search

zhongkaifu / RNNSharp

RNNSharp is a toolkit of deep recurrent neural network which is widely used for many different kinds of tasks, such as sequence labeling, sequence-to-sequence and so on. It's written by C# language and based on .NET framework 4.6 or above versions. RNNSharp supports many different types of networks, such as forward and bi-directional network, sequence-to-sequence network, and different types of layers, such as LSTM, Softmax, sampled Softmax and others.
BSD 3-Clause "New" or "Revised" License
285 stars 91 forks source link

Some kind of cold start #12

Open AxelUser opened 8 years ago

AxelUser commented 8 years ago

Hi! Thank you for your lib, it look's really cool. But i've got problems in understanding some concepts and approaches. For example what for is feature templating? How does all those POS parameters means (i dont get the meaning of those abbreviations) I want to test your tool with document classifications, what do i need to do in general? Or is there any tutorialt to start with? Sorry if my questions sounds really stupid

zhongkaifu commented 8 years ago

Hi AxeUser,

Feature templating is the set of feature definition for feature extraction from corpus. It has two dimensions (x, y). X is for the context of current token, and Y is for the different feature dimensions. According feature templates, RNNSharp is able to extract features automatically. For more detailed information, please access README.md file.

For POS tags meaning, it's related to specific tasks and how you define your POS tags. Here is an example: https://www.ling.upenn.edu/courses/Fall_2003/ling001/penn_treebank_pos.html

bzaar commented 8 years ago

Hi Zhong, I agree with AxelUser that a working example would make things a whole lot easier and bring your library more adopters. Would you mind adding an example corpus.txt file and a train-test.bat file containing the commands to train a model on the given corpus and produce the features for a test.txt file? Maybe add all three files to a folder called Example so they are easy to find? Many thanks. Sergey

zhongkaifu commented 8 years ago

Hi bzaar,

You could get demo and example there: https://github.com/zhongkaifu/RNNSharp/releases Sure, I will add examples along with code in next code updating.