140721 - Xing Chao

From cslt Wiki
Jump to: navigation, search

Last Week:

1. Tested vbgmm(Variational Inference for the Gaussian Mixture Model) model when alpha = 10.

2. Finish training translation model today by using Chinese and English corpus V 1.0.

3. Finish preprocess the corpus by Xinhua and Bible.

This Week:

1. Finish training translation model by using Chinese to English corpus V 2.0 when use new corpus.

2. I have interested in whether the Single Word has higher performance on translation model and If we training corpus as Single Words to English words shell we get another highly performance language model, or translation model?