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 |
What is a SCOBY in kombucha making?
|
A SCOBY is a symbiotic culture of bacteria and yeast. This is what allows the fermentation process to occur to make kombucha and is often referred to as the kombucha starter. The bacteria and yeast in the SCOBY convert the sugar present in the sweet tea used to make alcohol, and also creates the fizziness typical in kombucha. Often the pellicle that forms on the top of the kombucha liquid is referred to as the SCOBY but this is a heated topic amongst many kombucha makers if that is necessary to include as a starter when brewing your next batch of kombucha.
|
|
null | false
| 388
|
Position (3 C features): A sentence on the boundaries of a speaker's segment could contain a reaction to another statement or could provoke a reaction, which in turn could signal a check-worthy claim. Thus, we added information about the position of the target sentence in its segment: whether it is first/last, as well as its reciprocal rank in the list of sentences in that segment.
Segment sizes (3 C features): The size of the segment belonging to one speaker might indicate whether the target sentence is part of a long speech, makes a short comment or is in the middle of a discussion with lots of interruptions. The size of the previous and of the next segments is also important in modeling the dialogue flow. Thus, we include three features with the sizes of the previous, the current and the next segments.
Metadata (8 C features): Check-worthy claims often contain mutual accusations between the opponents, as the following example shows (from the 2nd presidential debate):
Thus, we use a feature that indicates whether the target sentence mentions the name of the opponent, whether the speaker is the moderator, and also who is speaking (3 features). We further use three binary features, indicating whether the target sentence is followed by a system message: applause, laugh, or cross-talk.
A sentence on the boundaries of a speaker’s segment could contain a reaction to another statement or could provoke a reaction, which in turn could signal a check-worthy claim. Thus, we added information about the position of the target sentence in its segment: whether it is first/last, as well as its reciprocal rank in the list of sentences in that segment.
|
Why did the authors add information about the position of the target sentence in its segment?
|
A sentence on the boundaries of a speaker’s segment could contain a reaction to another statement or could provoke a reaction, which in turn could signal a check-worthy claim.
|
null | false
| null |
Identify which instrument is string or percussion: Dabakan, Kwitra
|
Kwitra is string, Dabakan is percussion.
|
|
1910.10288
| false
| null |
FLOAT SELECTED: Table 3. MOS naturalness results along with 95% confidence intervals for the Lessac and LJ datasets.
FLOAT SELECTED: Table 3. MOS naturalness results along with 95% confidence intervals for the Lessac and LJ datasets.
|
Does DCA or GMM-based attention perform better in experiments?
|
About the same performance
|
null | false
| null |
Which city is the best to relocate in Texas
|
Texas offers some great places to live, if you like city life. The prominent choices can be Houston, Dallas, San Antonio, Austin, all of which are part of the Texas triangle, and are connected by 2-3 hours drive. Houston and Dallas are bigger and dense metropolitans. San Antonio and Austin have smaller footprints but the proximity to hill country makes them more charming in terms of geography. Some other places to consider also can be Corpus Christi, Lubbock and El Paso, but are a bit distant.
|
|
null | false
| null |
What are some common types of high heel shoes?
|
Common types of high heel shoes include the stiletto heel, block heel, flare heel, kitten heel, and wedge heel.
|
|
null | false
| null |
What is the minimum feature a cell phone should have to avoid addiction
|
1. Call facility
2. SMS
3.GPS
|
|
null | false
| null |
Identify which instrument is string or percussion: Cowbell, Crwth
|
Crwth is string, Cowbell is percussion.
|
|
null | false
| 78
|
Social media platforms have made the spreading of fake news easier, faster as well as able to reach a wider audience. Social media offer another feature which is the anonymity for the authors, and this opens the door to many suspicious individuals or organizations to utilize these platforms. Recently, there has been an increased number of spreading fake news and rumors over the web and social media BIBREF0. Fake news in social media vary considering the intention to mislead. Some of these news are spread with the intention to be ironic or to deliver the news in an ironic way (satirical news). Others, such as propaganda, hoaxes, and clickbaits, are spread to mislead the audience or to manipulate their opinions. In the case of Twitter, suspicious news annotations should be done on a tweet rather than an account level, since some accounts mix fake with real news. However, these annotations are extremely costly and time consuming – i.e., due to high volume of available tweets Consequently, a first step in this direction, e.g., as a pre-filtering step, can be viewed as the task of detecting fake news at the account level.
The main obstacle for detecting suspicious Twitter accounts is due to the behavior of mixing some real news with the misleading ones. Consequently, we investigate ways to detect suspicious accounts by considering their tweets in groups (chunks). Our hypothesis is that suspicious accounts have a unique pattern in posting tweet sequences. Since their intention is to mislead, the way they transition from one set of tweets to the next has a hidden signature, biased by their intentions. Therefore, reading these tweets in chunks has the potential to improve the detection of the fake news accounts.
In this work, we investigate the problem of discriminating between factual and non-factual accounts in Twitter. To this end, we collect a large dataset of tweets using a list of propaganda, hoax and clickbait accounts and compare different versions of sequential chunk-based approaches using a variety of feature sets against several baselines. Several approaches have been proposed for news verification, whether in social media (rumors detection) BIBREF0, BIBREF1, BIBREF2, BIBREF3, BIBREF4, or in news claims BIBREF5, BIBREF6, BIBREF7, BIBREF8. The main orientation in the previous works is to verify the textual claims/tweets but not their sources. To the best of our knowledge, this is the first work aiming to detect factuality at the account level, and especially from a textual perspective. Our contributions are:
[leftmargin=4mm]
We propose an approach to detect non-factual Twitter accounts by treating post streams as a sequence of tweets' chunks. We test several semantic and dictionary-based features together with a neural sequential approach, and apply an ablation test to investigate their contribution.
We benchmark our approach against other approaches that discard the chronological order of the tweets or read the tweets individually. The results show that our approach produces superior results at detecting non-factual accounts.
To this end, we collect a large dataset of tweets using a list of propaganda, hoax and clickbait accounts and compare different versions of sequential chunk-based approaches using a variety of feature sets against several baselines.
|
What do the authors use to collect the large dataset of tweets?
|
Using a list of propaganda, hoax and clickbait accounts.
|
null | false
| null |
Identify which animal species is alive or extinct: Ceratopsian, Sunda Tiger
|
Sunda Tiger is alive, Ceratopsian is extinct.
|
|
null | false
| null |
Please sort the following numbers in ascending order: 6, 88, 4, 2074, 34
|
4, 6, 34, 88, 2074
|
|
null | false
| null |
Toothpaste is a paste or gel dentifrice used with a toothbrush to clean and maintain the aesthetics and health of teeth. Toothpaste is used to promote oral hygiene: it is an abrasive that aids in removing dental plaque and food from the teeth, assists in suppressing halitosis, and delivers active ingredients (most commonly fluoride) to help prevent tooth decay (dental caries) and gum disease (gingivitis). Owing to differences in composition and fluoride content, not all toothpastes are equally effective in maintaining oral health. The decline of tooth decay during the 20th century has been attributed to the introduction and regular use of fluoride-containing toothpastes worldwide. Large amounts of swallowed toothpaste can be toxic. Common colors for toothpaste include white (sometimes with colored stripes or green tint) and blue.
Usefulness
Toothpastes are generally useful to maintain dental health. Toothpastes containing fluoride are effective at preventing tooth decay. Toothpastes may also help to control and remove plaque build-up, promoting healthy gums. A 2016 systematic review indicated that using toothpaste when brushing the teeth does not necessarily impact the level of plaque removal. However, the active ingredients in toothpastes are able to prevent dental diseases with regular use.
|
How useful is toothpaste
|
Toothpaste is a paste or gel dentifrice used with a toothbrush to clean and maintain the aesthetics and health of teeth. Toothpaste is used to promote oral hygiene: it is an abrasive that aids in removing dental plaque and food from the teeth, assists in suppressing halitosis, and delivers active ingredients (most commonly fluoride) to help prevent tooth decay (dental caries) and gum disease (gingivitis). Owing to differences in composition and fluoride content, not all toothpastes are equally effective in maintaining oral health. The decline of tooth decay during the 20th century has been attributed to the introduction and regular use of fluoride-containing toothpastes worldwide. Large amounts of swallowed toothpaste can be toxic. Common colors for toothpaste include white (sometimes with colored stripes or green tint) and blue.
Toothpastes are generally useful to maintain dental health. Toothpastes containing fluoride are effective at preventing tooth decay. Toothpastes may also help to control and remove plaque build-up, promoting healthy gums. A 2016 systematic review indicated that using toothpaste when brushing the teeth does not necessarily impact the level of plaque removal. However, the active ingredients in toothpastes are able to prevent dental diseases with regular use.
|
null | false
| 83
|
Our target was to build a large Arabic offensive language dataset that is representative of their appearance on Twitter and is hopefully not biased to specific dialects, topics, or targets. One of the main challenges is that offensive tweets constitute a very small portion of overall tweets. To quantify their proportion, we took 3 random samples of tweets from different days, with each sample composed of 1,000 tweets, and we found that between 1% and 2% of them were in fact offensive (including pornographic advertisement). This percentage is consistent with previously reported percentages BIBREF19. Thus, annotating random tweets is grossly inefficient. One way to overcome this problem is to use a seed list of offensive words to filter tweets. However, doing so is problematic as it would skew the dataset to particular types of offensive language or to specific dialects. Offensiveness is often dialect and country specific.
After inspecting many tweets, we observed that many offensive tweets have the vocative particle يا> (“yA” – meaning “O”), which is mainly used in directing the speech to a specific person or group. The ratio of offensive tweets increases to 5% if each tweet contains one vocative particle and to 19% if has at least two vocative particles. Users often repeat this particle for emphasis, as in: يا أمي يا حنونة> (“yA Amy yA Hnwnp” – O my mother, O kind one), which is endearing and non-offensive, and يا كلب يا قذر> (“yA klb yA q*r” – “O dog, O dirty one”), which is offensive. We decided to use this pattern to increase our chances of finding offensive tweets. One of the main advantages of the pattern يا ... يا> (“yA ... yA”) is that it is not associated with any specific topic or genre, and it appears in all Arabic dialects. Though the use of offensive language does not necessitate the appearance of the vocative particle, the particle does not favor any specific offensive expressions and greatly improves our chances of finding offensive tweets. It is clear, the dataset is more biased toward positive class. Using the dataset for real-life application may require de-biasing it by boosting negative class or random sampling additional data from Twitter BIBREF22.Using the Twitter API, we collected 660k Arabic tweets having this pattern between April 15, 2019 and May 6, 2019. To increase diversity, we sorted the word sequences between the vocative particles and took the most frequent 10,000 unique sequences. For each word sequence, we took a random tweet containing each sequence. Then we annotated those tweets, ending up with 1,915 offensive tweets which represent roughly 19% of all tweets. Each tweet was labeled as: offensive, which could additionally be labeled as vulgar and/or hate speech, or Clean. We describe in greater detail our annotation guidelines, which we made sure that they are compatible with the OffensEval2019 annotation guidelines BIBREF15. For example, if a tweet has insults or threats targeting a group based on their nationality, ethnicity, gender, political affiliation, religious belief, or other common characteristics, this is considered as hate speech BIBREF15. It is worth mentioning that we also considered insulting groups based on their sport affiliation as a form of hate speech. In most Arab countries, being a fan of a particularly sporting club is considered as part of the personality and ideology which rarely changes over time (similar to political affiliation). Many incidents of violence have occurred among fans of rival clubs.
Using Twitter APIs, we collected 660k Arabic tweets having this pattern between April 15 – May 6, 2019.
|
How do the authors collect the Arabic tweets?
|
The authors collected 660k Arabic tweets having this pattern between April 15 – May 6, 2019 by using Twitter APIs.
|
null | false
| 19
|
Figure FIGREF32 summarises the average MSD-prediction accuracy for the multi-tasking experiments discussed above. Accuracy here is generally higher than on the main task, with the multilingual finetuned setup for Spanish and the monolingual setup for French scoring best: 66.59% and 65.35%, respectively. This observation illustrates the added difficulty of generating the correct surface form even when the morphosyntactic description has been identified correctly.
We observe some correlation between these numbers and accuracy on the main task: for de, en, ru and sv, the brown, pink and blue bars here pattern in the same way as the corresponding INLINEFORM0 's in Figure FIGREF23 . One notable exception to this pattern is fr where inflection gains a lot from multilingual training, while MSD prediction suffers greatly. Notice that the magnitude of change is not always the same, however, even when the general direction matches: for ru, for example, multilingual training benefits inflection much more than in benefits MSD prediction, even though the MSD decoder is the only component that is actually shared between languages. This observation illustrates the two-fold effect of multi-task training: an auxiliary task can either inform the main task through the parameters the two tasks share, or it can help the main task learning through its regularising effect.
This observation illustrates the two-fold effect of multi-task training: an auxiliary task can either inform the main task through the parameters the two tasks share, or it can help the main task learning through its regularising effect.
|
What two-fold effect of multi-task training does the team conclude according to the observation?
|
An auxiliary task can either inform the main task through the parameters the two tasks share, or it can help the main task learning through its regularising effect.
|
1908.06083
| false
| null |
FLOAT SELECTED: Table 10: Results of experiments on the multi-turn adversarial task. We denote the average and one standard deviation from the results of five runs. Models that use the context as input (“with context”) perform better. Encoding this in the architecture as well (via BERT dialogue segment features) gives us the best results.
FLOAT SELECTED: Table 10: Results of experiments on the multi-turn adversarial task. We denote the average and one standard deviation from the results of five runs. Models that use the context as input (“with context”) perform better. Encoding this in the architecture as well (via BERT dialogue segment features) gives us the best results.
|
What evaluation metric is used?
|
F1 and Weighted-F1
|
null | false
| null |
What is the Shiba Inu
|
The Shiba Inu is a breed of hunting dog from Japan. A small-to-medium breed, it is the smallest of the six original and distinct spitz breeds of dog native to Japan. Its name literally translates to "brushwood dog", as it is used to flush game.
|
|
null | false
| null |
What re the differences between recruiters and sourcers at a company?
|
In typical companies, sourcers are responsible for the upfront outreach of new, passive candidates. Sourcers leverage a variety of tools including LinkedIn, blogs, GitHub, etc. to find potential candidates and then try to contact them directly. It is the sourcers responsibility to identify as many viable candidates as possible for a role. A recruiter is responsible for managing the entire recruiting process once a candidate is identified. Recruiters will help schedule interviews, help the candidate prepare for interviews, and ultimately manage the offer process for the candidate. Recruiters are also responsible for managing the relationship with the hiring manager for a role. They are the point of contact for the entire recruiting process and are responsible for ultimately filling the role.
|
|
null | false
| null |
Give me four ideas for my 7 year old daughters birthday party
|
Here are four ideas for a birthday party for a 7 year old girl:
- A trampoline park (make sure food is provided after!)
- A unicorn themed party with party games like pin the horn on the unicorn
- Hire an ice rink and give the party a frozen theme (Disney or otherwise!)
- Hire a community hall and do a treasure or scavenger hunt
|
|
null | false
| 277
|
AliMe Bot is a kind of retrieval-based online service of E-commerce which collects a lot of predefined question-answering pairs. Through data analysis, we find that many variants of a question exist which means a sort of questions can correspond to a same answer. Based on the observation, naturally, we can view these questions with the same answer as a bag. Obviously, the bag contains diverse expressions of a question, which may provide more matching evidence than only one question due to the rich information contained in the bag. Motivated by the fact, different from existing query-question (Q-Q) matching method, we propose a new query-bag matching approach for retrieval-based chatbots. Concretely, when a user raises a query, the query-bag matching model provides the most suitable bag and returns the corresponding answer of the bag. To our knowledge, there is no query-bag matching study exists, and we focus on the new approach in this paper.
Recalling the text matching task BIBREF0, recently, researchers have adopted the deep neural network to model the matching relationship. ESIM BIBREF1 judges the inference relationship between two sentences by enhanced LSTM and interaction space. SMN BIBREF2 performs the context-response matching for the open-domain dialog system. BIBREF3 BIBREF3 explores the usefulness of noisy pre-training in the paraphrase identification task. BIBREF4 BIBREF4 surveys the methods in query-document matching in web search which focuses on the topic model, the dependency model, etc. However, none of them pays attention to the query-bag matching which concentrates on the matching for a query and a bag containing multiple questions.
When a user poses a query to the bot, the bot searches the most similar bag and uses the corresponding answer to reply to the user. The more information in the query covered by the bag, the more likely the bag's corresponding answer answers the query. What's more, the bag should not have too much information exceeding the query. Thus modelling the bag-to-query and query-to-bag coverage is essential in this task.
In this paper, we propose a simple but effective mutual coverage component to model the above-mentioned problem. The coverage is based on the cross-attention matrix of the query-bag pair which indicates the matching degree of elements between the query and bag. The mutual coverage is performed by stacking the cross-attention matrix along two directions, i.e., query and bag, in the word level respectively. In addition to the mutual coverage, a bag representation in word level is issued to help discover the main points of a bag. The bag representation then provides new matching evidence to the query-bag matching model.
We conduct experiments on the AliMe and Quora dataset for the query-bag matching based information-seeking conversation. Compared with baselines, we verify the effectiveness of our model. Our model obtains 0.05 and 0.03 $\text{R}_{10}@1$ gains comparing to the strongest baseline in the two datasets. The ablation study shows the usefulness of the components. The contributions in this paper are summarized as follows: 1) To the best of our knowledge, we are the first to adopt query-bag matching in the information-seeking conversation. 2) We propose the mutual coverage model to measure the information coverage in the query-bag matching. 3) We release the composite Quora dataset to facilitate the research in this area.
In this paper, we propose a simple but effective mutual coverage component to model the above-mentioned problem. The coverage is based on the cross-attention matrix of the query-bag pair which indicates the matching degree of elements between the query and bag. The mutual coverage is performed by stacking the cross-attention matrix along two directions, i.e., query and bag, in the word level respectively.
|
What is the mutual coverage performed by?
|
By stacking the cross-attention matrix along two directions, i.e., query and bag, in the word level respectively.
|
null | false
| null |
Which 3 mens tennis players have the most grand slam singles championships?
|
Novak Djokovic, Rafael Nadal, and Roger Federer have the most grand slam singles championships.
|
|
null | false
| null |
Suggest some action movies I can watch?
|
Mission Impossible, Goldfinger, Speed, The Magnificent Seven, Die Hard and Top Gun are all great Action movies.
|
|
null | false
| null |
Matt Eskandari attended film school at the University of Southern California. His thesis film, "The Taking," won the award for Best Student Film at Screamfest. Following film school, Eskandari was selected as a contestant for the Steven Spielberg filmmaker competition On The Lot. He would go on to make his feature debut a year later with the independent film Victim released by IFC Films, starring Stacey Haiduk, and premiered at the Cannes Film Festival. The film received a theatrical release, and was distributed on IFC Midnight's Pay-per-view channel.
In 2013 Eskandari directed and produced his film titled The Gauntlet which starred Bai Ling, and Dustin Nguyen. It debuted that October at the Screamfest film festival. Filmed in Beijing, China, the movie "follows five strangers who awake in an underground, medieval dungeon." The film received distribution from Lionsgate Home Entertainment following a screening at Cannes and was retitled Game of Assassins. For the film, Eskandari received several accolades, including the Best Director award from the Asians On Film Festival and the Winter Film Award for Best Feature Film. The official release on DVD and digital VOD was September 23, 2014.
In 2016, Eskandari began filming a new film he co-wrote entitled The Deep End. The film's cast included Tobin Bell, Nora-Jane Noone, and Alexandra Park. It was re-titled to 12 Feet Deep and released wide on June 20, 2017 by Mar Vista Entertainment.
|
Using this paragraph about Matt Eskandari, did Matt Eskandri make any films before Victim?
|
Matt Eskandari filmed "The Taking" before his feature debut with the film Victim.
|
null | false
| null |
What are the differences between an upright piano and a grand piano?
|
The main differences between an upright piano and a grand piano are:
- action mechanisms: an upright piano utilizes a string action to allow vertically placed hammers to reset, while gravity returns the hammers in a grand piano, offering better responsiveness and repetition speed
- pedals: a grand piano usually have 3 pedals, and an upright piano has 2 or 3 pedals with different functions.
Though it depends on the specific models and conditions, a grand piano usually offers better dynamic range and richer resonance than an upright piano.
|
|
null | false
| null |
which of these words do not belong in the same category? peel, organic, fruit, yellow, mushy, crunchy.
|
Only crunchy does not belong in this category. This is because all these words describe a banana. However a banana and crunchy do not go together because a banana is mushy.
|
|
null | false
| null |
What would Abraham Lincoln think of Big Data and Agile Development?
|
In a speech to the House in June 1848, Abraham Lincoln shared his perspective on how to execute desired change, acknowledging that charting a path to a desired outcome is not a linear path and that facts are central to the journey.
A quote that reflects Lincoln's agile mindset: "Determine that the thing can and shall be done, and then we shall find the way." In Agile development, we embrace the fact that we do not know very much about how to reach our goal. We "find the way" by communicating often, sharing facts and executing tasks that we believe will lead us to accomplishing "the thing." We focus on the goal and adapt, as needed, with the facts at hand. And regarding facts, he wanted to count all the pigs and chickens...
A quote that reflects Lincoln's thoughts on "Big Data" (collection): "He particularly objects, as I understand him, to counting all the pigs and chickens in the land. I do not perceive much force in the objection." Abe understood that decisions are advised by facts and data, yet he too warned that some data may not end up being very useful. Regarding the collection of data to guide decisions: "In that information, we shall have a stern, unbending basis of facts---a basis, in nowise subject to whim, caprice, or local interest."
|
|
null | false
| 0
|
In this section, we will present a cross-lingual pre-training based transfer approach. This method is designed for a common zero-shot scenario where there are a lot of source$\leftrightarrow $pivot and pivot$\leftrightarrow $target bilingual data but no source$\leftrightarrow $target parallel data, and the whole training process can be summarized as follows step by step:
Pre-train a universal encoder with source/pivot monolingual or source$\leftrightarrow $pivot bilingual data.
Train a pivot$\rightarrow $target parent model built on the pre-trained universal encoder with the available parallel data. During the training process, we freeze several layers of the pre-trained universal encoder to avoid the degeneracy issue BIBREF27.
Directly translate source sentences into target sentences with the parent model, which benefits from the availability of the universal encoder.
The key difficulty of this method is to ensure the intermediate representations of the universal encoder are language invariant. In the rest of this section, we first present two existing methods yet to be explored in zero-shot translation, and then propose a straightforward but effective cross-lingual pre-training method. In the end, we present the whole training and inference protocol for transfer.
The key difficulty of this method is to ensure the intermediate representations of the universal encoder are language in variant.
|
What's the key difficulty of the proposed method?
|
To ensure the intermediate representations of the universal encoder are language in variant.
|
1910.03814
| false
| null |
Despite the model trained only with images proves that they are useful for hate speech detection, the proposed multimodal models are not able to improve the detection compared to the textual models. Besides the different architectures, we have tried different training strategies, such as initializing the CNN weights with a model already trained solely with MMHS150K images or using dropout to force the multimodal models to use the visual information. Eventually, though, these models end up using almost only the text input for the prediction and producing very similar results to those of the textual models. The proposed multimodal models, such as TKM, have shown good performance in other tasks, such as VQA. Next, we analyze why they do not perform well in this task and with this data:
[noitemsep,leftmargin=*]
Noisy data. A major challenge of this task is the discrepancy between annotations due to subjective judgement. Although this affects also detection using only text, its repercussion is bigger in more complex tasks, such as detection using images or multimodal detection.
Complexity and diversity of multimodal relations. Hate speech multimodal publications employ a lot of background knowledge which makes the relations between visual and textual elements they use very complex and diverse, and therefore difficult to learn by a neural network.
Small set of multimodal examples. Fig. FIGREF5 shows some of the challenging multimodal hate examples that we aimed to detect. But although we have collected a big dataset of $150K$ tweets, the subset of multimodal hate there is still too small to learn the complex multimodal relations needed to identify multimodal hate.
Next, we analyze why they do not perform well in this task and with this data:
[noitemsep,leftmargin=*]
Noisy data. A major challenge of this task is the discrepancy between annotations due to subjective judgement. Although this affects also detection using only text, its repercussion is bigger in more complex tasks, such as detection using images or multimodal detection.
Complexity and diversity of multimodal relations. Hate speech multimodal publications employ a lot of background knowledge which makes the relations between visual and textual elements they use very complex and diverse, and therefore difficult to learn by a neural network.
Small set of multimodal examples. Fig. FIGREF5 shows some of the challenging multimodal hate examples that we aimed to detect. But although we have collected a big dataset of $150K$ tweets, the subset of multimodal hate there is still too small to learn the complex multimodal relations needed to identify multimodal hate.
|
What is author's opinion on why current multimodal models cannot outperform models analyzing only text?
|
The answers are shown as follows:
* Noisy data
* Complexity and diversity of multimodal relations
* Small set of multimodal examples
|
null | false
| null |
The Beatles were an English rock band, formed in Liverpool in 1960, that comprised John Lennon, Paul McCartney, George Harrison and Ringo Starr. They are regarded as the most influential band of all time and were integral to the development of 1960s counterculture and popular music's recognition as an art form. Rooted in skiffle, beat and 1950s rock 'n' roll, their sound incorporated elements of classical music and traditional pop in innovative ways; the band also explored music styles ranging from folk and Indian music to psychedelia and hard rock. As pioneers in recording, songwriting and artistic presentation, the Beatles revolutionised many aspects of the music industry and were often publicised as leaders of the era's youth and sociocultural movements.
|
Name the four members of the Beatles.
|
John Lennon, Paul McCartney, George Harrison and Ringo Starr.
|
null | false
| null |
Where in the US are the most renowned Pinot Noir wines produced?
|
Oregon's Willamette Valley is known for producing award-winning Pinot Noir wine.
|
|
1911.03562
| false
| null |
We focus on the following aspects of NLP research: size, demographics, areas of research, impact, and correlation of citations with demographic attributes (age and gender).
We focus on the following aspects of NLP research: size, demographics, areas of research, impact, and correlation of citations with demographic attributes (age and gender).
|
What aspect of NLP research is examined?
|
The answers are shown as follows:
* size, demographics, areas of research, impact, and correlation of citations with demographic attributes (age and gender)
|
null | false
| null |
What’s the most over powered race in StarCraft 2?
|
Based on the mechanics of the worker units of the 3 Races in Starcraft 2; Zerg Drone, Protoss Probe, and Terran SCV, the Protoss are widely regarded as the most over powered race. Competitive Starcraft 2 requires players to gain an advantage by constantly managing their resources and building their army. The Protoss Probe has the simplest interaction to build new structures. Furthermore, the Protoss arsenal consists of some of the most expensive and powerful units in the game. The other races have means to counter the Protoss however the Community agrees the Protoss requires the least amount of skill to master.
|
|
1911.01799
| false
| null |
FLOAT SELECTED: Table 4. EER(%) results of the i-vector and x-vector systems trained on VoxCeleb and evaluated on three evaluation sets.
FLOAT SELECTED: Table 4. EER(%) results of the i-vector and x-vector systems trained on VoxCeleb and evaluated on three evaluation sets.
|
By how much is performance on CN-Celeb inferior to performance on VoxCeleb?
|
For i-vector system, performances are 11.75% inferior to voxceleb. For x-vector system, performances are 10.74% inferior to voxceleb
|
null | false
| null |
Arup Kumar Raychaudhuri (born 1 January 1952) is an Indian condensed matter physicist, materials scientist and a Distinguished Emeritus Professor at the S. N. Bose National Centre for Basic Sciences. Known for his pioneering work on the interplay of disorder and interaction, Raychaudhuri is an elected fellow of all the three major Indian science academies viz. Indian Academy of Sciences, National Academy of Sciences, India and Indian National Science Academy as well as the Asia-Pacific Academy of Materials. He is a recipient of a number of awards such as Millennium Medal of the Indian Science Congress, ICS Gold Medal of the Materials Research Society of India and FICCI Award. The Council of Scientific and Industrial Research, the apex agency of the Government of India for scientific research, awarded him the Shanti Swarup Bhatnagar Prize for Science and Technology, one of the highest Indian science awards, for his contributions to physical sciences in 1994.[note 1]
|
What type of physicist is Arup Kumar Raychaudhuri?
|
Arup Kumar Raychaudhuri is a condensed matter physicists.
|
null | false
| 31
|
purpleIn this paper, we propose a novel extractive summarization model especially designed for long documents, by incorporating the local context within each topic, along with the global context of the whole document.[2] purpleOur approach integrates recent findings on neural extractive summarization in a parameter lean and modular architecture.[3] purpleWe evaluate our model and compare with previous works in both extractive and abstractive summarization on two large scientific paper datasets, which contain documents that are much longer than in previously used corpora.[4] purpleOur model not only achieves state-of-the-art on these two datasets, but in an additional experiment, in which we consider documents with increasing length, it becomes more competitive for longer documents.[5] purpleWe also ran an ablation study to assess the relative contribution of the global and local components of our approach. [1] Rather surprisingly, it appears that the benefits of our model come only from modeling the local context.
For future work, we initially intend to investigate neural methods to deal with redundancy. Then, it could be beneficial to integrate explicit features, like sentence position and salience, into our neural approach. More generally, we plan to combine of traditional and neural models, as suggested by our results. Furthermore, we would like to explore more sophistical structure of documents, like discourse tree, instead of rough topic segments. As for evaluation, we would like to elicit human judgments, for instance by inviting authors to rate the outputs from different systems, when applied to their own papers. More long term, we will study how extractive/abstractive techniques can be integrated; for instance, the output of an extractive system could be fed into an abstractive one, training the two jointly.
Our model not only achieves state-of-the-art on these two datasets, but in an additional experiment, in which we consider documents with increasing length, it becomes more competitive for longer documents
|
What are the results of this paper?
|
Their model not only achieves state-of-the-art on these two datasets, but in an additional experiment, in which they consider documents with increasing length, it becomes more competitive for longer documents.
|
null | false
| null |
The 2011 ICC Cricket World Cup was the tenth Cricket World Cup. It was played in India, Sri Lanka, and for the first time in Bangladesh. India won the tournament, defeating Sri Lanka by 6 wickets in the final at Wankhede Stadium in Mumbai, thus becoming the first country to win the Cricket World Cup final on home soil. India's Yuvraj Singh was declared the man of the tournament. This was the first time in World Cup history that two Asian teams had appeared in the final. It was also the first time since the 1992 World Cup that the final match did not feature Australia.
Fourteen national cricket teams took part in this tournament, including 10 full members and four associate members of the International Cricket Council (ICC). The opening ceremony was held on 17 February 2011 at Bangabandhu National Stadium, Dhaka, and the tournament was played between 19 February and 2 April. The first match was played between India and Bangladesh at the Sher-e-Bangla National Stadium in Mirpur, Dhaka.
|
Short Summary about 2011 Cricket World Cup
|
- 2011 Cricket World Cup was played in India, Sri Lanka and Bangladesh
- India won the tournament defeating Sri Lanka.
- Yuvraj Singh was declared the Man of the Tournament
- Opening Ceremony was held on 17 February 2011 at Dhaka
|
null | false
| null |
Is there an evolutionary root cause for self-deception?
|
Diverse evolutionary psychologists have conjectured about the potential advantages of self-deception in a social context, however some Terror Management theorists suggest that self-deception is actually at the core of what it means to be human. Hunter-gatherers may have used their increasing cognitive abilities to their own evolutionary benefit, but increased cognition necessarily brought along the awareness of death, which could become a major obstacle by making the dangers of day-to-day so explicit and linked to the possibility of death that would cause paralysis -thus it would not be evolutionary anymore. In conclusion, awareness of death could only have emerged in combination with self-deception, so that the terror of death would be managed and diminished. In other words, the theory dismisses the possibility of the advent of a breed of intelligent beings, unless those beings, no matter how clever they come to be, have as well the ability to conceal their own mortality.
|
|
null | false
| null |
This is a list of Billboard magazine's Top Hot 100 songs of 1980.
No. Title Artist(s)
1 "Call Me" Blondie
2 "Another Brick in the Wall, Part II" Pink Floyd
3 "Magic" Olivia Newton-John
4 "Rock with You" Michael Jackson
5 "Do That to Me One More Time" Captain & Tennille
6 "Crazy Little Thing Called Love" Queen
7 "Coming Up" Paul McCartney
8 "Funkytown" Lipps Inc
9 "It's Still Rock and Roll to Me" Billy Joel
10 "The Rose" Bette Midler
|
What is the billboard number one in the top 10 of 1980?
|
"Call Me" is a song by the American new wave band Blondie and the theme to the 1980 film American Gigolo. "Call Me" was No. 1 for six consecutive weeks on the Billboard Hot 100 chart, where it became the band's biggest single and second No. 1.
|
null | false
| null |
My refrigerator broke for six days while I was on vacation. Which of the following foods would be still safe to eat unrefrigerated for that long: raw chicken, a jar of olives, hard cheese, milk, and fruit jelly.
|
Olives, hard cheese, and fruit jelly are not likely to spoil unrefrigerated for a couple of days. However, raw chicken and milk would be unsafe to eat.
|
|
null | false
| 10
|
The BERT models perform remarkably well on all the syntactic test cases. I expected the attention-based mechanism to fail on these (compared to the LSTM-based models), and am surprised by these results. The BIBREF2 and BIBREF3 conditions rule out the possibility of overly relying on selectional preference cues or memorizing the wikipedia training data, and suggest real syntactic generalization is taking place. Exploring the extent to which deep purely-attention-based architectures such as BERT are capable of capturing hierarchy-sensitive and syntactic dependencies—as well as the mechanisms by which this is achieved—is a fascinating area for future research.
Exploring the extent to which deep purely-attention-based architectures such as BERT are capable of capturing hierarchy-sensitive and syntactic dependencies—as well as the mechanisms by which this is achieved—is a fascinating area for future research.
|
What is the author's research direction for his future work?
|
Exploring the extent to which deep purely-attention-based architectures such as BERT are capable of capturing hierarchy-sensitive and syntactic dependencies—as well as the mechanisms by which this is achieved.
|
1906.03538
| false
| null |
We use crowdsourcing to annotate different aspects of the dataset. We used Amazon Mechanical Turk (AMT) for our annotations, restricting the task to workers in five English-speaking countries (USA, UK, Canada, New Zealand, and Australia), more than 1000 finished HITs and at least a 95% acceptance rate. To ensure the diversity of responses, we do not require additional qualifications or demographic information from our annotators.
We used Amazon Mechanical Turk (AMT) for our annotations, restricting the task to workers in five English-speaking countries (USA, UK, Canada, New Zealand, and Australia), more than 1000 finished HITs and at least a 95% acceptance rate.
|
What crowdsourcing platform did they use?
|
The answers are shown as follows:
* Amazon Mechanical Turk (AMT)
|
null | false
| null |
Porthcurno is largely not farms or fisherman's cottages today, having its linear centre inland, centred 6.6 miles (10.6 km) west of the railway, market and resort town of Penzance and 2.5 miles (4.0 km) from Land's End, the most westerly point of the English mainland. In the most local-level organisational and community bodies, it is in civil and ecclesiastical parishes named Saint Levan and usually spelt as St Leven since the 18th century.
|
Currently, does Porthcurno mainly comprise farms and fisherman's cottages?
|
No
|
null | false
| 37
|
With this work we present a resource that will be extremely useful for building language systems in an endangered, under-represented language, Mapudungun. We benchmark NLP systems for speech synthesis, speech recognition, and machine translation, providing strong baseline results. The size of our resource (142 hours, more than 260k total sentences) has the potential to alleviate many of the issues faced when building language technologies for Mapudungun, in contrast to other indigenous languages of the Americas that unfortunately remain low-resource.
Our resource could also be used for ethnographic and anthropological research into the Mapuche culture, and has the potential to contribute to intercultural bilingual education, preservation activities and further general advancement of the Mapudungun-speaking community.
The size of our resource (142 hours, more than 260k total sentences) has the potential to alleviate many of the issues faced when building language technologies for Mapudungun, in contrast to other indigenous languages of the Americas that unfortunately remain low-resource
|
What is the size of the resource in this paper?
|
142 hours, more than 260k total sentences.
|
null | false
| null |
What is a surf board used for?
|
A surfboard is shaped like a narrow plank. Surfers stand on the top of the surfboard and use it to surf on the ocean. They tend to be light and are strong enough to get hit by a large wave without breaking in half. Surfboards were originally invented in Hawaii and were made using the wood from local trees. Early surfboards from Hawaii were over 15 feet long and were referred to as longboards. Surfboards got a lot shorter as surfers had wanted to be more agile with them in the ocean.
|
|
null | false
| null |
In their travels, Arthur comes to learn that the Earth was actually a giant supercomputer, created by another supercomputer, Deep Thought. Deep Thought had been built by its creators to give the answer to the "Ultimate Question of Life, the Universe, and Everything", which, after eons of calculations, was given simply as "42". Deep Thought was then instructed to design the Earth supercomputer to determine what the Question actually is. The Earth was subsequently destroyed by the Vogons moments before its calculations were completed, and Arthur becomes the target of the descendants of the Deep Thought creators, believing his mind must hold the Question. With his friends' help, Arthur escapes and they decide to have lunch at The Restaurant at the End of the Universe, before embarking on further adventures.
|
What is the meaning of life?
|
42
|
1611.02550
| false
| null |
An acoustic word embedding is a function that takes as input a speech segment corresponding to a word, INLINEFORM0 , where each INLINEFORM1 is a vector of frame-level acoustic features, and outputs a fixed-dimensional vector representing the segment, INLINEFORM2 . The basic embedding model structure we use is shown in Fig. FIGREF1 . The model consists of a deep RNN with some number INLINEFORM3 of stacked layers, whose final hidden state vector is passed as input to a set of INLINEFORM4 of fully connected layers; the output of the final fully connected layer is the embedding INLINEFORM5 .
An acoustic word embedding is a function that takes as input a speech segment corresponding to a word, INLINEFORM0 , where each INLINEFORM1 is a vector of frame-level acoustic features, and outputs a fixed-dimensional vector representing the segment, INLINEFORM2 .
|
How do they represent input features of their model to train embeddings?
|
The answers are shown as follows:
* a vector of frame-level acoustic features
|
null | false
| null |
Psilocybin mushrooms, commonly known as magic mushrooms, are a polyphyletic informal group of fungi that contain psilocybin which turns into psilocin upon ingestion. Biological genera containing psilocybin mushrooms include Psilocybe, Panaeolus (including Copelandia), Inocybe, Pluteus, Gymnopilus, and Pholiotina. Psilocybin mushrooms have been and continue to be used in indigenous New World cultures in religious, divinatory, or spiritual contexts. Psilocybin mushrooms are also used as recreational drugs. They may be depicted in Stone Age rock art in Africa and Europe but are most famously represented in the Pre-Columbian sculptures and glyphs seen throughout North, Central, and South America.
|
What is the biological term for Magic Mushrooms?
|
Magic Mushrooms are the colloquial term for Psilocybin mushrooms
|
null | false
| null |
2022 Mazar-i-Sharif mosque bombing
On 21 April 2022, a powerful bomb rocked the Shia Seh Dokan mosque in Mazar-i-Sharif, Balkh Province, Afghanistan killing at least 31 people and injuring more than 87 others. The Islamic State – Khorasan Province claimed responsibility via Telegram.
Worshippers were performing the Zuhr prayer inside the mosque when the bomb exploded. Authorities counted 31 dead and at least 87 injured, six of whom were wounded critically.
The bombing was part of a series of attacks on the same day, including two explosions targeting Taliban officers in Kunduz and Khogyani District and a bombing that injured two Shia children in Kabul.
|
Given a reference text about the 2022 Mazar-i-Sharif mosque bombing, how many people were injured and killed?
|
At least 31 people were killed and more than 87 others were injured in the 2022 Mazar-i-Sharif mosque bombing.
|
null | false
| null |
The MECE principle, (mutually exclusive and collectively exhaustive) is a grouping principle for separating a set of items into subsets that are mutually exclusive (ME) and collectively exhaustive (CE). It was developed in the late 1960s by Barbara Minto at McKinsey & Company and underlies her Minto Pyramid Principle, and while she takes credit for MECE, according to her interview with McKinsey, she says the idea for MECE goes back as far as to Aristotle.
The MECE principle has been used in the business mapping process wherein the optimum arrangement of information is exhaustive and does not double count at any level of the hierarchy. Examples of MECE arrangements include categorizing people by year of birth (assuming all years are known), apartments by their building number, letters by postmark, and dice rolls. A non-MECE example would be categorization by nationality, because nationalities are neither mutually exclusive (some people have dual nationality) nor collectively exhaustive (some people have none).
|
Please summarize the MECE principle, give an example and explain how it can be used in business
|
MECE stands for Mutually Exclusive, Collectively Exhaustive and serves as an organizing framework used by management consulting firms, such as McKinsey and BCG as a way to organize reports and complicated data. Due to its effectiveness, it has made it's way into the larger business community. An example MECE arrangement would be departments within an organization, year hired for employees, the choice to eat in or eat out. MECE can be used to simplify complex ideas.
|
null | false
| null |
Think up some locations for me to ask my partner out
|
- On a greek island
- At Disney World
- At their favorite pizza place
- On a cruise ship
- At the end of a hike
- At the Burj Khalifa
- On a private charter helicopter ride
- At Times Square, NYC
- At the Taj Mahal
- At a surprise party in their favorite place growing up
- where you first met your partner
- where you first confessed that you like each other
- at your favorite date spot
- at the northern lights in Norway
|
|
null | false
| null |
Air Marshal Sir Edward “Tap” Gordon Jones, KCB, CBE, DSO, DFC (31 August 1914 – 20 February 2007) was an officer in the Royal Air Force for 34 years, from 1935 to 1969. He commanded a squadron of obsolescent biplane Gladiator fighters during the Greek Campaign in the Second World War, where he shot down five Italian Fiat CR.42 fighters. He served in mainly operational posts in Europe until he retired. Unusually, he never served a post in the Air Ministry or the UK Ministry of Defence.
|
Who was Tap Gordon Jones?
|
Tap Gordon Jones, full name Air Marshal Sir Edward "Tap" Gordon Jones, KCB, CBE, DSO, DFC was an officer in the UK Royal Air Force from 1935 to 1969. He commanded a squadron of Gladiators during the Greek Campaign in Wold War II. At the time Gladiators were considered to be obsolete. During the Greek Campaign, he shot down Five Italian fighters. After the war until his retirement he mainly served in operational posts in Europe.
|
null | false
| null |
Give me a list of good psychedelic rock bands that I'd like
|
- Tame Impala
- Pond
- Kikagaku Moyo
- Unknown Mortal Orchestra
- King Gizzard & the Lizard Wizard
- Pink Floyd
- The Doors
- Grateful Dead
- The Flaming Lips
- The Beatles
- Post Animal
- The Who
|
|
null | false
| null |
What is a sight-reading contest?
|
In musical education, sight-reading is the ability to perform a newly-encountered musical composition within just a few minutes. Music students typically enter a rehearsal room with their assigned instruments, sit down, and are given a just a few moments to learn a new piece. Oftentimes, directors are allowed to walk them quickly through the composition, perhaps clapping or singing, but the students are not allowed to practice the piece or mark up the musical notation. The ensemble that delivers the best performance (according to the contest's judges) in the given amount of time will win the contest.
|
|
null | false
| null |
classify each of the following as a US state or US state capital: Atlanta, Washington, Boston, Iowa, Sacramento, Alaska
|
US States: Washington, Iowa, Alaska
US State Capitals: Atlanta, Boston, Sacramento
|
|
null | false
| null |
Why should I cheer for the Sacramento Kings in this year's NBA playoffs?
|
Until this season, the Sacramento Kings held the longest playoff drought in all of North American professional sports, lasting 16 years. Despite this, the fanbase for the Kings remained loyal and regularly attended games to watch their team lose year after year. In the midst of this drought, the previous owners of the franchise attempted to move the team to a new city, with significant traction at various points, to both Anaheim and Seattle. It was only the passion of the fans, and the combined support from local businesses and politicians, that prevented the move. In addition to the Kings having one of the most supportive fanbases in the NBA getting their first taste of success, the Kings are also a highly enjoyable team to watch. The players are friendly and likable, and the offense is one of the most successful in NBA history. Lastly, the team also has a fun gimmick of lighting a large purple laser beam that can be seen for miles after every victory, leading to a chorus of 'Light the beam!' from the Kings fanbase in the arena at the conclusion of every win. Pulling all of these factors together make the Sacramento Kings a fun team to cheer for in the upcoming NBA playoffs.
|
|
null | false
| 6
|
We explored improving the performance of neural abstractive summarizers when applied to the low resource domain of student reflections using three approaches: domain transfer, data synthesis and the combination of both. For domain transfer, state of the art abstractive summarization model was pretrained using out-of-domain data (CNN/DM), then tuned using in-domain data (student reflections). The process of tuning improved ROUGE scores on the student reflection data, and at the same time produced more readable summaries. To incorporate synthetic data, we proposed a new template based synthesis model to synthesize new summaries. We showed that enriching the training data with this synthesized data can further increase the benefits of using domain transfer / tuning to increase ROUGE scores. We additionally showed that the proposed synthesis model outperformed a word replacement synthesis baseline. Future plans include trying domain adaptation, enhancing the synthesising process by using other models, further exploring template-based methods, and extending the analysis of the synthesis model to cover other types of data like reviews and opinions.
We additionally showed that the proposed synthesis model outperformed a word replacement synthesis baseline.
|
Did the synthesis model proposed by the authors outperform the word replacement synthesis baseline?
|
Yes, it did.
|
null | false
| null |
Give me a list of 3 good ways to gain weight.
|
Here is a list of 3 good ways to gain weight:
1. Eat more food, especially highly processed foods and foods high in sugars
2. Don't exercise, sit and do nothing for as long as you can
3. Drink sugary drinks for extra calories
|
|
null | false
| null |
Why is the sky blue?
|
The sky appears to be blue because of the way sunlight scatters in the atmosphere. The white light from the sun is composed of many frequencies and as it travels through the atmosphere, the blue light, which has a shorter frequency, scatters more. When looking at an area of the sky away from the sun we see the scattered blue light.
|
|
null | false
| null |
How do plants get their energy
|
They use photosynthesis and turn the sunlight into glucose as food
|
|
null | false
| null |
When is the best time to catch Bass in Michigan?
|
You have the best chance of hooking into a Bass in the early morning or evening on inland lakes in Michigan. Since Bass spawn in the spring, you also will increase your chances of catching bass in Spring (Typically April/May/June) when the water temperature hits 63 degrees F. Bass like shallow water, grassy areas, and spawn in sandy areas of the lake. Use a spinner bait, fake green worm, or top water bait for your best chances of catching a big bass!
|
|
null | false
| 31
|
Single-document summarization is the task of generating a short summary for a given document. Ideally, the generated summaries should be fluent and coherent, and should faithfully maintain the most important information in the source document. purpleThis is a very challenging task, because it arguably requires an in-depth understanding of the source document, and current automatic solutions are still far from human performance BIBREF0 .
Single-document summarization can be either extractive or abstractive. Extractive methods typically pick sentences directly from the original document based on their importance, and form the summary as an aggregate of these sentences. Usually, summaries generated in this way have a better performance on fluency and grammar, but they may contain much redundancy and lack in coherence across sentences. In contrast, abstractive methods attempt to mimic what humans do by first extracting content from the source document and then produce new sentences that aggregate and organize the extracted information. Since the sentences are generated from scratch they tend to have a relatively worse performance on fluency and grammar. Furthermore, while abstractive summaries are typically less redundant, they may end up including misleading or even utterly false statements, because the methods to extract and aggregate information form the source document are still rather noisy.
In this work, we focus on extracting informative sentences from a given document (without dealing with redundancy), especially when the document is relatively long (e.g., scientific articles).
Most recent works on neural extractive summarization have been rather successful in generating summaries of short news documents (around 650 words/document) BIBREF1 by applying neural Seq2Seq models BIBREF2 . However when it comes to long documents, these models tend to struggle with longer sequences because at each decoding step, the decoder needs to learn to construct a context vector capturing relevant information from all the tokens in the source sequence BIBREF3 .
Long documents typically cover multiple topics. In general, the longer a document is, the more topics are discussed. As a matter of fact, when humans write long documents they organize them in chapters, sections etc.. Scientific papers are an example of longer documents and they follow a standard discourse structure describing the problem, methodology, experiments/results, and finally conclusions BIBREF4 .
To the best of our knowledge only one previous work in extractive summarization has explicitly leveraged section information to guide the generation of summaries BIBREF5 . However, the only information about sections fed into their sentence classifier is a categorical feature with values like Highlight, Abstract, Introduction, etc., depending on which section the sentence appears in.
In contrast, in order to exploit section information, in this paper we propose to capture a distributed representation of both the global (the whole document) and the local context (e.g., the section/topic) when deciding if a sentence should be included in the summary
Our main contributions are as follows: (i) In order to capture the local context, we are the first to apply LSTM-minus to text summarization. LSTM-minus is a method for learning embeddings of text spans, which has achieved good performance in dependency parsing BIBREF6 , in constituency parsing BIBREF7 , as well as in discourse parsing BIBREF8 . With respect to more traditional methods for capturing local context, which rely on hierarchical structures, LSTM-minus produces simpler models i.e. with less parameters, and therefore faster to train and less prone to overfitting. (ii) We test our method on the Pubmed and arXiv datasets and results appear to support our goal of effectively summarizing long documents. In particular, while overall we outperform the baseline and previous approaches only by a narrow margin on both datasets, the benefit of our method become much stronger as we apply it to longer documents. purpleFurthermore, in an ablation study to assess the relative contributions of the global and the local model we found that, rather surprisingly, the benefits of our model seem to come exclusively from modeling the local context, even for the longest documents.[6] (iii) In order to evaluate our approach, we have created oracle labels for both Pubmed and arXiv BIBREF9 , by applying a greedy oracle labeling algorithm. The two datasets annotated with extractive labels will be made public.
In order to evaluate our approach, we have created oracle labels for both Pubmed and arXiv, by applying a greedy oracle labeling algorithm.
|
What has been created by the authors for Pubmed and arXiv?
|
Oracle labels.
|
null | false
| null |
What is the FDA doing to respond to the COVID-19 pandemic?
|
The FDA, along with other federal, state, and local agencies and public health officials across the country and internationally, plays a critical role in protecting public health during the COVID-19 pandemic. FDA staff are working around the clock to support development of medical countermeasures and are providing regulatory advice, guidance, and technical assistance to advance the development and availability of vaccines, therapies, diagnostic tests and other medical devices for use diagnosing, treating, and preventing this novel virus. The FDA continues to monitor the human and animal food supply and take swift action on fraudulent COVID-19 products.
|
|
null | false
| null |
Write a review of Brandon Sanderson's book Warbreaker
|
Can't believe what a consistent hitter Sanderson is.
The characters really grew throughout the novel. Siri and Vivenna switch places, and also switch personalities, reveling in their new preferences. Lightsong owes up to the man he once was, and I found his humor quite dad-joke-y and well, good. Loved Denth. And Susebron, well, grows up.
Biochroma was a fun concept, and made for some interesting power plays.
|
|
null | false
| null |
what was the biggest tax change from tax reform
|
competitive US corp tax rates when compared to rest of world
|
|
null | false
| null |
Which of the following does not fit with the others? Please explain your reasoning.
Galaxy M81
Black Eye Galaxy
Pinwheel Galaxy
Andromeda Galaxy
Sombrero Galaxy
|
Galaxy M81 does not fit with the other 4 in the list as M81is a designation rather than a proper name. The other 4 galaxies listed also have a designation, e.g Sombrero Galaxy is designated M104. M81 also has a proper name, Bode's Galaxy. But as listed, Galaxy M81 is distinct from the rest of the list.
|
|
null | false
| 140
|
Antisocial behavior. Antisocial behavior online comes in many forms, including harassment BIBREF30, cyberbullying BIBREF31, and general aggression BIBREF32. Prior work has sought to understand different aspects of such behavior, including its effect on the communities where it happens BIBREF33, BIBREF34, the actors involved BIBREF35, BIBREF36, BIBREF37, BIBREF38 and connections to the outside world BIBREF39.
Post-hoc classification of conversations. There is a rich body of prior work on classifying the outcome of a conversation after it has concluded, or classifying conversational events after they happened. Many examples exist, but some more closely related to our present work include identifying the winner of a debate BIBREF40, BIBREF41, BIBREF42, identifying successful negotiations BIBREF21, BIBREF43, as well as detecting whether deception BIBREF44, BIBREF45, BIBREF46 or disagreement BIBREF47, BIBREF48, BIBREF49, BIBREF50, BIBREF51 has occurred.
Our goal is different because we wish to forecast conversational events before they happen and while the conversation is still ongoing (potentially allowing for interventions). Note that some post-hoc tasks can also be re-framed as forecasting tasks (assuming the existence of necessary labels); for instance, predicting whether an ongoing conversation will eventually spark disagreement BIBREF18, rather than detecting already-existing disagreement.
Conversational forecasting. As described in Section SECREF1, prior work on forecasting conversational outcomes and events has largely relied on hand-crafted features to capture aspects of conversational dynamics. Example feature sets include statistical measures based on similarity between utterances BIBREF16, sentiment imbalance BIBREF20, flow of ideas BIBREF20, increase in hostility BIBREF8, reply rate BIBREF11 and graph representations of conversations BIBREF52, BIBREF17. By contrast, we aim to automatically learn neural representations of conversational dynamics through pre-training.
Such hand-crafted features are typically extracted from fixed-length windows of the conversation, leaving unaddressed the problem of unknown horizon. While some work has trained multiple models for different window-lengths BIBREF8, BIBREF18, they consider these models to be independent and, as such, do not address the issue of aggregating them into a single forecast (i.e., deciding at what point to make a prediction). We implement a simple sliding windows solution as a baseline (Section SECREF5).
Pre-training for NLP. The use of pre-training for natural language tasks has been growing in popularity after recent breakthroughs demonstrating improved performance on a wide array of benchmark tasks BIBREF53, BIBREF54. Existing work has generally used a language modeling objective as the pre-training objective; examples include next-word prediction BIBREF55, sentence autoencoding, BIBREF56, and machine translation BIBREF57. BERT BIBREF58 introduces a variation on this in which the goal is to predict the next sentence in a document given the current sentence. Our pre-training objective is similar in spirit, but operates at a conversation level, rather than a document level. We hence view our objective as conversational modeling rather than (only) language modeling. Furthermore, while BERT's sentence prediction objective is framed as a multiple-choice task, our objective is framed as a generative task.
Such hand-crafted features are typically extracted from fixed-length windows of the conversation, leaving unaddressed the problem of unknown horizon. While some work has trained multiple models for different window-lengths (Liu et al., 2018; Hessel and Lee, 2019), they consider these models to be independent and, as such, do not address the issue of aggregating them into a single forecast (i.e., deciding at what point to make a prediction). We implement a simple sliding windows solution as a baseline (Section 5).
|
Why don't they solve the issue of aggregating models into a single forecast when some work train multiple models for different window lengths?
|
Because they consider these models to be independent.
|
null | false
| null |
What is Accrued income security
|
This is a security (investment) which pays interest at regular intervals. When it is sold, interest may have built up and this interest will be paid to the new owner. Interest built up like this is called accrued interest.
|
|
null | false
| null |
What is the birthday problem?
|
The Birthday Problem is a probability problem with results that many find counterintuitive. The problem is formulated as "how many people must you consider such that the probability of at least two people sharing a birthday is at least 50%?". The answer, which many find surprising and counterintuitive, is 23. The answer is lower than most people expect and it is due to the number of pairwise permutations that exists for 23 people.
|
|
null | false
| 7
|
One interesting pattern is that the ethnicity/race of babies doesn't seem to be mentioned unless the baby is black or asian. In other words: white seems to be the default, and others seem to be marked. How can we tell whether or not the data is actually biased?
We don't know whether or not an entity belongs to a particular social class (in this case: ethnic group) until it is marked as such. But we can approximate the proportion by looking at all the images where the annotators have used a marker (in this case: adjectives like black, white, asian), and for those images count how many descriptions (out of five) contain a marker. This gives us an upper bound that tells us how often ethnicity is indicated by the annotators. Note that this upper bound lies somewhere between 20% (one description) and 100% (5 descriptions). Figure TABREF22 presents count data for the ethnic marking of babies. It includes two false positives (talking about a white baby stroller rather than a white baby). In the Asian group there is an additional complication: sometimes the mother gets marked rather than the baby. E.g. An Asian woman holds a baby girl. I have counted these occurrences as well.
The numbers in Table TABREF22 are striking: there seems to be a real, systematic difference in ethnicity marking between the groups. We can take one step further and look at all the 697 pictures with the word `baby' in it. If there turn out to be disproportionately many white babies, this strengthens the conclusion that the dataset is biased.
I have manually categorized each of the baby images. There are 504 white, 66 asian, and 36 black babies. 73 images do not contain a baby, and 18 images do not fall into any of the other categories. While this does bring down the average number of times each category was marked, it also increases the contrast between white babies (who get marked in less than 1% of the images) and asian/black babies (who get marked much more often). A next step would be to see whether these observations also hold for other age groups, i.e. children and adults. INLINEFORM0
The numbers in Table 1 are striking: there seems to be a real, systematic difference in ethnicity marking between the groups. We can take one step further and look at all the 697 pictures with the word ‘baby’ in it. If there turn out to be disproportionately many white babies, this strengthens the conclusion that the dataset is biased.
|
What does the author analyze from the numbers in Table 1?
|
The numbers in Table 1 are striking: there seems to be a real, systematic difference in ethnicity marking between the groups. They can take one step further and look at all the 697 pictures with the word ‘baby’ in it. If there turn out to be disproportionately many white babies, this strengthens the conclusion that the dataset is biased.
|
null | false
| null |
Caroline Vreeland was born Caroline Olivia Zickerick in Washington, D.C. to Daisy Vreeland and Michael Zickerick. Her father, a German diplomat, served as the German Ambassador to Moldova and as the Director General of the German Institute Taipei in Taiwan. Her maternal great-grandmother was fashion editor Diana Vreeland. She is a grandniece of American diplomat Frederick Vreeland and a first cousin once removed of the Buddhist abbot Nicholas Vreeland. Vreeland spent four months of her childhood living at the German Embassy in Kingston, Jamaica while her father was stationed there. Her father later had an affair and her parents divorced. Vreeland's mother moved them to a Siddha Yoga ashram in Upstate New York. They left New York and moved in with her grandparents in Belvedere, Marin County in the San Francisco Bay Area. After graduating from high school she moved to Los Angeles to pursue a music career.
|
Who was Caroline Olivia Zickerick's grandmother?
|
Caroline Olivia Zickerick grandmother's was fashion editor Diana Vreeland
|
1909.09484
| false
| null |
We adopt the DSTC2 BIBREF20 dataset and Maluuba BIBREF21 dataset to evaluate our proposed model. Both of them are the benchmark datasets for building the task-oriented dialog systems. Specifically, the DSTC2 is a human-machine dataset in the single domain of restaurant searching. The Maluuba is a very complex human-human dataset in travel booking domain which contains more slots and values than DSTC2. Detailed slot information in each dataset is shown in Table TABREF34.
We adopt the DSTC2 BIBREF20 dataset and Maluuba BIBREF21 dataset to evaluate our proposed model.
|
What two benchmark datasets are used?
|
The answers are shown as follows:
* DSTC2
* Maluuba
|
null | false
| null |
Why do leaves on houseplants turn yellow and brown?
|
Often houseplant leaves turn yellow and brown for a few common reasons. First is overwatering. Houseplants are kept in containers that can retain water which can lead to root rot if not watered carefully. Some houseplants prefer the soil to dry out between waterings and many home gardeners end up watering before the soil has dried up sufficiently. Too much water in a container can lead to root rot which first can show symptoms in browning, dropping, and yellowing of leaves. Another reason for leaf discoloration is under watering as some plants prefer the soil to have a level of moisture at all times. Other reasons include too little sunlight which limits the plants ability to photosynthesize. Also, even too much sunlight can cause burned leaves.
|
|
null | false
| null |
Which letters denote Jesus Nazareth King of the Jews
|
INRI
|
|
null | false
| 127
|
Adversarial examples, a term introduced in BIBREF0, are inputs transformed by small perturbations that machine learning models consistently misclassify. The experiments are conducted in the context of computer vision (CV), and the core idea is encapsulated by an illustrative example: after imperceptible noises are added to a panda image, an image classifier predicts, with high confidence, that it is a gibbon. Interestingly, these adversarial examples can also be used to improve the classifier — either as additional training data BIBREF0 or as a regularisation objective BIBREF1 — thus providing motivation for generating effective adversarial examples.
The germ of this paper comes from our investigation of adversarial attack methods for natural language processing (NLP) tasks, e.g. sentiment classification, which drives us to quantify what is an “effective” or “good” adversarial example. In the context of images, a good adversarial example is typically defined according two criteria:
it has successfully fooled the classifier;
it is visually similar to the original example.
In NLP, defining a good adversarial example is a little more involving, because while criterion (b) can be measured with a comparable text similarity metric (e.g. BLEU or edit distance), an adversarial example should also:
be fluent or natural;
preserve its original label.
These two additional criteria are generally irrelevant for images, as adding minor perturbations to an image is unlikely to: (1) create an uninterpretable image (where else changing one word in a sentence can render a sentence incoherent), or (2) change how we perceive the image, say from seeing a panda to a gibbon (but a sentence's sentiment can be reversed by simply adding a negative adverb such as not). Without considering criterion (d), generating adversarial examples in NLP would be trivial, as the model can learn to simply replace a positive adjective (amazing) with a negative one (awful) to attack a sentiment classifier.
To the best of our knowledge, most studies on adversarial example generation in NLP have largely ignored these additional criteria BIBREF2, BIBREF3, BIBREF4, BIBREF5. We believe the lack of a rigorous evaluation framework partially explains why adversarial training for NLP models has not seen the same extent of improvement compared to CV models. As our experiments reveal, examples generated from most attacking methods are successful in fooling the classifier, but their language is often unnatural and the original label is not properly preserved.
The core contribution of our paper is to introduce a systematic, rigorous evaluation framework to assess the quality of adversarial examples for NLP. We focus on sentiment classification as the target task, as it is a popular application that highlights the importance of criteria discussed above. We test a number of attacking methods and also propose an alternative approach (based on an auto-encoder) for generating adversarial examples. We learn that a number of factors can influence the performance of adversarial attacks, including architecture of the classifier, sentence length and input domain.
In the context of images, a good adversarial example is typically defined according to two criteria: (a) it has successfully fooled the target classifier; (b) it is visually similar to the original example.
|
What are the criteria for a good adversarial example in the context of images?
|
(a) it has successfully fooled the target classifier; (b) it is visually similar to the original example.
|
1912.13072
| true
| null |
Implementation & Models Parameters. For all our tasks, we use the BERT-Base Multilingual Cased model released by the authors . The model is trained on 104 languages (including Arabic) with 12 layer, 768 hidden units each, 12 attention heads, and has 110M parameters in entire model. The model has 119,547 shared WordPieces vocabulary, and was pre-trained on the entire Wikipedia for each language. For fine-tuning, we use a maximum sequence size of 50 tokens and a batch size of 32. We set the learning rate to $2e-5$ and train for 15 epochs and choose the best model based on performance on a development set. We use the same hyper-parameters in all of our BERT models. We fine-tune BERT on each respective labeled dataset for each task. For BERT input, we apply WordPiece tokenization, setting the maximal sequence length to 50 words/WordPieces. For all tasks, we use a TensorFlow implementation. An exception is the sentiment analysis task, where we used a PyTorch implementation with the same hyper-parameters but with a learning rate $2e-6$.
We presented AraNet, a deep learning toolkit for a host of Arabic social media processing. AraNet predicts age, dialect, gender, emotion, irony, and sentiment from social media posts. It delivers state-of-the-art and competitive performance on these tasks and has the advantage of using a unified, simple framework based on the recently-developed BERT model. AraNet has the potential to alleviate issues related to comparing across different Arabic social media NLP tasks, by providing one way to test new models against AraNet predictions. Our toolkit can be used to make important discoveries on the wide region of the Arab world, and can enhance our understating of Arab online communication. AraNet will be publicly available upon acceptance.
For all tasks, we use a TensorFlow implementation.
AraNet predicts age, dialect, gender, emotion, irony, and sentiment from social media posts. It delivers state-of-the-art and competitive performance on these tasks and has the advantage of using a unified, simple framework based on the recently-developed BERT model.
|
Did they experiment on all the tasks?
|
Yes.
|
1808.03986
| false
| null |
The comparison of our method with various baselines and state-of-the-art methods is provided in table TABREF26 for VQA 1.0 and table TABREF27 for VQG-COCO dataset. The comparable baselines for our method are the image based and caption based models in which we use either only the image or the caption embedding and generate the question. In both the tables, the first block consists of the current state-of-the-art methods on that dataset and the second contains the baselines. We observe that for the VQA dataset we achieve an improvement of 8% in BLEU and 7% in METEOR metric scores over the baselines, whereas for VQG-COCO dataset this is 15% for both the metrics. We improve over the previous state-of-the-art BIBREF35 for VQA dataset by around 6% in BLEU score and 10% in METEOR score. In the VQG-COCO dataset, we improve over BIBREF5 by 3.7% and BIBREF36 by 3.5% in terms of METEOR scores.
We improve over the previous state-of-the-art BIBREF35 for VQA dataset by around 6% in BLEU score and 10% in METEOR score. In the VQG-COCO dataset, we improve over BIBREF5 by 3.7% and BIBREF36 by 3.5% in terms of METEOR scores.
|
What were the previous state of the art benchmarks?
|
The answers are shown as follows:
* BIBREF35 for VQA dataset
* BIBREF5
* BIBREF36
|
null | false
| null |
Write about apple
|
Apple is a fruit.
In history, there are three apples that profoundly changed the world.
- Eve's apple
- Newton's apple
- Steve Jobs' Apple
Biblical Adam and Eve ate the forbidden fruit and committed the original sin.
Newton discovered the law of gravity when an apple fell from tree.
Steve Jobs co-founded Apple and several products like the iPhone changed the world.
|
|
null | false
| null |
Why are protein waffles so dry?
|
Protein waffles lack butter causing them to be incredibly dry. Butter is flavorful, you should make waffles with butter and forget this protein waffle nonsense.
|
|
null | false
| null |
Major Himalayan-origin rivers that substantially flow through India include the Ganges and the Brahmaputra, both of which drain into the Bay of Bengal. Important tributaries of the Ganges include the Yamuna and the Kosi; the latter's extremely low gradient, caused by long-term silt deposition, leads to severe floods and course changes. Major peninsular rivers, whose steeper gradients prevent their waters from flooding, include the Godavari, the Mahanadi, the Kaveri, and the Krishna, which also drain into the Bay of Bengal; and the Narmada and the Tapti, which drain into the Arabian Sea. Coastal features include the marshy Rann of Kutch of western India and the alluvial Sundarbans delta of eastern India; the latter is shared with Bangladesh. India has two archipelagos: the Lakshadweep, coral atolls off India's south-western coast; and the Andaman and Nicobar Islands, a volcanic chain in the Andaman Sea.
|
Extract name and the type of water body from this text and list them using bullets in the format {Name}-{River}
|
Ganges - River
Brahmaputra - River
Bay of Bengal - Sea
Yamuna - River
Kosi - River
Godavari - River
Mahanadi - River
Kaveri - River
Krishna - River
Narmada - River
Tapti - River
Andaman Sea - Sea
|
null | false
| null |
The Eurovision Song Contest 1987 was the 32nd edition of the annual Eurovision Song Contest. It took place in Brussels, Belgium, following the country's victory at the 1986 contest with the song "J'aime la vie" by Sandra Kim. Organised by the European Broadcasting Union (EBU) and host broadcaster Radio-télévision belge de la Communauté française (RTBF), the contest was held at the Centenary Palace on 9 May 1987 (also Europe Day) and hosted by French-Belgian singer Viktor Lazlo.
Twenty-two countries took part in the contest with Greece and Italy returning to the competition after their absences the previous year. This set the record for the highest number of competing countries up until that point.
The winner was Ireland with the song "Hold Me Now" by Johnny Logan, who had also won the 1980 contest. He remains the only performer to have won the Eurovision Song Contest twice.
|
Given a reference text about the Eurovision Song Contest 1987, tell me where it took place, and how many countries participated.
|
The Eurovision Song Contest 1987 took place in Brussels, Belgium and twenty-two countries participated.
|
null | false
| null |
What is the total number of goals scored by this player when he played for the Sevilla team?
|
https://en.wikipedia.org/wiki/Wissam_Ben_Yedder#Club
|
38
|
null | false
| null |
Show me the current P4P boxer ratings in the world
|
Here's a list of P4P boxing rankings as of March 23
1 Saul Alvarez
2 Dmitrii Bivol
3 Terence Crawford
4 Naoya Inoue
5 Gennadiy Golovkin
6 Shakur Stevenson
7 Errol Spence Jr
8 Stephen Fulton
9 Brandon Figueroa
10 Tyson Fury
|
|
null | false
| null |
Which airport should I fly if I go to San Francisco Bay Area?
|
There are 3 major airports in San Francisco Bay Area: San Francisco International Airport (SFO), Oakland International Airport (OAK), and San Jose International Airport (SJC). They are located in different areas in Bay Area and serves different neighborhoods.
If you are flying to San Francisco or Peninsula area, San Francisco Internal Airport is the most convenient airport.
If you go to East Bay, such as UC Berkeley or Oakland, Oakland International Airport is the best option.
If you go to South Bay, such as Palo Alto, Mountain View, San Jose, then San Jose International Airport would be your destination.
|
|
null | false
| null |
There are two subgroups of austenitic stainless steel. 300 series stainless steels achieve their austenitic structure primarily by a nickel addition while 200 series stainless steels substitute manganese and nitrogen for nickel, though there is still a small nickel content.
|
Given this paragraph about stainless steel, which of the two series contains the most nickel content?
|
The 300 series contains the most nickel content.
|
null | false
| null |
Explain to your 6 year son why its ok to fail few times only to get better later
|
dear son, when you were born you were so small, you didnt know how to eat or sit. But mommy kept trying to feed you and slowly slowly you learned to eat, and now you see you eat food by yourself so fast. There are so many things you have learned like talking, reading, writing, it all came from practise. Remember the first time when you did cycling you could not balance and kept falling? But Daddy told you to keep trying again and again and you kept practicing every Saturday and now you know how to ride cycle and balance well. Always remember this, everytime you do something new, it will be difficult at first, it might not work, but you should not lose hope and be sad and give up. You should not think you cant do this. Remember if you keep trying , you will figure it out. So next you find something difficult, stay calm, keep trying and you will get it done.
|
|
null | false
| null |
Which of the following bodies of water are below sea level: Dead Sea, Lake Ontario, Sea of Galilee, the Hudson Bay, Lake Assal?
|
The Dead Sea, the Sea of Galilee, and Lake Assal are all below sea level.
|
|
null | false
| null |
Over the past 3,000 years, the city of Beijing has had numerous other names. The name Beijing, which means "Northern Capital" (from the Chinese characters 北 for north and 京 for capital), was applied to the city in 1403 during the Ming dynasty to distinguish the city from Nanjing (the "Southern Capital"). The English spelling Beijing is based on the government's official romanization (adopted in the 1980s) of the two characters as they are pronounced in Standard Mandarin. An older English spelling, Peking(北平 in Chinese meaning), was used by Jesuit missionary Martino Martini in a popular atlas published in Amsterdam in 1655. Although Peking is no longer the common name for the city, some of the city's older locations and facilities, such as Beijing Capital International Airport, with the IATA Code PEK, and Peking University, still retain the former romanization.
|
Given this paragraph about the history of Beijing, when did Beijing get the current name?
|
In 1403, during the Ming dynasty, Beijing got its current name.
|
null | false
| null |
Lasantha Rodrigo (born 28 May 1938) is a former cricketer who played 14 matches of first-class cricket for Ceylon between 1959 and 1971.
Life and career
Lasantha Rodrigo was born in Moratuwa and attended Prince of Wales' College, Moratuwa, where he captained the cricket team in 1958 and 1959, and also captained the Ceylon schools team. His father, J. B. C. Rodrigo, was principal of Prince of Wales' College from 1933 to 1959.
He made his highest first-class score on his debut, in the Gopalan Trophy match in 1958–59, when he scored 89. Batting at number three, he top-scored with 44 for Ceylon in their one-day match against the touring Australians in April 1961. He toured India with the Ceylon team in 1964-65, playing in all three matches against India, but with only moderate success.
He worked for Ceylon Cold Stores for 31 years as a mechanical engineer. Inability to take time off work to play cricket shortened his cricket career. In 2014 he was formally honoured by Sri Lanka Cricket for his services to cricket in Sri Lanka, and awarded 300,000 rupees. In September 2018, he was one of 49 former Sri Lankan cricketers felicitated by Sri Lanka Cricket, to honour them for their services before Sri Lanka became a full member of the International Cricket Council (ICC).
Rodrigo lives with his wife Sweenie in Moratuwa, opposite Prince of Wales' College. They have a daughter and a son.
|
Given a reference text about Lasantha Rodrigo, tell me how long he worked as a mechanical engineer and who worked for.
|
Lasantha Rodrigo worked as a mechanical engineer for Ceylon Cold Stores for 31 years.
|
null | false
| null |
What is an audible?
|
An audible is a football term for changing the play call on the fly. A quarterback will read the defense and the coverage and will make a determination if the play called will be an effective one. If they decide the play will not work well against what the defense is showing they will call an audible to a play with a higher probability of success.
|
|
null | false
| null |
John Henry Moore II (August 5, 1927 – July 19, 2013) was an American lawyer and United States district judge of the United States District Court for the Middle District of Florida.
Education and career
Born in Atlantic City, New Jersey, he received his Bachelor of Science degree from Syracuse University in 1952 and his Juris Doctor from the University of Florida College of Law in 1961. Moore served in the Naval Reserve from 1948 to 1971, and served on active duty in Korea before retiring with the rank of commander. Moore was in private practice in Atlanta, Georgia in 1961 with the firm of Fisher and Phillips before moving to Fort Lauderdale, Florida where he was a partner with the firms of Fleming, O'Bryan and Fleming and then Turner, Shaw and Moore from 1961 to 1967. Moore also served as attorney for the School Board of Broward County.
State judicial service
Moore served as a judge of the 17th Judicial Circuit of Florida from 1967 to 1977, serving as chief judge of that court from 1975 to 1977. Moore was nominated to United States District Court for the Southern District of Florida in 1976, but the Senate took no action. Moore served as a judge on the Florida Fourth District Court of Appeal from 1977 to 1981.
|
Tell me about John Henry Moore II
|
John Henry Moore II was an American lawyer and United States district judge of the United States District Court for the Middle District of Florida. He received his Bachelor of Science degree from Syracuse University in 1952 and his Juris Doctor from the University of Florida College of Law in 1961. Moore served in the Naval Reserve from 1948 to 1971, and served on active duty in Korea before retiring with the rank of commander. Moore served as a judge of the 17th Judicial Circuit of Florida from 1967 to 1977, serving as chief judge of that court from 1975 to 1977.
|
null | false
| null |
What are the best beach family vacation spots on the east coast?
|
International: Aruba, Mexico, Dominican Republic. Domestic: Florida, Texas, outerbanks in North Carolina
|
|
null | false
| null |
Who is the patron saint of traveling?
|
Saint Christopher is the patron saint of travelers, including motorists, who sometimes hang a small image of him in their vehicle for luck. As legend tells us, he was crossing a river when a child asked to be carried across.
|
|
null | false
| null |
The Vedas (/ˈveɪdə(z)/ or /ˈviːdə(z)/, IAST: veda, Sanskrit: वेदः, lit. 'knowledge') are a large body of religious texts originating in ancient India. Composed in Vedic Sanskrit, the texts constitute the oldest layer of Sanskrit literature and the oldest scriptures of Hinduism.
There are four Vedas: the Rigveda, the Yajurveda, the Samaveda and the Atharvaveda. Each Veda has four subdivisions – the Samhitas (mantras and benedictions), the Aranyakas (text on rituals, ceremonies, sacrifices and symbolic-sacrifices), the Brahmanas (commentaries on rituals, ceremonies and sacrifices), and the Upanishads (texts discussing meditation, philosophy and spiritual knowledge). Some scholars add a fifth category – the Upasanas (worship). The texts of the Upanishads discuss ideas akin to the heterodox sramana-traditions.
|
What are Vedas?
|
Vedas are a religious texts from ancient India written in Sanskrit. There are primarily 4 vedas - Rigveda, Yajurveda, Samaveda and Atharvaveda.
|
null | false
| null |
Which characters belong to DC or Marvel Universe? Kent Nelson, Doctor Strange
|
Kent Nelson is DC, Doctor Strange is Marvel
|
|
null | false
| null |
Identify the musical interval (e.g perfect fifth, major third) associated with the following pairs of notes:
C-G
G-B
E-A
C-Fsharp
Eflat-C
D-Csharp
B-F
|
C-G : perfect fifth
G-B: major third
E-A: perfect fourth
C-Fsharp: augmented fourth
Eflat-C: major sixth
D-Csharp: major seventh
B-F: diminished fifth
|
|
null | false
| 377
|
The Recognizing Textual Entailment (RTE) challenges first appeared in 2004 as a means to test textual entailment, i.e. relations between a premise text and a hypothesis text ( "Recognizing Textual Entailment" ):
An entailment example from RTE1.
Budapest again became the focus of national political drama in the late 1980s, when Hungary led the reform movement in eastern Europe that broke the communist monopoly on political power and ushered in the possibility of multiparty politics.
In the late 1980s Budapest became the center of the reform movement.
Entailment [RTE702]
In contrast to the FraCaS test suite, the RTE challenges use naturally occurring data as premises. The hypothesis text is then constructed based on this premise text. There is either a binary or a tripartite classification of entailment — depending on the version of RTE. The first two RTE challenges follow the former scheme and make a binary classification of entailment (entailed or not entailed). Tripartite classification (entailment, negation of the hypothesis entailment or no entailment) is added in the later datasets, retaining two way classification versions as well. Seven RTE challenges have been created altogether.
The main advantages of the RTE challenges is their use of examples from natural text and the inclusion of cases that require presupposed information, mostly world knowledge. Indeed, the very definition of inference assumed in a number of the examples is problematic. As BIBREF1 have pointed out, RTE platforms suffer from cases of inference that should not be categorized as such. For these cases, a vast amount of world knowledge needs to be taken into consideration (that most importantly not every linguistic agent has). In this paper, and having the RTE as our starting point challenges, we claim that RTE is insufficiently precise to perform logical reasoning or precise reasoning tasks and we take up the task of validating our working hypothesis and proposing a method for doing proper collection of precise entailment pairs in the style of RTE. Of course, the creators of RTE had in mind a more loose definition of inference where both a precise and an imprecise definition of entailment would be at play. BIBREF2 mention that “our applied notion of textual entailment is also related, of course, to classical semantic entailment in the linguistics literature... a common definition of entailment specifies that a text t entails another text h (hypothesis, in our terminology) if h is true in every circumstance (possible world) in which t is true." This is close to what we want to capture in this paper. But, at the same time, BIBREF2 also mention that “however, our applied definition allows for cases in which the truth of the hypothesis is highly plausible, for most practical purposes, rather than certain". It is these cases we want to make more precise, in the sense of making the supporting hidden inferences that are at play in many of the RTE examples explicit. In a way, what we are aiming at is a methodology of constructing entailment datasets in the style of RTE, that will involve a more precise definition of entailment and will further record any missing/hidden premises are used in justifying or not an entailment pattern.
The first two RTE challenges follow the former scheme and make a binary classification of entailment (entailed or not entailed).
|
Do the first two RTE challenges make a binary classification of entailment?
|
Yes, they do.
|
null | false
| null |
Why do people watch the English Premier League more than any other soccer league in the world?
|
The English Premier League, which I've been watching for the past 20 years, is one of the fastest and most entertaining leagues in international football.
The Premier League provides us with a full dose of excitement and the most physical league in the world. The rules for fouls aren't as strict as they are in other leagues, which leads to more physicality and allows for more tackles, rougher games, and a huge variation in footballing styles amongst the teams.
Compare this with other leagues in Europe, and you'll see that the teams are far more technical and tactical. Any Premier League match you watch, you'll have some kind of action.
Another advantage of watching the Premier League is that unlike, say, La Liga or the Bundesliga, there aren't just two teams dominating the whole league, there are at least 6 teams in the EPL vying for 4 Champions League spots, which is a lot of competition. Games tend to be more brutal for this reason.
|
|
null | false
| null |
What is the objective of hockey?
|
Ice hockey is a sport played by 2 opposing teams. The objective is for a team to score more goals than their opponents. Goals are scored by shooting a puck into the opposing team's net.
|
|
null | false
| 102
|
Intuitively, when a human writes a description from a table she keeps track of information at two levels. At the macro level, it is important to decide which is the appropriate field to attend to next and at a micro level (i.e., within a field) it is important to know which values to attend to next. To capture this behavior, we use a bifocal attention mechanism as described below.
Macro Attention: Consider the INLINEFORM0 -th field INLINEFORM1 which has values INLINEFORM2 . Let INLINEFORM3 be the representation of this field in the infobox. This representation can either be (i) the word embedding of the field name or (ii) some function INLINEFORM4 of the values in the field or (iii) a concatenation of (i) and (ii). The function INLINEFORM5 could simply be the sum or average of the embeddings of the values in the field. Alternately, this function could be a GRU (or LSTM) which treats these values within a field as a sequence and computes the field representation as the final representation of this sequence (i.e., the representation of the last time-step). We found that bidirectional GRU is a better choice for INLINEFORM6 and concatenating the embedding of the field name with this GRU representation works best. Further, using a bidirectional GRU cell to take contextual information from neighboring fields also helps (these are the orange colored cells in the top-left block in Figure FIGREF3 with macro attention). Given these representations INLINEFORM7 for all the INLINEFORM8 fields we compute an attention over the fields (macro level). DISPLAYFORM0
where INLINEFORM0 is the state of the decoder at time step INLINEFORM1 . INLINEFORM2 and INLINEFORM3 are parameters, INLINEFORM4 is the total number of fields in the input, INLINEFORM5 is the macro (field level) context vector at the INLINEFORM6 -th time step of the decoder.
Micro Attention: Let INLINEFORM0 be the representation of the INLINEFORM1 -th value in a given field. This representation could again either be (i) simply the embedding of this value (ii) or a contextual representation computed using a function INLINEFORM2 which also considers the other values in the field. For example, if INLINEFORM3 are the values in a field then these values can be treated as a sequence and the representation of the INLINEFORM4 -th value can be computed using a bidirectional GRU over this sequence. Once again, we found that using a bi-GRU works better then simply using the embedding of the value. Once we have such a representation computed for all values across all the fields, we compute the attention over these values (micro level) as shown below : DISPLAYFORM0
where INLINEFORM0 is the state of the decoder at time step INLINEFORM1 . INLINEFORM2 and INLINEFORM3 are parameters, INLINEFORM4 is the total number of values across all the fields.
Fused Attention: Intuitively, the attention weights assigned to a field should have an influence on all the values belonging to the particular field. To ensure this, we reweigh the micro level attention weights based on the corresponding macro level attention weights. In other words, we fuse the attention weights at the two levels as: DISPLAYFORM0
where INLINEFORM0 is the field corresponding to the INLINEFORM1 -th value, INLINEFORM2 is the macro level context vector.
We found that bidirectional GRU is a better choice for f and concatenating the embedding of the field name with this GRU representation works best. Further, using a bidirectional GRU cell to take contextual information from neighboring fields also helps (these are the orange colored cells in the top-left block in Figure 2 with macro attention).
|
Why is GRU chosen in the method?
|
GRU is a better choice for f and concatenating the embedding of the field name with this GRU representation works best. Further, using a bidirectional GRU cell to take contextual information from neighboring fields also helps.
|
null | false
| null |
When was the first Academy Awards?
|
The first Academy Awards was Thursday, May 16, 1929
|
|
null | false
| null |
how to solve a Rubik's cube
|
Start with one side: Begin by choosing one side of the cube to solve. Usually, people choose the white or yellow side as it's easier to see the colors. Look for a center piece of the color you've chosen and place it in the center of that side.
Build the first layer: Now, turn the cube so that the side you just solved is facing down. Look for the edges that match the center color and put them in their respective places on the first layer. This layer will be completed when you have all four matching edge pieces in place.
Build the second layer: Once the first layer is complete, turn the cube over and find the middle row. Look for the edge pieces that belong on this layer, and move them into place with an algorithm. The goal is to build a complete layer with the same color as the center.
Build the final layer: The final layer can be tricky, but with practice, it becomes easier. First, align the corners of the final layer with the corresponding edges. Then, apply an algorithm to move the corners to their correct positions. Once the corners are in place, use another algorithm to position the edges.
Finish the cube: After completing the final layer, you may have some misaligned pieces. Don't worry, as a simple algorithm can fix them. Keep using the algorithm until all the sides of the cube are in place.
With practice, solving a Rubik's Cube can be achieved in under a minute. It's important to remember that it's all about pattern recognition and memorizing algorithms, so keep practicing and soon you'll be able to solve it without even thinking!
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.