Khánh đẹp trai
Khánh đẹp trai
Hi SongJinXue, can you share the real-time code? I would be so appreciated it. Many thanks for considering my request.
No, you just have to prepare the English dataset
I can see that your dataset is relatively small, so the number of update steps per epoch is only 5. Have your try a longer run and check if the...
I have not tried on other language datasets yet. Can you share more information about your dataset, config, tensorboard,…
i can see that your model did not converge yet, train loss is still high. Try increase the lr higher for faster training
Ping me at mail [email protected] for better debugging since I rarely check the GitHub notifications
I will take a look at my codes and run some experiments on english datasets and response to you soon @Shaobo-Z
@robin1001 @xingchensong @Mddct Hi there, all checks are now passing, review appreciated.
I added the implementation that enables joint training for both full and limited contexts. The configuration remains unchanged, and the attention decoder works as expected for chunk context inferencing. Encoder:...