Hi, I have a question about the language model. I mentioned that language model is given as a parameter to the training process.
Let say that after training we got more resources (texts) about the target language and computed new language model. Can we use this new language model directly for decoding or first we should retrain the system using the new language model ? How language model is used for training the system (creating the phrase table) ? I thought that it is used only for decoding. If someone can provide some reference it would be appreciated. thanks, Igor
_______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
