Open lyzhang1999 opened 1 month ago
+1
Hey @lyzhang1999 - the model is currently only trained on English data, but the training recipes are language-agnostic. This means you can drop-in your own Chinese training data and reproduce the model in your language. Feel free to try it out using the training recipes linked in the repo!
It should also be possible to start from the English-only checkpoint, and fine-tune a Chinese model, again using the training recipes.
Does it support Chinese?