Should i need to create lm.binary and trie with my own data

Hi,

I understand how to create a lm.binary and trie files as per the documentation.
Now, i am training with pre-trained models 0.5.1.
So, Here I need to create a new lm.binary file and trie file with the transcript in csv file?

wav_filename,wav_filesize,transcript
common_voice_en_18885793.wav,97580,she attended the chapin school
common_voice_en_18885788.wav,142124,today their programs are available on the internet

Or is it enough created from the document which is from here --> https://github.com/mozilla/DeepSpeech/tree/master/data/lm.