• Do the baseline experiment but the performance is poor, so I may conduct it wrong
• Help Lantian for ICASSP with codes

• Getting familiar with ASR engine and prepare for the document

• Try different settings for rare word baseline experiment but still got wrong performance
• Read the paper "Pointing the unknown words"

• Read ASR code and write the document

• Read the paper "Neural Machine Translation By Jointly Learning"

• Read the paper "Memory Networks" and finish the document

• Review the two papers I've read last week and ask questions
• Read the paper "Recurrent Neural Network Grammars"

• install scrapy & beautifulSoup
• learn scrapy

• continue preparing for the paper sharing
• set up env for MemN2N code lua version(for language model)

• Review the paper "Recurrent Neural Network Grammars" and ask questions
• Try to run the code of RNNG

• tried to finish running the source code of MemN2N but it crashed
• tried to install torch on the server without root but failed
• will finish running the source code at home

• finish the ppt for paper sharing and pre-share it with Leader Feng

• try using scrapy download 'data learn' and 'blog' from sina(115MB)
• realize renew automatically everyday

• help Lantian, Wangyang and Zhiyuan to check papers

• try using scrapy download 'Wall news' & 'zqsb'(400M)

• reviewed paper that is to be shared tomorrow
• tried to set the env for a c++ code but failed

• review the non-parametric model written by Maoning Wang and learn about Latex
• help Wangyang check his paper's language problems

• help Wang Yang and Tang Zhiyuan to check papers
• try to understand and run the code of RNNG

• continue download news(2G)

• set the env for rnng code
• shared a paper

• share a paper
• keep understanding what Maoning Wang has written about Chapter 8

• Try to compile the rnng code, ask Zhiyong for help

• modify the spider

• continue work on the env setting of rnng, had small progress, aborted it
• find several open sources of NTM
• read the source code of MemN2N in lua(torch)

• Compile the code successfully
• Try to run the code

• spider QA(failed in zhihu)

• read the code of MemN2N in lua(torch)

• Run the code sucessfully using a fragment of data
• Try to format the whole dataset

• sprder QA

• format the full version of data
• debug the discriminative model code

• program script to test the chatting model
• spider QA from tianya (unsolved question)
• spider QA from zhihu.sogou.com(still run, but slow)
• filter the deteset for word vector

• managed to understand the source code

• run discriminative model, find there is one step missing, and ask author for missing code
• run generative model successfully, but need to run again on newly formated data
• reformat the data, stemming the nonterminal tokens

• run the two models of rnng successfully

• continued working on the source code
• ran memn2n several times, got a better result(closer to that in the paper)

• get the perfromance results of rnng, which agree with the original paper
• try to calculate the F1 score of generative model

• learnt torch

• get the mid states of discriminative model for memory network use

• finish the writing of chapter8 and update it to cvs

}

Monthly Summary

Time Off Table

Past progress

nlp-progress 2016/08

nlp-progress 2016/05-07

nlp-progress 2016/04