Python Context Free Grammar和PCFG生成基准测试？
[英]Python Context Free Grammar and PCFG generation benchmarks?
问题描述
I know there are various functions to use to general CFGs and PCFGs in Python; 
我知道在Python中有一些用于通用CFG和PCFG的函数; however they all seem to differ in speed. 然而它们似乎都有不同的速度。
Eg: NLTK, PyParsing. 例如：NLTK，PyParsing。
Are there any recent benchmarks comparing various attributes related to speed and memory usage? 是否有最近的基准测试比较与速度和内存使用相关的各种属性？
1 个解决方案
解决方案1
0 2013-06-14 11:06:10
If you're looking at official benchmarking of NLP tools like other computing software, you may be deeply discouraged. 如果你正在寻找像其他计算软件一样的NLP工具的官方基准测试，你可能会深感沮丧。 Sadly the NLP researchers are trying to push the accuracy system above making it realtime. 可悲的是，NLP的研究人员正在努力推动精确度系统的实现。 (It's surely nicer to say that I score X% more than the state-of-art rather than I save Y hrs / days training my model in research). （当然， I score X% more than the state-of-art而不是I save Y hrs / days training my model在研究中I save Y hrs / days training my model ，这肯定更好。）
Often they have like 1 sentence in their research paper that says how long it takes to train their system eg In average, the sampling program run on the Wikipedia dump consumed 20G memory, and each round took about one week on a single AMD Dual-Core 1000MHZ processor. 通常他们在他们的研究论文中有一句话说明训练他们的系统需要多长时间。例如， In average, the sampling program run on the Wikipedia dump consumed 20G memory, and each round took about one week on a single AMD Dual-Core 1000MHZ processor. from www.aclweb.org/anthology/P10-1116.pdf 来自www.aclweb.org/anthology/P10-1116.pdf
Anyway, since you want some benchmarks, so here's some of the homework that you can do with some googling =) www.aclweb.org/anthology/I11-1100 . 无论如何，既然你想要一些基准，所以这里有一些你可以用google搜索做的功课=） www.aclweb.org/anthology/I11-1100 。 But once again you realized that they are benchmarking accuracy not speed =) 但是你又一次意识到他们是基准测试精度而不是速度=）
声明:本站的技术帖子网页，遵循CC BY-SA 4.0协议，如果您需要转载，请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.