Pham Van Ngoan
Pham Van Ngoan
Yes, you can. I trained with my own language (~1B words). However you have to adjust batch size from 4096 to ~512 with other configs is set default. The running...
Yes, I use Sentencepiece to build vocab and use dup_factor = 10 is ok.
@peregilk I really run with batch size 512. @stefan-it I run with python module ```python import sentencepiece as spm spm.SentencePieceTrainer.Train('--input= --vocab_size=30000 --model_prefix=prefix_name --pad_id=0 --unk_id=1 --pad_piece= --unk_piece= --bos_id=-1 --eos_id=-1 --control_symbols=[CLS],[SEP],[MASK], --user_defined_symbols=(,),",-,.,–,£,€')...
I see the README on TFHub have not updated yet, these parameters are as same as V1. As the author mentioned on README on this project (github), they use no...
@joongkyunKim @tienthegainz Google drive link: https://drive.google.com/open?id=1F_du_iarTepKKYgXpk_cJNGRb34rlJ5c
Hi @kasakh, I have the same problem with the weights do not produce any output. Have you found any solutions? please share with me.
Sorry for the late response and I hope you found the solution. More examples, you can find at official ALBERT repository: https://github.com/google-research/ALBERT
Sorry for late response! I don't maintain this work for a long time, I think you could send an email to Zalo for require the correct result.
I am also experiencing this issue with other STT services, including self-hosted models. I tried warming up these services with initial requests, but it doesn't seem to work. The initial...
What LLM model do you use? and what prompt you use to create the community report?