Datasets:

lp
stringclasses
3 values
src
stringclasses
415 values
mt
stringlengths
1
388
⌀
ref
stringlengths
3
370
raw
int64
0
100
annotators
int64
1
1
domain
stringclasses
1 value
year
int64
2.02k
2.02k
task
stringclasses
2 values
sys_id
int64
0
34
en-zh
And we can easily incorporate some prior knowledge as constraint which can help improve the performance.
我们可以很容易地结合一些先前的知识作为约束,这可以帮助提高性能。
我们可以简单地将一些先前的知识作为约束,这样就可以帮助提高表现。
93
1
ACL
2,023
offline
34
en-zh
Which must be left to Jira, the issue tracker system, and can only be applied to projects that use Jira.
它必须连接到JiLO ,也就是问题标记系统,并且只能应用到使用JiLO的项目中。
这必须留给问题跟踪器系统 JIRA ,并且只能应用于使用JIRA的项目。
57
1
ACL
2,023
offline
11
en-zh
This penalty results in a lower BLEU value in the experiment results described next.
个定罪导致了下一个实验结果中一个超额的句子。
在接下来描述的实验结果中,这种惩罚导致了较低的BLEU值。
67
1
ACL
2,023
offline
25
en-zh
Lastly, statutory articles aren't small paragraphs which usually is the typical retrieval unit in most retrieval works.
最后,法律条款是小段落,通常是大多数检索工作中典型的检索单元。
最后,法定条款不是小段落,而小段落通常是大多数检索作品中的典型检索单元。
50
1
ACL
2,023
offline
25
en-zh
So, we need to ah feature extraction phase ah which includes text analysis.
所以,我们需要一个特征提取阶段,包括文本分析。
因此,我们需要进行特征提取阶段,其中包括文本分析。
87
1
ACL
2,023
offline
6
en-zh
The first is a system called ARENA released in twenty fourteen.
第一个是2014年发布的名为Arena的系统。
第一个是被称为ARENA的系统 ,发布于2014年。
96
1
ACL
2,023
offline
34
en-zh
And around eighty percent of these cited articles come from either the civil code, judicial codes, criminal investigation codes or penal codes.
大约80%的被引用文章来自民法、司法法规、刑事调查法规或刑法。
这些被引用的条款中约有80%来自民法典、司法法典、刑事调查法典或刑法典。
100
1
ACL
2,023
offline
28
en-zh
For instance, the question, can I evict my tenants if they make too much noise?
例如,这个问题,“如果我的租户制造了太多噪音,我可以驱逐他们吗?”
例如这个问题:如果我的租户制造太多噪音,我可以驱逐他们吗?
100
1
ACL
2,023
offline
28
en-zh
Two of our most important evaluation metrics are the accuracy of the models in classifying image sentence pairs into captions and foils.
我们最重要的两个评估指标是将图像句子对分类为字幕和文件的模型的准确性。
我们最重要的两个评估指标是模型在将图像句子对分类为标题和干扰词的准确性。
79
1
ACL
2,023
offline
33
en-zh
Dataset fed into FeSTE.
null
数据集送入FeSTE。
0
1
ACL
2,023
offline
17
en-zh
In the figure on the right, the reference sentence has three or four sentences, while CAS has only one.
在右边的图中,参考句子有三四个句子,而CAS只有一个句子。
在右图中,引用句子有3或4个句子 ,而CAS只有1个。
95
1
ACL
2,023
offline
28
en-zh
And we can also see the number of all the possible ah expression is different from the previous step.
我们还可以看到,所有可能的表达式的数量与前一步不同,所以
我们还可以看到,所有可能的表达式的数字与之前的步骤不同。
95
1
ACL
2,023
offline
17
en-zh
When each entity, in this example the university name, is linked to an entity within the knowledge base.
在每个实体中,在这个例子中,大学的名字与知识库中的一个实体相
当每个实体,在这个例子中是指大学名称,被链接到知识库中的一个实体。
87
1
ACL
2,023
offline
28
en-zh
So in a formal deductive system, at a time step t, we apply the operator between the q_i and q_j pair, and then we obtain this new expression.
在一个形式演绎系统中,在时间t ,我们在qi和qj对之间应用算子,然后我们得到这个新的表达式。
在形式演绎系统中,在时间步骤t ,我们在q_i和q_j对之间应用运算符,然后我们得到这个新表达式。
73
1
ACL
2,023
offline
34
en-zh
We add it to the next state to become a new quantity.
我们把它添加到下一个状态,成为一个新的数量。所以,
我们把它添加到下一个状态,成为一个新的数量。
90
1
ACL
2,023
offline
6
en-zh
For instance, the question, can I evict my tenants if they make too much noise?
例如,这个问题,“如果我的租户制造了太多噪音,我可以驱逐他们吗?”
例如这个问题:如果我的租户制造太多噪音,我可以驱逐他们吗?
75
1
ACL
2,023
offline
6
en-zh
This image is a sample usage that returns a corrective or bug fixes label.
图像是一个样本用法,返回纠正或错误修复标签。
此图像是一个返回纠正或错误修复标签的示例用法。
62
1
ACL
2,023
offline
6
en-zh
In particular, on the clean test set, the score gap between the proposed method and the baselines jumped to more than twenty points.
特别是,在绿色测试中,提议方法和基准之间的平方差距跳跃到超过20点。
特别是在干净的测试集上,建议的方法和基线之间的分数差距跃升到20分以上。
76
1
ACL
2,023
offline
25
en-zh
First, I will introduce automatic release note generation that we are working on in this research.
首先,我将介绍我们在这项研究中正在研究的自动减少营养。 这
首先,我将介绍我们在这项研究中正在进行的自动发行说明生成。
70
1
ACL
2,023
offline
25
en-zh
The two-tower model improves over its Siamese variants on recall at one hundred, but performs similarly on the other metrics.
双塔模型在100的回忆率上优于其Siamese变体,但在其他指标上表现相似。
双塔模型在召回率上比其siamese模型高出了100,但在其他指标上表现相似。
69
1
ACL
2,023
offline
34
en-zh
The results with pairwise accuracy are shown here and they are consistent with the results we got from the other metrics is that the best zero shot performance is achieved by ViLBERT twelve in one, followed by ViLBERT, LXMert, CLIP, and finally VisualBERT.
成对准确度的结果在这里显示,它们与我们从其他度量标准得到的结果是一致的。最好的零镜头性能是由Wilbert 12-in-1实现的,其次是Wilbert , AlexMert , Clip ,最后是VisualBERT 。
这里显示了成对的准确度的结果,它们与我们从其他指标中得到的结果一致,即ViLBERT十二分之一取得了最佳的零样本表现,其次是ViLBERT、LXMert、CLIP,最后是VisualBERT。
84
1
ACL
2,023
offline
34
en-zh
So here our model actually makes a wrong prediction at the first step.
在这里,我们的模型实际上在第一步做出了错误的预测。
在这里,我们的模型实际上在第一步就做出了错误的预测。
86
1
ACL
2,023
offline
25
en-zh
So here if we look at this expression, eight times three plus three is actually generated twice, but in fact we should reuse the results.
如果我们看这个表达式, “ 8乘以3 + 3 ”实际上产生了两次,但实际上我们应该重用结果。因此,
如果我们看一下这个表达式,八乘三加三实际上生成两次,但实际上我们应该重复使用结果。
77
1
ACL
2,023
offline
17
en-zh
Ah, if we have a large number of operators or constants, the memory consumption could be pretty high.
如果我们有少量操作员, 或者随机存量, 记忆消耗可能会很高,
如果我们有大量 数量的运算符或常量,内存消耗可能相当高。
92
1
ACL
2,023
offline
5
en-zh
I will be presenting my joint work with Jerry which is about a New Dataset for Statutory Article Retrieval.
我将和杰里一起展示我的合作成果,这是关于一个新的法定文章回收数据集。
我将展示我与Jerry的合作结果,它是针对法定条款检索的新数据集。
60
1
ACL
2,023
offline
11
en-zh
MTDNN maintained number of classes, heads, output layers.
空的DNA, 保留了大量的类别头, 输出文件夹
MTDNN保持了类、头、输出层的数量。
24
1
ACL
2,023
offline
5
en-zh
Each of them has been labeled by experienced jurists with references to relevant articles from a corpus of more than twenty-two thousand six hundred legal articles from Belgian codes of law.
每个人都被经验丰富的法学家标记过,他们参考了比利时法规中超过22600条法律条款的相关文章。
每个问题都由经验丰富的法学家标记,并参考了比利时法典中超过22,600篇法律条款的语料库中的相关条款。
80
1
ACL
2,023
offline
28
en-zh
Here are an error analysis.
英雄和错误的护士。
这里有一个错误分析。
0
1
ACL
2,023
offline
25
en-zh
Regarding evaluation, in some cases, release notes are output in multiple sentences.
在数学句子中, 由于很难区分句子的数目, 它
关于评估,在某些情况下,发行说明是以多个句子形式输出的。
31
1
ACL
2,023
offline
5
en-zh
Then, we generate the new features and evaluate them using five evaluation classifiers.
然后,我们生成新的特征,并使用五个评估分类器对它们进行评估。
然后,我们生成新的特征,并使用五个评估分类器对其进行评估。
72
1
ACL
2,023
offline
17
en-zh
During the dataset construction, all references to these uncollected articles are ignored, which causes some questions to end up with only a fraction of the initial number of relevant articles.
在数据集的构建过程中,所有对这些未编辑文章的引用都被忽略了,这导致一些问题最终只得到了相关文章初始数量的一小部分。
在构建数据集的过程中,所有对这些未收集的条款的引用都被忽略,这导致一些问题最终只有最初相关条款数量的一小部分。
90
1
ACL
2,023
offline
32
en-zh
And we can also refer back to the original questions to find the relevant contents.
我们也可以回到最初的问题,找到相关的内容,
我们也可以回头参考原始问题来查找相关内容。
86
1
ACL
2,023
offline
25
en-zh
During the dataset construction, all references to these uncollected articles are ignored, which causes some questions to end up with only a fraction of the initial number of relevant articles.
在数据集构建过程中, 对这些未被收集的文章的参考 都会被忽略, 这就导致了一些问题, 最终只剩初始可引用的文章数量的一小部分。
在构建数据集的过程中,所有对这些未收集的条款的引用都被忽略,这导致一些问题最终只有最初相关条款数量的一小部分。
90
1
ACL
2,023
offline
5
en-zh
For example, the tenants makes two parties a week until two AM.
例如,租户每周举行两次聚会,直到凌晨2点。
例如,租户每周举行两次派对,直到凌晨2点。
100
1
ACL
2,023
offline
11
en-zh
So, finally we want to show the interpretability through a question perturbation example.
最后,我们想通过一个碰撞和参与的例子来展示可解释性。
最后,我们想通过一个问题扰动的例子来展示可解释性。
58
1
ACL
2,023
offline
34
en-zh
Although promising, these results suggest ample opportunity for improvement compared to a skilled legal expert who can eventually retrieve all relevant articles to any question and thus get perfect scores.
个任务。尽管这些结果很有希望,但与技能法律专家相比,这些专家最终可以将所有相关文章检索到任何问题上,从而获得完美的
虽然有希望,但这些结果表明,与一个熟练的法律专家相比,还有很多改进的机会,因为他最终可以检索到任何问题的所有相关条款,从而获得满分。
60
1
ACL
2,023
offline
32
en-zh
And in these steps we obtain the divisors.
在这些步骤中, 我们使用了设备,
在这些步骤中,我们得到了除数。
57
1
ACL
2,023
offline
5
en-zh
A higher specificity means that the model correctly outputs an empty text in cases where the release notes assume empty.
高特异性意味着在释放节点为空的情况下,模型校正输出为空文本。
更高的特异性意味着,在发行说明假定为空的情况下,模型会正确地输出一个空文本。
76
1
ACL
2,023
offline
34
en-zh
You can check out our paper, dataset and code at the following links. Thank you.
你可以查看以下这些 代码的文件: 谢谢。
您可以在以下链接查看我们的论文、数据集和法典。谢谢。
70
1
ACL
2,023
offline
5
en-zh
For more metrics and results on them, do check out our paper.
对称 是否能检查我们的纸质。
有关更多指标及其结果,请查看我们的论文。
20
1
ACL
2,023
offline
5
en-zh
So, after filtering and human evaluation, we have as many test instances as described in this table.
所以,在过滤和人类评估之后,我们有了和这张表格中描述的一样多的测试实例。
因此,经过过滤和人工评估后,我们拥有与本表中所述的测试实例一样多的测试实例。
89
1
ACL
2,023
offline
11
en-zh
This is not easy to do because a foiled caption may be less likely than the original caption.
这并不容易做到,因为锡箔标题可能比原来的标题更不可能。
这并不容易,因为被干扰的标题可能比原始标题更不可能。
50
1
ACL
2,023
offline
28
en-zh
And and the text of the entities of the knowledge base is extracted and added to the dataset.
连,知识库的实体的文本被提取并添加到数据集中。
并且知识库的实体文本被提取出来,并添加到数据集中。
89
1
ACL
2,023
offline
6
en-zh
We modeled the classwise abstractive summarization approach by two different methods.
我们用两个精确的方法 模拟了攻击性声波的 横向移
我们通过两种不同的方法对分类抽象总结方法进行建模。
50
1
ACL
2,023
offline
5
en-zh
So let's see the full framework.
让我们来看看完整的框架。一个
让我们来看看完整的框架。
89
1
ACL
2,023
offline
32
en-zh
The second method, method, which we call CAS-Multi, consists of four different seq2seq networks, each of which correspond to one of the fixed release note classes.
第二种方法,我们称之为“她是多”的方法,由四个不同的扇区到扇区网络组成,每个扇区对应一个最小的节点。
第二种方法,方法,我们称之为CAS-Multi,由四个不同的seq2seq网络组成,每个网络都对应于一个固定的发行说明类别。
40
1
ACL
2,023
offline
17
en-zh
So we got a message, the accuracies on these tasks and specific benchmarks are increasing steadily.
所以,我们得到了一个信息,这些任务特定基准的准确性正在稳步增加。
因此,我们得到了一个信息,这些任务的准确性和特定基准正在稳步提升。
85
1
ACL
2,023
offline
11
en-zh
It ah it extracts the text from the knowledge base, which in this example is the abstract of the Wikipedia page.
它从知识基础中提取文本,在这个例子中,这是维基百科页面的抽象。
它从知识库中提取文本,在本例子中,它是维基百科页面的摘要。
64
1
ACL
2,023
offline
32
en-zh
The recent advances in NLP have sparked huge interest in many legal tasks, such as legal judgment prediction or automated contact contract review.
NRP的最新进展激发了人们对许多法律任务的巨大兴趣,比如法律判断预测或自动合同审查。
自然语言处理中的最新进展引发了对许多法律任务的巨大兴趣,例如法律判断预测或自动联系人合同审查。
89
1
ACL
2,023
offline
11
en-zh
We experiment with five vision and language models on VALSE, namely with CLIP, LXMert, ViLBERT, ViLBERT twelve in one, and VisualBERT.
艾利克斯莫特、维尔伯特、 维尔伯特121和维尔伯特,
我们在VALSE上尝试了五种视觉和语言模型,即使用CLIP、LXMert、ViLBERT、ViLBERT十二合一和VisualBERT。
18
1
ACL
2,023
offline
5
en-zh
So, certain assumptions ah also apply as in previous work.
某些假设也适用,就像之前的工作一样,
某些假设也适用于之前的工作。
70
1
ACL
2,023
offline
32
en-zh
Each of them has been labeled by experienced jurists with references to relevant articles from a corpus of more than twenty-two thousand six hundred legal articles from Belgian codes of law.
每一个都被经验丰富的法学家标记为相关文章,这些文章来自比利时法律法规的22600多篇法律文章。
每个问题都由经验丰富的法学家标记,并参考了比利时法典中超过22,600篇法律条款的语料库中的相关条款。
80
1
ACL
2,023
offline
32
en-zh
We also, we also split each dataset into four folds and apply four folds cross validation.
我们也把每个数据单元分成四个折叠,并应用四个折叠的交叉验证。
我们还将每个数据集分成四折,并应用四折交叉验证。
40
1
ACL
2,023
offline
25
en-zh
So, in this example there are four tasks in the training set, so MTDNN maintain four heads as you can see at the image.
在这个例子中,训练集有四个任务。所以MTDNN保持四个领先,正如你在图片中看到的。
因此,在这个例子中,训练集里有四个任务,所以MTDNN维持四个头,正如在图像上看到的那样。
88
1
ACL
2,023
offline
34
en-zh
So, in this example there are four tasks in the training set, so MTDNN maintain four heads as you can see at the image.
在这个例子中,训练集中有四个任务,所以, MT-DNC保持着四个头,就像你在图片中看到的。
因此,在这个例子中,训练集里有四个任务,所以MTDNN维持四个头,正如在图像上看到的那样。
89
1
ACL
2,023
offline
11
en-zh
So let's see an example.
让我们看一个例子,
让我们来看一个例子。
85
1
ACL
2,023
offline
28
en-zh
So, the expression is represented by e i j o p.
表达方式是由eijop表示的,
表达式由e i j o p表示。
95
1
ACL
2,023
offline
11
en-zh
So here if we look at this expression, eight times three plus three is actually generated twice, but in fact we should reuse the results.
如果我们看这个表达式, 8乘以3加3 ,它实际上生成了两次。事实上,我们应该重用结果。所以
如果我们看一下这个表达式,八乘三加三实际上生成两次,但实际上我们应该重复使用结果。
76
1
ACL
2,023
offline
34
en-zh
So in our datasets we are also given the mathematical expression which leads to the ah to this particular answer as well.
在我们的数据集中,我们也被赋予了数学表达,这也导致了这个特定的答案。所以,
在我们的数据集中,还向我们提供了数学表达式,该表达式也导向这个特定的答案。
94
1
ACL
2,023
offline
6
en-zh
We see from the plurality and counting instruments that vision and language models have trouble distinguishing references to single versus multiple objects, or counting them in an image.
我们从多元化和计数工具中看到,视觉和语言模型很难区分单一对象和多个物体的参考,或者在图像中数数它们。
我们从复数和计数工具中看到,视觉和语言模型难以区分对单个和多个对象的引用,或在图像中计算它们。
59
1
ACL
2,023
offline
33
en-zh
Lastly, statutory articles aren't small paragraphs which usually is the typical retrieval unit in most retrieval works.
最后,法定期刊是一小段,这通常是大多数检索工作中的典型检索单位。
最后,法定条款不是小段落,而小段落通常是大多数检索作品中的典型检索单元。
60
1
ACL
2,023
offline
17
en-zh
We collected thousands of questions annotated with categories, subcategories and legal references to relevant statutes.
我们收集了成千上万的问题,用类别、子类和相关法规的法律参考来注释。
我们收集了数以千计的问题,并用类别、子类别和相关法规的法律参考进行了注释。
96
1
ACL
2,023
offline
11
en-zh
And even more, VALSE could be used as an indirect assessment of datasets, as models could be evaluated before and after training or finetuning to see whether a dataset helps models improve on any of the aspects tested by VALSE.
更重要的是, valSE可以被用作对数据集的间接评估,因为模型可以在训练前和训练后进行评估,或者微调,看看数据集是否能帮助模型在valSE测试的任何方面改进。
更重要的是,VALSE可以作为数据集的间接评估,因为可以在训练或微调前后对模型进行评估,以了解数据集是否有助于模型在VALSE测试的任何方面得到改善。
95
1
ACL
2,023
offline
28
en-zh
This system has a simple learning based text classification model and outputs one of five labels such as features or bug fixes for each input commit message.
这个系统有一个简单的基于运行的文本分类模型,并输出了所有五个图书馆,例如功能或公共修复,或每个输入承诺信息。
这个系统有一个简单的基于学习的文本分类模型,并为每个输入的提交信息输出五个标签之一,如特征或错误修复。
80
1
ACL
2,023
offline
33
en-zh
First, it deals with two types of language.
首先,它涉及两种语言:
首先,它涉及两种类型的语言。
100
1
ACL
2,023
offline
25
en-zh
In particular, on the clean test set, the score gap between the proposed method and the baselines jumped to more than twenty points.
特别是,在干净的数据集中,提出的方法和基准线之间的分数差距跃升到了20分以上。
特别是在干净的测试集上,建议的方法和基线之间的分数差距跃升到20分以上。
80
1
ACL
2,023
offline
11
en-zh
Or in other words, we trained the language model to classify an abstract and class ah to abstract and class ah, if the abstract belongs to the class or not.
或者换句话说,我们训练语言模型将抽象和类分类为抽象和类,如果抽象是否属于类的
或者换句话说,我们训练了语言模型,将一个抽象和类划分为抽象和类,无论抽象属不属于类。
70
1
ACL
2,023
offline
34
en-zh
So we conduct experiments on the commonly used math word problem datasets, MAWPS, Math23K, MathQA and SVAMP.
我们对常用的问题数据集方法进行了实验, MAWPS, Math23k, MathQA ,和SvaMP ,
我们对常用的数学文字问题 数据集、MAWPS、Math23K、MathQA和SVAMP进行实验。
95
1
ACL
2,023
offline
11
en-zh
All right. So, the best approaches are often tree based model.
好吧。所以,最好的方法通常是基于树的模型。所以,
没错。最好的方法通常是基于树的模型。
77
1
ACL
2,023
offline
11
en-zh
But if the dataset has five classes, FeSTE generates five new features.
但如果数据集有五类,那么FeSTE就产生了五个新的特征,
但如果数据集有五个类,FeSTE就会生成五个新的特征。
62
1
ACL
2,023
offline
28
en-zh
We created a heuristic matching rule to get the previous and next versions.
我们创造了一个“人脸”的标识, 来得到我们下一个病人,
我们创建了一个启发式匹配规则来获取上一个和下一个版本。
20
1
ACL
2,023
offline
5
en-zh
So on the left hand side, we can see if we give some examples with just question and answers, we might not be able to obtain the correct answers.
上,我们可以看到, 如果我们给一些样本 只是破折号和答案, 我们可能无法得到正确的答案,
在左侧我们可以看到,如果我们给出一些只有问题和答案的例子,我们可能无法获得正确的答案。
56
1
ACL
2,023
offline
5
en-zh
So, previous work in math word problem solving ah actually can ah be categorized into sequence to sequence and sequence to tree model.
之前在数学单词问题解决方面的工作实际上可以分类为序列到序列和序列到树的模型。所以,
之前解决数学文字问题的工作实际上可以分为序列到序列和序列到树模型。
82
1
ACL
2,023
offline
28
en-zh
So our best performing variant is Roberta-DeductiveReasoner.
我们最好的方法是机器人演绎推理,
我们表现最好的变体是Roberta-DeductiveReasoner。
56
1
ACL
2,023
offline
17
en-zh
Therefore, to train the classifier, that's why we reassigned surveys to each input commit message using the first ten characters of each commit message.
因此,为了训练分类器,我们使用每个提交消息的前10个字符为每个输入提交消息分配伪标签。
所以,为了训练分类器,这就是为什么我们使用每条提交信息的前十个字符,来对每个输入的提交信息重新分配调查。
56
1
ACL
2,023
offline
34
en-zh
In this example, the dataset is university dataset.
在这个例子中,数据集是大学数据集,
在这个例子中,数据集是大学数据集。
98
1
ACL
2,023
offline
25
en-zh
This can be regarded as a summarization task.
这可以被看作是一个总结任务。
这可以看作是一项总结 任务。
80
1
ACL
2,023
offline
28
en-zh
Legal issues are an integral part of many people's lives.
法律问题是很多人生活中不可或缺的一部分,
法律问题是许多人生活中不可或缺的一部分。
96
1
ACL
2,023
offline
11
en-zh
And we can easily incorporate some prior knowledge as constraint which can help improve the performance.
我们可以很容易地将一些先前的知识作为限制,这可以帮助提高性能。
我们可以简单地将一些先前的知识作为约束,这样就可以帮助提高表现。
94
1
ACL
2,023
offline
11
en-zh
So, ah and then at this third step we actually get the quotient.
然后在第三步,我们实际上得到了方程
然后在第三步,我们实际上得到了商。
90
1
ACL
2,023
offline
17
en-zh
The article are also very diverse as they come from thirty two different Belgian codes that cover a large number of legal topics.
文章的内容也很多元化, 因为文章来自32个不同的保加利亚密码, 涵盖了大量的法律话题,
这些法律条款也非常多样化,因为它们来自32个不同的比利时法典,涵盖了大量 数量的法律主题。
50
1
ACL
2,023
offline
5
en-zh
But if the dataset has five classes, FeSTE generates five new features.
但如果数据集有五类,那么FeSTE就会产生五个新的特征,
但如果数据集有五个类,FeSTE就会生成五个新的特征。
95
1
ACL
2,023
offline
11
en-zh
The BLEU is penalized when the system outputs a short sentence.
当系统输出一个短句子时,蓝色被惩罚。这种惩罚导致
当系统输出一个短句时,BLEU会受到惩罚。
59
1
ACL
2,023
offline
34
en-zh
And as we said, we are interested in assessing what capabilities the vision and language models have after pretraining.
正如我们所说,我们有兴趣评估视觉和语言模型在预先训练后有什么能力。
正如我们所说,我们有兴趣评估视觉和语言模型在预训练后具有哪些能力。
88
1
ACL
2,023
offline
11
en-zh
Hence, some question are better suited than others to the statutory article retrieval task, and the domain of the less suitable ones remains to be determined.
因此,有些问题比其他问题更适合——法定条款检索任务,而那些不适用的领域仍需被确定。
因此,有些问题比其他问题更适合于法定条款检索任务,而不太适合的问题所在的领域还有待确定。
93
1
ACL
2,023
offline
25
en-zh
Our paper solves these two problems and automatically generates high quality release notes.
我们的论文解决了这两个问题,自动生成了高质量的听众。
我们的论文解决了这两个问题,并自动生成了高质量的发行说明。
73
1
ACL
2,023
offline
33
en-zh
Instead, it's a structured collection of legal provisions that have a whole meaning only when considered in the overall context, that is, together with the supplementary information from the neighboring articles, the fields and subfields they belong to, and their place in the structure of the law.
相反,它是一个结构化的法律条款集合,只有在整体情况下考虑时才有完整的意义,这与他们邻近文章的补充信息一起,他们属于的领域和子领域,以及他们在法律结构中的位置。
相反,它是法律条款的结构化集合,只有在整体上下文中考虑时才具有完整的意义,也就是说,连同相邻条款的补充信息,它们所属的字段和子字段,以及它们在法律结构中的位置。
87
1
ACL
2,023
offline
11
en-zh
So, overall our reasoner is able to significantl significantly outperform this tree based model.
总而言之,我们的理疗师 能够从海浪中筛选出明显的海浪,
总的来说,我们的推理能够显著优于这个基于树的模型。
59
1
ACL
2,023
offline
5
en-zh
And the low score for this one?
这个问题的低分数是:
而这一个的低分呢?
89
1
ACL
2,023
offline
5
en-zh
So here planting another thirty five makes the model makes the model think it should be an addition operator.
在这里,再种 35 种,让模型认为它应该是一个额外的操作员,所以
在这里再植入一个35,会使得模型以为它是一个加法运算符。
55
1
ACL
2,023
offline
33
en-zh
The first is a system called ARENA released in twenty fourteen.
第一个是一个叫做Arena的系统,在2014年发布。
第一个是被称为ARENA的系统 ,发布于2014年。
80
1
ACL
2,023
offline
28
en-zh
With a limited applicability problem, we propose a high quality classwise summarization method using only commit messages as input.
对于一个有限的求解程序,我们提出了一个高质量的校验和总结方法,只使用委员会信息作为输入。
面对适用性有限的问题,我们提出了一种只使用提交信息作为输入的高质量的分类总结方法。
57
1
ACL
2,023
offline
25
en-zh
This proposed method can be used for all English repositories.
这个提议的方法可以用于所有英语书库。
这个提议的方法可以用于所有英语存储库。
100
1
ACL
2,023
offline
11
en-zh
Legal issues are an integral part of many people's lives.
法律问题是许多人生活中不可或缺的一部分,
法律问题是许多人生活中不可或缺的一部分。
99
1
ACL
2,023
offline
32
en-zh
Let's look at some characteristic of our dataset.
让我们来看看我们数据集的一些特征。
让我们来看看我们的数据集的一些特征。
92
1
ACL
2,023
offline
6
en-zh
For example, if this expression is not allowed, we can simply remove this expression in our search space.
例如,如果这个表情是被禁止的, 我们就可以在搜索空间中 简单地删除这个表情。
例如,如果此表达式不被允许,那么我们可以简单地在我们的搜索空间中删除此表达式。
90
1
ACL
2,023
offline
5
en-zh
CEAS and CAS achieved ROUGE-L scores more than ten points higher than the baselines.
GES 和 GES 的成绩比基数高出 10 分。
CEAS和CAS的ROUGE-L得分比基线高10分以上。
80
1
ACL
2,023
offline
33
en-zh
But do we know what the models have actually learned?
但我们知道模型到底学到了什么吗?当一个
但我们是否知道模型实际上学到了什么?
84
1
ACL
2,023
offline
28
en-zh
The article are also very diverse as they come from thirty two different Belgian codes that cover a large number of legal topics.
这篇文章也非常多样化,因为它们来自 32 种不同的比利时法规,这些法规包含了大量的非法话题。
这些法律条款也非常多样化,因为它们来自32个不同的比利时法典,涵盖了大量 数量的法律主题。
89
1
ACL
2,023
offline
33
en-zh
I present two related researches, but their problems are limited applicability and scarce data resources.
我提出了两个相关的研究,但存在适用性有限和数据资源稀缺的问题。
我提出了两个相关的研究,但它们的问题是适用性有限和数据资源稀缺。
87
1
ACL
2,023
offline
34