14-10-19 Dongxu Zhang
来自cslt Wiki
Last week
1. Train LSTM-Rnn LM with 200MB corpus(vocabulary 10k, classes 100). when using 2 kernels, it takes aroung 200min per epoch. 2. Train 5-gram LM using Baiduzhidao_corpus(~30GB after preprocess) with new lexicon. There is a mistake when counted possiblity after merge. 3. An idea occured to me which may improve word2vec with much more semantic information. But there is huge computation complexity problem that bothers me, which I wish we can discuss. 4. Read paper "Learning Long-Term Dependencies with Gradient Descent is Difficult". Still in progress.
This week
1. Test LSTM-Rnn LM. 2. Finished building lexion. 3. Understand the paper. 4. May have time to achieve my baseline idea on text8.
