to as refinement. In each step, a random position is mutated. If it improves the IG-score then it is kept, otherwise it is discarded. We use 3000 steps, a fraction of the steps done in DARK 0 , as this was found to improve the average IG-score of x s * to approximately equal to the IG-score of x s 0 . For simplicity, this is fixed for all steps. The result of optimizing x s * is then combined with x s 0 to make x s 1 . In the second iteration, x s 1 is used to train a new sampling model p γ 2 (x s 1 ). After some number of iterations N , the final sampling model is viewed as a strong prior over x d , also making it a powerful generative model that can be used for de novo design tasks. Here we perform 3 iterations with training set sizes of 15K, 100K, and 500K, where the difference is sampled and refined between iterations. These sample sizes are arbitrary and they were entirely dictated by computational resources available at the time. Three iterations were performed as we believed it sufficient to demonstrate DARK's effectiveness. We note that the iterative aspect of DARK shares similarities to methods like Estimation of Distribution Algorithms (EDAs), a common approach in model based optimization (MBO). Viewed through this lens, it suggests a number of ways that DARK can be adapted and potentially improved on, which we leave for future work.
In DARK, we use deep autoregressive language models for learning protein sequence distributions. Specifically, we use a standard Transformer decoder architecture which has been used extensively to train state-of-the-art language models capable of generating high-quality synthetic sequence samples. In the first iteration of DARK we use a small decoder model, termed DARK 1 , with 4 layers, 4 heads, and a feed-forward size of H = 128. To affirm the choice of self-attention based architecture, we compare it to a 1layer LSTM with a hidden dimension H = 128. We also include results for a variant of DARK 1 referred to as DARK 1 -Adversarial, that uses DMPfold2 as an adversarial regularizer during training (Appendix A.3). After the first iteration, the second and third iteration models, DARK 2 and DARK 3 respectively, use a decoder with 12 layers, 12 heads, and H = 768. The parameters are increased to coincide with the order of magnitude change in the size of the training set from DARK 1 to DARK 2 .
To provide a even comparison to models trained on natural sequences, we train a model, termed UF50, on 50M natural sequences from the UniRef50 sequence database (The UniProt Consortium, 2021) using the same architecture as DARK 3 (Appendix A.5). For all trained models we perform early stopping and a small amount of hyper-parameter optimization on the validation set.
Quantitatively measuring sample quality We define high-quality samples as those that are diverse, have sequences unlike any natural sequence, and are predicted to have a stable and ordered structure. In de novo design, the most important test of a candidate sequence is that it is confidently predicted to have a stable and ordered atomistic structure. We provide a direct measure of both confidence and order jointly using AlphaFold's pLDDT score, being its confidence metric, which ranges from 0 to 100. This is a measure of confidence, and an indirect measure of order. It was found that low scores (pLDDT< 50) strongly suggest disorder and vice versa. Thus, we use the proportion of samples with a pLDDT> 70 as a measure of high quality for the predicted structures, which we refer to as Good+ pLDDT, indicating "good or better" quality according to criteria in. When discussing results we use 'ordered' to mean 'confidently predicted and ordered'. Using AlphaFold also provides a check to ensure that the model has not learned to produce adversarial examples against the original DMPfold2 oracle. We also measure the diversity of sequences and predicted structures using the estimated number of clusters in samples of sequences, and samples of predicted structures (See Appendix B.7 & B.8). Measuring the structural generality of DARK in a strict setting We take two key approaches to providing a stringent measure of how well DARK generalizes with regards to structure. Both are enabled by using AlphaFold to generate all-atom structure predictions for all 100,000 sequences contained in the DARK 2 training set, which covers the 15,000 seed samples and the 85,000 samples from the first iteration of DARK. We then assign each sample's structure to a topology. This process is the repeated for the 950 sequence validation set; we ignore the 950 sequence test set for what is described here. The first approach we take is to carefully construct a strict training and test set split based on structure, from the 100K samples (See Appendix B.2 for exhaustive details). This has been recently suggested for machine learning-based protein design studies and is considered a gold-standard approach in protein structure prediction. Our strict test set contains no sequences with overlapping topologies in either the training set or validation set. We also remove any sequences from the validation and training set that are detected to be similar to those in the test set with the MMseqs2 search tool . We train and evaluate a model on these sets, termed DARK 2 -STRICT. In the limit of working with synthetic, we believe this constitutes a stringent test of generality. The second approach we take is to compare the number of unique topologies represented by the 15,000 seed samples and those represented by the 85,000 sequences from the first iteration of DARK to show that DARK iterations generate sequences with unseen structures.
Table 1: Perplexities (PPLs), IG-scores, & Sampling Speed for protein language models. Lower PPL is better and higher IG-score is better. IG-scores are calculated from 1000 sequence samples except the 15K DARK0-Seed Samples. IG-scores for refined (Ref) samples are included for DARK models. Sampling Speed is for a single example (See Appendix A.6 for further details).
|
You present your work as a faster alternative to the trDesign approach. Why is there no computational times reported? How long does it take to your approach to match the perplexity or IG-Score of the seed sample generated by trDesign?
|
Thank you, we agree this should have been a part of the text. We have added significant timing information throughout and we believe it now forms a part of our argument. DARK3 without refinement can produce samples with similar IG-scores to our version of trDesign (DARK0 ) in 0.05 seconds. In contrast, DARK0 takes over 11 minutes (Table 1).
|
null | false
| null |
Classify these dishes into the cuisines that they belong to.
Panipuri, Risotto, Butter Chicken, Lasagna, Naan, Ravioli, Jalebi
|
Indian - Panipuri, Butter Chicken, Naan, Jalebi
Italian - Risotto, Lasagna, Ravioli
|
|
null | false
| null |
Write a letter to my landlord, Dave, asking them to hire a plumber to repair a leaking faucet.
|
Hi Dave, I'm reaching out about our leaking sink faucet. The sink has been leaking for about a week now, and I've tried tightening the faucet and making sure that the faucet is fully closed, but it still seems to be leaking. It is a pretty constant drip, and is filling a cup up every hour. I think it probably needs to be seen by a plumber who can decide if it is fixable or needs to be replaced. Could you call a plumber and ask them to step by?
|
|
null | false
| null |
Why am I so tired?
|
If you're frequently tired without obvious reasons, such as lack of sleep, you should consult a doctor. If you suffer from regular sleep deprivation, you should also consult a doctor.
|
|
null | false
| null |
List some steps I'll need to take in order to buy a car
|
- Search the web for car models that interest you
- Go to the dealership for a test drive
- Think about whether to finance, buy or lease your car
- Order a car from the dealership
- Enjoy your new car
|
|
1807.08204
| false
| null |
Unification Module. In backward chaining, unification between two atoms is used for checking whether they can represent the same structure. In discrete unification, non-variable symbols are checked for equality, and the proof fails if the symbols differ. In NTP, rather than comparing symbols, their embedding representations are compared by means of a RBF kernel. This allows matching different symbols with similar semantics, such as matching relations like ${grandFatherOf}$ and ${grandpaOf}$ . Given a proof state $= (_, _)$ , where $_$ and $_$ denote a substitution set and a proof score, respectively, unification is computed as follows:
The resulting proof score of $g$ is given by:
$$ \begin{aligned} \max _{f \in \mathcal {K}} & \; {unify}_(g, [f_{p}, f_{s}, f_{o}], (\emptyset , )) \\ & = \max _{f \in \mathcal {K}} \; \min \big \lbrace , \operatorname{k}(_{\scriptsize {grandpaOf}:}, _{f_{p}:}),\\ &\qquad \qquad \qquad \operatorname{k}(_{{abe}:}, _{f_{s}:}), \operatorname{k}(_{{bart}:}, _{f_{o}:}) \big \rbrace , \end{aligned}$$ (Eq. 3)
where $f \triangleq [f_{p}, f_{s}, f_{o}]$ is a fact in $\mathcal {K}$ denoting a relationship of type $f_{p}$ between $f_{s}$ and $f_{o}$ , $_{s:}$ is the embedding representation of a symbol $s$ , $$ denotes the initial proof score, and $\operatorname{k}({}\cdot {}, {}\cdot {})$ denotes the RBF kernel. Note that the maximum proof score is given by the fact $f \in \mathcal {K}$ that maximises the similarity between its components and the goal $\mathcal {K}$0 : solving the maximisation problem in eq:inference can be equivalently stated as a nearest neighbour search problem. In this work, we use ANNS during the forward pass for considering only the most promising proof paths during the construction of the neural network.
Given a proof state $= (_, _)$ , where $_$ and $_$ denote a substitution set and a proof score, respectively, unification is computed as follows:
The resulting proof score of $g$ is given by:
$$ \begin{aligned} \max _{f \in \mathcal {K}} & \; {unify}_(g, [f_{p}, f_{s}, f_{o}], (\emptyset , )) \\ & = \max _{f \in \mathcal {K}} \; \min \big \lbrace , \operatorname{k}(_{\scriptsize {grandpaOf}:}, _{f_{p}:}),\\ &\qquad \qquad \qquad \operatorname{k}(_{{abe}:}, _{f_{s}:}), \operatorname{k}(_{{bart}:}, _{f_{o}:}) \big \rbrace , \end{aligned}$$ (Eq. 3)
where $f \triangleq [f_{p}, f_{s}, f_{o}]$ is a fact in $\mathcal {K}$ denoting a relationship of type $f_{p}$ between $f_{s}$ and $f_{o}$ , $_{s:}$ is the embedding representation of a symbol $s$ , $$ denotes the initial proof score, and $\operatorname{k}({}\cdot {}, {}\cdot {})$ denotes the RBF kernel.
|
How are proof scores calculated?
|
The answers are shown as follows:
* '= ( , { ll k(h:, g:) if hV, gV
1 otherwise } )
where $_{h:}$ and $_{g:}$ denote the embedding representations of $h$ and $g$ , respectively.
|
null | false
| null |
What tools do people use to communicate in 2023?
|
There are many tools people use to communicate in 2023. Email, phone, text message, and direct message on social media platforms are some of the most common; however, people still write letters and send post cards to one another at times as they have for generations. Another form of communication is in-person communication where groups of people will get together and communicate on a given topic. This could be at work in a meeting or a group of friends getting together for dinner to catch up. This form of communication requires no other tools than one's own voice.
|
|
null | false
| 328
|
Neural network based models have been widely exploited with the prosperities of Deep Learning BIBREF0 and achieved inspiring performances on many NLP tasks, such as text classification BIBREF1 , BIBREF2 , semantic matching BIBREF3 , BIBREF4 and machine translation BIBREF5 . These models are robust at feature engineering and can represent words, sentences and documents as fix-length vectors, which contain rich semantic information and are ideal for subsequent NLP tasks.
One formidable constraint of deep neural networks (DNN) is their strong reliance on large amounts of annotated corpus due to substantial parameters to train. A DNN trained on limited data is prone to overfitting and incapable to generalize well. However, constructions of large-scale high-quality labeled datasets are extremely labor-intensive. To solve the problem, these models usually employ a pre-trained lookup table, also known as Word Embedding BIBREF6 , to map words into vectors with semantic implications. However, this method just introduces extra knowledge and does not directly optimize the targeted task. The problem of insufficient annotated resources is not solved either.
Multi-task learning leverages potential correlations among related tasks to extract common features, increase corpus size implicitly and yield classification improvements. Inspired by BIBREF7 , there are a large literature dedicated for multi-task learning with neural network based models BIBREF8 , BIBREF9 , BIBREF10 , BIBREF11 . These models basically share some lower layers to capture common features and further feed them to subsequent task-specific layers, which can be classified into three types:
In this paper, we propose a generalized multi-task learning architecture with four types of recurrent neural layers for text classification. The architecture focuses on Type-III, which involves more complicated interactions but has not been researched yet. All the related tasks are jointly integrated into a single system and samples from different tasks are trained in parallel. In our model, every two tasks can directly interact with each other and selectively absorb useful information, or communicate indirectly via a shared intermediate layer. We also design a global memory storage to share common features and collect interactions among all tasks.
We conduct extensive experiments on five benchmark datasets for text classification. Compared to learning separately, jointly learning multiple relative tasks in our model demonstrate significant performance gains for each task.
Our contributions are three-folds:
In this paper, we propose a generalized multi-task learning architecture with four types of recurrent neural layers for text classification.
|
How many types of recurrent neural layers in the authors' multi-task learning architecture?
|
Four.
|
null | false
| null |
Who gave the UN the land in NY to build their HQ
|
John D Rockerfeller
|
|
null | false
| null |
What is the Taylor rule?
|
The Taylor rule is a monetary policy targeting rule. The rule was proposed in 1992 by American economist John B. Taylor for central banks to use to stabilize economic activity by appropriately setting short-term interest rates. The rule considers the federal funds rate, the price level and changes in real income. The Taylor rule computes the optimal federal funds rate based on the gap between the desired (targeted) inflation rate and the actual inflation rate; and the output gap between the actual and natural output level. According to Taylor, monetary policy is stabilizing when the nominal interest rate is higher/lower than the increase/decrease in inflation. Thus the Taylor rule prescribes a relatively high interest rate when actual inflation is higher than the inflation target.
|
|
null | false
| null |
Hudson was born in Quincy, Massachusetts, in 1819, and attended the town school in Concord, Massachusetts. When he was 17, he moved to New York City, where his brothers had opened "Hudson's News Room". In 1836, he there met James Gordon Bennett Sr., who had founded the Herald in 1835, and soon went to work for him, becoming the third full-time employee of the paper.
|
How did Frederic Hudson start working at the Herald?
|
While working at "Hudson's News Room" in 1836, Hudson met James Gordon Bennett Sr., who had founded the Herald just one year prior in 1835. The 17 year old Hudson went to work for the Herald shortly thereafter, becoming just the third full-time employee of the Herald.
|
null | false
| 485
|
Consider a task Y , which can correspond to a variety of objectives. For instance, it could be an auxiliary task of predicting the next state when interacting with an RL environment, or simply a classification task on a vision benchmark. We are interested in training a model that can do well on task Y , where the model consists of two components, a backbone representation f θ and a predictor/classifier h φ which is attached over the representation. In a lot of tasks that involve pixel input, spurious correlations can arise due to various reasons such as lack of non-iid data, irrelevant information, confounders etc. that can result in poor generalization at test time. An attractive property for better generalization has been the idea of sparse representations, in that the mutual information of any two dimensions of the representation must be low. However, in order to avoid learning trivial solutions, the representation as a whole should encode enough information about the downstream task as well. A combination of both of these objectives then leads to a model with better generalization capability. In this paper, we balance the above two objectives using the idea of model invariance (MODINV). MODINV deploys multiple predictors over a single representation, while training each predictor independently, each capturing different spurious correlations. The common representation acts as a implicit invariance loss which ensures that only the optimal representation remains at convergence. Intuitively, each predictor can be looked at as an augmented version of the optimal model, where the augmentation is over the model space and refers to particular spurious correlations arising in the model that differ it from the optimal model. Note that each predictor head is the same in architecture, and we only diversify the learning procedure of each head (through different initialization, independent training, different learning rates). This is so that we eventually converge to an optimal representation for a particular predictor/classifier and not for all predictor/classifier families, which is a much more hard to optimize in practise. For all experiments, we use a random routing to decide which data sample is used to train which predictor head. Furthermore, the learning rates of the different predictor heads are chosen intuitively, i.e. if the base rate is 3e-3 then we choose one slightly higher rate (5e-3) and one slightly smaller rate (1e-3) for K=3 predictors.
Furthermore, the learning rates of the different predictor heads are chosen intuitively, i.e. if the base rate is 3e-3 then we choose one slightly higher rate (5e-3) and one slightly smaller rate (1e-3) for K=3 predictors.
|
Can the authors comment on how the learning rates and initializations for different predictor heads in all settings were chosen?
|
For now, this is done simply by intuition, i.e. if the base lr is 3e-3 then we choose one lr slightly higher, i.e. 5e-3 and one lr slightly smaller, i.e. 1e-3. We believe in future work there is a lot of scope in developing better engineering solutions for such choices. We have noted this in the updated version.
|
null | false
| null |
On 11 April 2001, the Australian and American Samoan national association football teams played each other in an Oceanian qualifying match for the 2002 FIFA World Cup. The match was played at the International Sports Stadium in Coffs Harbour, Australia. Australia set a world record for the largest victory in an international football match, winning the game 31–0. Australia's Archie Thompson also broke the record for most goals scored by a player in an international match by scoring 13 goals. David Zdrilic, the scorer of eight goals in the match, scored the second-highest number of goals in an international match since World War I.
|
How many goals did Archie Thompson's team mates score?
|
Since Archie Thompson scored 13 goals, his teammates scored 18 goals
|
null | false
| null |
Where is John Rahm from?
|
Barrika, Spain
|
|
null | false
| 283
|
Scripts were developed as a means of representing stereotypical event sequences and interactions in narratives. The benefits of scripts for encoding common sense knowledge, filling in gaps in a story, resolving ambiguous references, and answering comprehension questions have been amply demonstrated in the early work in natural language understanding BIBREF0 . The earliest attempts to learn scripts were based on explanation-based learning, which can be characterized as example-guided deduction from first principles BIBREF1 , BIBREF2 . While this approach is successful in generalizing from a small number of examples, it requires a strong domain theory, which limits its applicability.
More recently, some new graph-based algorithms for inducing script-like structures from text have emerged. “Narrative Chains” is a narrative model similar to Scripts BIBREF3 . Each Narrative Chain is a directed graph indicating the most frequent temporal relationship between the events in the chain. Narrative Chains are learned by a novel application of pairwise mutual information and temporal relation learning. Another graph learning approach employs Multiple Sequence Alignment in conjunction with a semantic similarity function to cluster sequences of event descriptions into a directed graph BIBREF4 . More recently still, graphical models have been proposed for representing script-like knowledge, but these lack the temporal component that is central to this paper and to the early script work. These models instead focus on learning bags of related events BIBREF5 , BIBREF6 .
While the above approches demonstrate the learnability of script-like knowledge, they do not offer a probabilistic framework to reason robustly under uncertainty taking into account the temporal order of events. In this paper we present the first formal representation of scripts as Hidden Markov Models (HMMs), which support robust inference and effective learning algorithms. The states of the HMM correspond to event types in scripts, such as entering a restaurant or opening a door. Observations correspond to natural language sentences that describe the event instances that occur in the story, e.g., “John went to Starbucks. He came back after ten minutes.” The standard inference algorithms, such as the Forward-Backward algorithm, are able to answer questions about the hidden states given the observed sentences, for example, “What did John do in Starbucks?”
There are two complications that need to be dealt with to adapt HMMs to model narrative scripts. First, both the set of states, i.e., event types, and the set of observations are not pre-specified but are to be learned from data. We assume that the set of possible observations and the set of event types to be bounded but unknown. We employ the clustering algorithm proposed in BIBREF4 to reduce the natural language sentences, i.e., event descriptions, to a small set of observations and states based on their Wordnet similarity.
The second complication of narrative texts is that many events may be omitted either in the narration or by the event extraction process. More importantly, there is no indication of a time lapse or a gap in the story, so the standard forward-backward algorithm does not apply. To account for this, we allow the states to skip generating observations with some probability. This kind of HMMs, with insertions and gaps, have been considered previously in speech processing BIBREF7 and in computational biology BIBREF8 . We refine these models by allowing state-dependent missingness, without introducing additional “insert states” or “delete states” as in BIBREF8 . In this paper, we restrict our attention to the so-called “Left-to-Right HMMs” which have acyclic graphical structure with possible self-loops, as they support more efficient inference algorithms than general HMMs and suffice to model most of the natural scripts. We consider the problem of learning the structure and parameters of scripts in the form of HMMs from sequences of natural language sentences. Our solution to script learning is a novel bottom-up method for structure learning, called SEM-HMM, which is inspired by Bayesian Model Merging (BMM) BIBREF9 and Structural Expectation Maximization (SEM) BIBREF10 . It starts with a fully enumerated HMM representation of the event sequences and incrementally merges states and deletes edges to improve the posterior probability of the structure and the parameters given the data. We compare our approach to several informed baselines on many natural datasets and show its superior performance. We believe our work represents the first formalization of scripts that supports probabilistic inference, and paves the way for robust understanding of natural language texts.
In this paper we present the first formal representation of scripts as Hidden Markov Models (HMMs), which support robust inference and effective learning algorithms.
|
How to present the first formal representation of scripts in this paper?
|
The authors present the first formal representation of scripts as Hidden Markov Models (HMMs).
|
null | false
| null |
Why wasn't the Treaty of Versailles effective?
|
The Treaty of Versailles was created almost vindictively by the West since they were seeking justice from Germany. This caused major social and economic problems in Germany as they were not only forced to take blame and return territories, but also pay large reparations to Britain and France. These harsh terms inevitably led to World War II, as bitter and resentful German citizens were easily swayed by Hitler's promises of revenge. The Allies attempt to restrict Germany’s power also failed, as they used a strategy of appeasement when Hitler tried to expand, rendering the rules of the treaty useless.
|
|
1909.12642
| false
| null |
The performance of our models across different languages for sub-task A are shown in table TABREF19. Our model got the first position in the German sub-task with a macro F1 score of 0.62. The results of sub-task B and sub-task C is shown in table TABREF20 and TABREF21 respectively.
Our model got the first position in the German sub-task with a macro F1 score of 0.62.
|
What is the performance of the model for the German sub-task A?
|
The answers are shown as follows:
* macro F1 score of 0.62
|
null | false
| null |
Hindi Diwas is observed every year on which date?
|
September 14
|
|
null | false
| 249
|
In this section, we present the results of our experiments on the automatic detection of cyberbullying-related posts in an English (EN) and Dutch (NL) corpus of ASKfm posts. Ten-fold cross-validation was performed in exhaustive grid-search over different feature type and hyperparameter combinations (see Section SECREF4 ). The unoptimised word INLINEFORM0 -gram-based classifier and keyword-matching system serve as baselines for comparison. Precision, Recall and F INLINEFORM1 performance metrics were calculated on the positive class (i.e., `binary averaging'). We also report Area Under the ROC curve (AUC) scores, a performance metric that is more robust to data imbalance than precision, recall and micro-averaged F-score BIBREF74 .
Table TABREF45 gives us an indication of which feature type combinations score best and hence contribute most to this task. A total of 31 feature type combinations, each with 28 different hyperparameter sets have been tested. Table TABREF45 shows the results for the three best scoring systems by included feature types with optimised hyperparameters. The maximum attained F INLINEFORM0 -score in cross-validation is 64.26% for English and 61.20% for Dutch and shows that the classifier benefits from a variety of feature types. The results on the holdout test set show that the trained systems generalise well on unseen data, indicating little under- or overfitting. The simple keyword-matching baseline system has the lowest performance for both languages even though it obtains high recall for English, suggesting that profane language characterises many cyberbullying-related posts. Feature group and hyperparameter optimisation provides a considerable performance increase over the unoptimised word INLINEFORM1 -gram baseline system. The top-scoring systems for each language do not differ a lot in performance, except the best system for Dutch, which trades recall for precision when compared to the runner-ups.
Table TABREF47 presents the scores of the (hyperparameter-optimised) single feature type systems, to gain insight into the performance of these feature types when used individually. Analysis of the combined and single feature type sets reveals that word INLINEFORM0 -grams, character INLINEFORM1 -grams, and subjectivity lexicons prove to be strong features for this task. In effect, adding character INLINEFORM2 -grams always improved classification performance for both languages. They likely provide robustness to lexical variation in social media text, as compared to word INLINEFORM3 -grams. While subjectivity lexicons appear to be discriminative features, term lists perform badly on their own as well as in combinations for both languages. This shows once again (cf. profanity baseline) that cyberbullying detection requires more sophisticated information sources than profanity lists. Topic models seem to do badly for both languages on their own, but in combination, they improve Dutch performance consistently. A possible explanation for their varying performance in both languages would be that the topic models trained on the Dutch background corpus are of better quality than the English ones. In effect, a random selection of background corpus texts reveals that the English scrape contains more noisy data (i.e., low word-count posts and non-English posts) than the Dutch data.
A shallow qualitative analysis of the classification output provided insight into some of the classification mistakes.
Table TABREF52 gives an overview of the error rates per cyberbullying category of the best performing and baseline systems. This could give an indication of which types of bullying the current system has trouble classifying. All categories are always considered positive for cyberbullying (i.e., the error rate equals the false negative rate), except for Sexual and Insult which can also be negative (in case of harmless sexual talk and `socially acceptable' insulting language like `hi bitches, in for a movie?' the corresponding category was indicated, but the post itself was not annotated as cyberbullying) and Not cyberbullying, which is always negative. Error rates often being lowest for the profanity baseline confirms that it performs particularly well in terms of recall (at the expense of precision, see Table TABREF47 ) When looking at the best system for both languages, we see that Defense is the hardest category to correctly classify. This should not be a surprise as the category comprises defensive posts from bystanders and victims, which contain less aggressive language than cyberbullying attacks and are often shorter in length than the latter. Assertive defensive posts (i.e., a subcategory of Defense) that attack the bully) are, however, more often correctly classified. There are not enough instances of Encouragement for either language in the holdout to be representative. In both languages, threats, curses and incidences of sexual harassment are most easily recognisable, showing (far) lower error rates than the categories Defamation, Defense, Encouragements to the harasser, and Insult.
Qualitative error analysis of the English and Dutch predictions reveals that false positives often contain aggressive language directed at a second person, often denoting personal flaws or containing sexual and profanity words. We see that misclassifications are often short posts containing just a few words and that false negatives often lack explicit verbal signs of cyberbullying (e.g. insulting or profane words) or are ironic (examples 2 and 3). Additionally, we see that cyberbullying posts containing misspellings or grammatical errors and incomplete words are also hard to recognise as such (examples 4 and 5). The Dutch and English data are overall similar with respect to qualitative properties of classification errors.
In short, the experiments show that our classifier clearly outperforms both a keyword-based and word INLINEFORM0 -gram baseline. However, analysis of the classifier output reveals that false negatives often lack explicit clues that cyberbullying is going on, indicating that our system might benefit from irony recognition and integrating world knowledge to capture such implicit realisations of cyberbullying.
Given that we present the first elaborate research on detecting signals of cyberbullying regardless of the author role instead of bully posts alone, crude comparison with the state of the art would be irrelevant. We observe, however, that our classifier obtains competitive results compared to BIBREF32 , BIBREF33 , BIBREF35 , BIBREF34 , BIBREF37 .
single feature type sets reveals that word n-grams, character n-grams, and subjectivity lexicons prove to be strong features for this task.
|
What are the strong features of this task？
|
Word n-grams, character n-grams, and subjectivity lexicons.
|
1602.01208
| false
| null |
The proposed method can learn words related to places from the utterances of sentences. We use an unsupervised word segmentation method latticelm that can directly segment words from the lattices of the speech recognition results of the uttered sentences BIBREF22 . The lattice can represent to a compact the set of more promising hypotheses of a speech recognition result, such as N-best, in a directed graph format. Unsupervised word segmentation using the lattices of syllable recognition is expected to be able to reduce the variability and errors in phonemes as compared to NPYLM BIBREF13 , i.e., word segmentation using the 1-best speech recognition results.
We use an unsupervised word segmentation method latticelm that can directly segment words from the lattices of the speech recognition results of the uttered sentences BIBREF22 .
|
Which method do they use for word segmentation?
|
The answers are shown as follows:
* unsupervised word segmentation method latticelm
|
null | false
| 130
|
Secondly, we fine-tune multilingual Bert (M-Bert) on the task BIBREF5 which has been pretrained jointly in 104 languages and has established itself as a state of the art for various multilingual tasks BIBREF18, BIBREF19. Within the field of stance detection, Bert can outperform both feature-based and other neural approaches in a monolingual English setting BIBREF10.
In the context of Bert we interpret the x-stance task as sequence pair classification inspired by natural language inference tasks BIBREF20. We follow the procedure outlined by BIBREF5 for such tasks. We designate the question as segment A and the comment as segment B. The two segments are separated with the special token [SEP], and the special token [CLS] is prepended to the sequence. The final hidden state corresponding to [CLS] is then classified by a linear layer.
We fine-tune the full model with a cross-entropy loss, using the AllenNLP library BIBREF21 as a basis for our implementation.
We upsampled the `favor' class so that the two classes are balanced when summing over all questions and topics. A maximum sequence length of 512 subwords and a batch size of 16 was chosen for all training runs. We then performed a grid search within the following range of hyperparameters based on the validation accuracy:
Learning rate: 5e-5, 3e-5, 2e-5
Number of epochs: 3, 4
The grid search was repeated independently for every variant that we tested. Furthermore, the standard recommendations for fine-tuning Bert were used: Adam with ${\beta }_1=0.9$ and ${\beta }_2=0.999$; an L2 weight decay of $0.01$; a learning rate warmup over the first 10% of the steps; and a linear decay of the learning rate. A dropout probability of 0.1 was set on all layers.
Table TABREF36 shows the results for the cross-lingual setting. M-Bert performs consistently better than the majority class baselines. Even the zero-short performance in Italian, while significantly lower than the supervised scores, is much better than the target-wise majority class baseline.
Results for the cross-target setting are given in Table TABREF37. Similar to the cross-lingual setting, M-Bert performs worse in a cross-target setting but easily surpasses the majority class baselines. Furthermore, the cross-question score of M-Bert is slightly lower than the cross-topic score.
Secondly, we fine-tune multilingual Bert (M-Bert) on the task BIBREF5 which has been pretrained jointly in 104 languages and has established itself as a state of the art for various multilingual tasks BIBREF18, BIBREF19. Within the field of stance detection, Bert can outperform both feature-based and other neural approaches in a monolingual English setting BIBREF10.
In the context of Bert we interpret the x-stance task as sequence pair classification inspired by natural language inference tasks BIBREF20. We follow the procedure outlined by BIBREF5 for such tasks. We designate the question as segment A and the comment as segment B. The two segments are separated with the special token [SEP], and the special token [CLS] is prepended to the sequence. The final hidden state corresponding to [CLS] is then classified by a linear layer.
We fine-tune the full model with a cross-entropy loss, using the AllenNLP library BIBREF21 as a basis for our implementation.
We upsampled the `favor' class so that the two classes are balanced when summing over all questions and topics. A maximum sequence length of 512 subwords and a batch size of 16 was chosen for all training runs. We then performed a grid search within the following range of hyperparameters based on the validation accuracy:
Learning rate: 5e-5, 3e-5, 2e-5
Number of epochs: 3, 4
The grid search was repeated independently for every variant that we tested. Furthermore, the standard recommendations for fine-tuning Bert were used: Adam with ${\beta }_1=0.9$ and ${\beta }_2=0.999$; an L2 weight decay of $0.01$; a learning rate warmup over the first 10% of the steps; and a linear decay of the learning rate. A dropout probability of 0.1 was set on all layers.
Table TABREF36 shows the results for the cross-lingual setting. M-Bert performs consistently better than the majority class baselines. Even the zero-short performance in Italian, while significantly lower than the supervised scores, is much better than the target-wise majority class baseline.
Results for the cross-target setting are given in Table TABREF37. Similar to the cross-lingual setting, M-Bert performs worse in a cross-target setting but easily surpasses the majority class baselines. Furthermore, the cross-question score of M-Bert is slightly lower than the cross-topic score.
Table 3 shows the results for the crosslingual setting. M-BERT performs consistently better than the previous baselines.****Results for the cross-target setting are given in Table 4. Similar to the cross-lingual setting, model performance drops in the cross-target setting, but M-BERT remains the strongest baseline and easily surpasses the majority class baselines.
|
Which baseline performs best according to the results?
|
M-BERT.
|
null | false
| null |
Spider-Man is a 2002 American superhero film based on the Marvel Comics superhero of the same name. Directed by Sam Raimi from a screenplay by David Koepp, it is the first installment in Raimi's Spider-Man trilogy, and stars Tobey Maguire as the titular character, alongside Willem Dafoe, Kirsten Dunst, James Franco, Cliff Robertson, and Rosemary Harris. The film chronicles Spider-Man's origin story and early superhero career. After being bitten by a genetically-altered spider, outcast teenager Peter Parker develops spider-like superhuman abilities and adopts a masked superhero identity to fight crime and injustice in New York City, facing the sinister Green Goblin (Dafoe) in the process.
Development on a live-action Spider-Man film began in the 1980s. Filmmakers Tobe Hooper, James Cameron, and Joseph Zito were all attached to direct the film at one point. However, the project would languish in development hell due to licensing and financial issues. After progress on the film stalled for nearly 25 years, it was licensed for a worldwide release by Columbia Pictures in 1999 after it acquired options from Metro-Goldwyn-Mayer (MGM) on all previous scripts developed by Cannon Films, Carolco, and New Cannon. Exercising its option on just two elements from the multi-script acquisition (a different screenplay was written by James Cameron, Ted Newsom, John Brancato, Barney Cohen, and Joseph Goldman), Sony hired Koepp to create a working screenplay (credited as Cameron's), and Koepp received sole credit in final billing. Directors Roland Emmerich, Ang Lee, Chris Columbus, Barry Sonnenfeld, Tim Burton, Michael Bay, Jan de Bont, M. Night Shyamalan, Tony Scott, and David Fincher were considered to direct the project before Raimi was hired as director in 2000. The Koepp script was rewritten by Scott Rosenberg during pre-production and received a dialogue polish from Alvin Sargent during production. Filming took place in Los Angeles and New York City from January to June 2001. Sony Pictures Imageworks handled the film's visual effects.
|
Based on the passage, which directors were considered for directing the Spider-Man movie apart from Raimi?
|
Roland Emmerich, Ang Lee, Chris Columbus, Barry Sonnenfeld, Tim Burton, Michael Bay, Jan de Bont, M. Night Shyamalan, Tony Scott, and David Fincher.
|
null | false
| null |
Why is "The quick brown fox jumps over the lazy dog" used so often to show fonts?
|
"The quick brown fox jumps over the lazy dog" contains all the letters in the English alphabet in a complete sentence. By showing that sentence in a particular font style, you see all the letters, and also see how it displays a particular sentence.
|
|
null | false
| 485
|
Consider a task Y , which can correspond to a variety of objectives. For instance, it could be an auxiliary task of predicting the next state when interacting with an RL environment, or simply a classification task on a vision benchmark. We are interested in training a model that can do well on task Y , where the model consists of two components, a backbone representation f θ and a predictor/classifier h φ which is attached over the representation. In a lot of tasks that involve pixel input, spurious correlations can arise due to various reasons such as lack of non-iid data, irrelevant information, confounders etc. that can result in poor generalization at test time. An attractive property for better generalization has been the idea of sparse representations, in that the mutual information of any two dimensions of the representation must be low. However, in order to avoid learning trivial solutions, the representation as a whole should encode enough information about the downstream task as well. A combination of both of these objectives then leads to a model with better generalization capability. In this paper, we balance the above two objectives using the idea of model invariance (MODINV). MODINV deploys multiple predictors over a single representation, while training each predictor independently, each capturing different spurious correlations. The common representation acts as a implicit invariance loss which ensures that only the optimal representation remains at convergence. Intuitively, each predictor can be looked at as an augmented version of the optimal model, where the augmentation is over the model space and refers to particular spurious correlations arising in the model that differ it from the optimal model. Note that each predictor head is the same in architecture, and we only diversify the learning procedure of each head (through different initialization, independent training, different learning rates). This is so that we eventually converge to an optimal representation for a particular predictor/classifier and not for all predictor/classifier families, which is a much more hard to optimize in practise. For all experiments, we use a random routing to decide which data sample is used to train which predictor head. Furthermore, the learning rates of the different predictor heads are chosen intuitively, i.e. if the base rate is 3e-3 then we choose one slightly higher rate (5e-3) and one slightly smaller rate (1e-3) for K=3 predictors.
For all experiments, we use a random routing to decide which data sample is used to train which predictor head.
|
The paper suffers from a few clarity issues. Highlighting them here. For instance, the ModInv structure implies that each predictor head is trained independently (a given sample only trains one head), implying that there is some underlying routing of the sample to a predictor head is going on. It’s unclear from the draft right now how this routing is performed. As in, given a training sample, how do the authors decide which head to feed the corresponding representation? If this routing is random, can the authors comment if they investigated other routing strategies?
|
Yes, this is done by randomly choosing a predictor head all throughout the paper. We did not experiment with more sophisticated schemes but do think it is an interesting direction to improve along.
|
null | false
| 37
|
Recent years have seen unprecedented progress for Natural Language Processing (NLP) on almost every NLP subtask. Even though low-resource settings have also been explored, this progress has overwhelmingly been observed in languages with significant data resources that can be leveraged to train deep neural networks. Low-resource languages still lag behind.
Endangered languages pose an additional challenge. The process of documenting an endangered language typically includes the creation of word lists, audio and video recordings, notes, or grammar fragments, with the created resources then stored in large online linguistics archives. This process is often hindered by the Transcription Bottleneck: the linguistic fieldworker and the language community may not have time to transcribe all of the recordings and may only transcribe segments that are linguistically salient for publication or culturally significant for the creation of community resources.
With this work we make publicly available a large corpus in Mapudungun, a language of the indigenous Mapuche people of southern Chile and western Argentina. We hope to ameliorate the resource gap and the transcription bottleneck in two ways. First, we are providing a larger data set than has previously been available, and second, we are providing baselines for NLP tasks (speech recognition, speech synthesis, and machine translation). In providing baselines and datasets splits, we hope to further facilitate research on low-resource NLP for this language through our data set. Research on low-resource speech recognition is particularly important in relieving the transcription bottleneck, while tackling the research challenges that speech synthesis and machine translation pose for such languages could lead to such systems being deployed to serve more under-represented communities.
We hope to ameliorate the resource gap and the transcription bottle - neck in two ways. First, we are providing a larger data set than has previously been available, and second, we are providing baselines for NLP tasks (speech recognition, speech synthesis, and machine translation).
|
How to ameliorate the resource gap and the transcription bottleneck by the authors?
|
First, the authors provide a larger data set than has previously been available, and second, the authors provide baselines for NLP tasks (speech recognition, speech synthesis, and machine translation).
|
null | false
| null |
What is a Fully paid share?
|
A share becomes fully paid when the company issuing it has received all the money due for the share.
|
|
null | false
| null |
What city is the capital of Alberta Canada?
|
The capital city of the province of Alberta in Canada is Edmonton.
|
|
null | false
| null |
Name some characters from the first "Saved by the Bell" series.
|
Zach, Kelly, Lisa, Slater, Screech, Mr. Belding
|
|
null | false
| null |
Taylor Swift (born December 13, 1989) is an American singer-songwriter. Her genre-spanning discography, songwriting and artistic reinventions have received critical praise and wide media coverage. Born in West Reading, Pennsylvania, Swift moved to Nashville at age 14 to become a country artist. She signed a songwriting deal with Sony/ATV Music Publishing in 2004 and a recording contract with Big Machine Records in 2005. Her 2006 self-titled debut album made her the first female country singer to write a U.S. platinum-certified album.
Swift's next albums, Fearless (2008) and Speak Now (2010), explored country pop. The former's "Love Story" and "You Belong with Me" were the first country songs to top the U.S. pop and all-genre airplay charts, respectively. She experimented with rock and electronic styles on Red (2012), which featured her first Billboard Hot 100 number-one song, "We Are Never Ever Getting Back Together", and eschewed her country image in her synth-pop album, 1989 (2014), supported by chart-topping songs "Shake It Off", "Blank Space", and "Bad Blood". Media scrutiny inspired the urban-flavored Reputation (2017) and its number-one single "Look What You Made Me Do".
Exiting Big Machine, Swift signed with Republic Records in 2018 and released her seventh studio album, Lover (2019), followed by the autobiographical documentary Miss Americana (2020). She ventured into indie folk and alternative rock in her 2020 albums Folklore and Evermore, whose singles "Cardigan" and "Willow" topped the Hot 100. Swift began re-recording her first six albums after a dispute over their masters, re-releasing two in 2021—Fearless (Taylor's Version) and Red (Taylor's Version). The latter's "All Too Well (10 Minute Version)" became the longest song to top the Hot 100. Her 2022 album Midnights and single "Anti-Hero" broke all-time streaming records. Swift has directed music videos and films, such as All Too Well: The Short Film (2021), and played supporting roles in others.
Having sold over 200 million records globally, Swift is one of the best-selling musicians in history. She is the most streamed woman on Spotify, and the only act to have five albums open with over one million copies sold in the US. Among her accolades are 12 Grammy Awards, including three Album of the Year wins; an Emmy Award; 40 American Music Awards; 29 Billboard Music Awards; three IFPI Global Recording Artist of the Year awards; and 92 Guinness World Records. Swift has been featured in rankings such as Rolling Stone's 100 Greatest Songwriters of All Time, Billboard's Greatest of All Time Artists, the Time 100 and Forbes Celebrity 100. Honored with titles such as Artist of the Decade and Woman of the Decade, Swift is an advocate for artists' rights and women's empowerment. Her music is credited with influencing a generation of singer-songwriters.
|
Please list all of Taylor Swift's albums, the year each one was released, and the genre of the album if it is mentioned, according to the paragraph I give.
|
Here is the list in the format of "album name: (year released) genre":
Taylor Swift: (2006) country
Fearless (2008): country pop
Speak Now (2010): country pop
Red (2012): rock and electronic
1989 (2014): synth-pop
Reputation (2017): urban
Lover (2019): not mentioned
folklore (2020): indie folk and alternative rock
evermore (2020): alternative rock
|
null | false
| null |
What is a casual hobby that would be fun to get into?
|
There are many great answers, and what is most fun for you depends on what your interests are. For instance, if you enjoy running and spending time with others, a good hobby that combines these interests could be ultimate frisbee. Ultimate frisbee involves a lot of running as well as teamwork and strategy to score points during games! It can also be played casually with friends without needing much experience beforehand.
|
|
null | false
| null |
How should I choose what cheese to buy?
|
There are many different types of cheese, so choosing what cheese to buy depends on personal preference, availability, and intended use. Washed-rind cheeses are often strongly flavored and runny so are best as a topping for bread or crackers. Harder cheeses can be eaten as a snack or grated on top of pasta.
|
|
null | false
| null |
What are some common characteristics of Maine Coon cats?
|
Maine Coon cats are a large breed that tends to have long-haired cats and exceptionally fluffy tails. They have distinct facial characteristics, including larger head, more pronounced chins, and tufted ear tips. They are also perceived to be loyal and more intelligent than average cats.
|
|
null | false
| 182
|
Personalized dialogue agents have been shown efficient in conducting human-like conversation. This progress has been catalyzed thanks to existing conversational dataset such as Persona-chat BIBREF0, BIBREF1. However, the training data are provided in a single language (e.g., English), and thus the resulting systems can perform conversations only in the training language. For wide, commercial dialogue systems are required to handle a large number of languages since the smart home devices market is increasingly international BIBREF2. Therefore, creating multilingual conversational benchmarks is essential, yet challenging since it is costly to perform human annotation of data in all languages.
A possible solution is to use translation systems before and after the model inference, a two-step translation from any language to English and from English to any language. This comes with three major problems: 1) amplification of translation errors since the current dialogue systems are far from perfect, especially with noisy input; 2) the three-stage pipeline system is significantly slower in terms of inference speed; and 3) high translation costs since the current state-of-the-art models, especially in low resources languages, are only available using costly APIs.
In this paper, we analyze two possible workarounds to alleviate the aforementioned challenges. The first is to build a cross-lingual transferable system by aligning cross-lingual representations, as in BIBREF3, in which the system is trained on one language and zero-shot to another language. The second is to learn a multilingual system directly from noisy multilingual data (e.g., translated data), thus getting rid of the translation system dependence at inference time.
To evaluate the aforementioned systems, we propose a dataset called Multilingual Persona-Chat, or XPersona, by extending the Persona-Chat corpora BIBREF1 to six languages: Chinese, French, Indonesian, Italian, Korean, and Japanese. In XPersona, the training sets are automatically translated using translation APIs with several human-in-the-loop passes of mistake correction. In contrast, the validation and test sets are annotated by human experts to facilitate both automatic and human evaluations in multiple languages.
Furthermore, we propose competitive baselines in two training settings, namely, cross-lingual and multilingual, and compare them with translation pipeline models. Our baselines leverage pre-trained cross-lingual BIBREF4 and multilingual BIBREF5 models.
An extensive automatic and human evaluation BIBREF6 of our models shows that a multilingual system is able to outperform strong translation-based models and on par with or even improve the monolingual model. The cross-lingual performance is still lower than other models, which indicates that cross-lingual conversation modeling is very challenging. The main contribution of this paper are summarized as follows:
We present the first multilingual non-goal-oriented dialogue benchmark for evaluating multilingual generative chatbots.
We provide both cross-lingual and multilingual baselines and discuss their limitations to inspire future research.
We show the potential of multilingual systems to understand the mixed language dialogue context and generate coherent responses.
To evaluate the aforementioned systems, we propose a dataset called Multilingual Persona-Chat, or XPersona, by extending the Persona-Chat corpora to six languages: Chinese, French, Indonesian, Italian, Korean, and Japanese.
|
What dataset do they propose?
|
Multilingual Persona-Chat.
|
null | false
| null |
What foods are good sources of protein?
|
Protein is essential for body function and is great for repairing and building muscles in your body. Eggs, chicken breast, and lean beef all have a lot of protein. So do dairy items like greek yogurt, cottage cheese, and milk. Vegetarians can find high protein in legumes like lentils, black beans, and chickpeas.
If you find you aren't getting enough protein in your daily diet, consider getting a protein supplement and adding it to smoothies.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.