Closed danpovey closed 7 years ago
this is an issue in train_lm.py. For a small data set I asked someone to test the --warm-start-ratio to 1, and the script did not work. Can someone please implement something whereby if that's set to 1, the warm-start optimization is skipped?
Dan
Forgot that I had already implemented this but not pushed it... closing.
this is an issue in train_lm.py. For a small data set I asked someone to test the --warm-start-ratio to 1, and the script did not work. Can someone please implement something whereby if that's set to 1, the warm-start optimization is skipped?
Dan