search

baoy-nlp / CNAT

Non-autoregressive Translation by Learning Target Categorical Codes
MIT License
10 stars 3 forks source link

invalid choice: 'nat' #1

Closed ustcdane closed 3 years ago

ustcdane commented 3 years ago

Hi, thanks for your great work! I tried run train.py, but, I get the following error: train.py: error: argument --task: invalid choice: 'nat' (choose from 'audio_pretraining', 'cross_lingual_lm', 'denoising', 'language_modeling', 'legacy_masked_lm', 'masked_lm', 'multilingual_denoising', 'multilingual_masked_lm', 'translation', 'multilingual_translation', 'semisupervised_translation', 'sentence_prediction', 'sentence_ranking', 'speech_to_text', 'translation_from_pretrained_bart', 'translation_from_pretrained_xlm', 'translation_lev', 'translation_multi_simple_epoch', 'dummy_lm', 'dummy_masked_lm', 'dummy_mt')

PyTorch == 1.8 fairseq==0.10.2

baoy-nlp commented 3 years ago

Can you share your fully training command here? I suspect that you may not specify the correct path for user-dir. If you run the ``python3 train.py --user-dir latent_nat ...'', you may correctly train the model.

ustcdane commented 3 years ago

Can you share your fully training command here? I suspect that you may not specify the correct path for user-dir. If you run the ``python3 train.py --user-dir latent_nat ...'', you may correctly train the model.

@baoy-nlp Thanks ,you are right, my user-dir path is wrong

ustcdane commented 3 years ago

hi, @baoy-nlp I get another error : ImportError: cannot import name 'flat' from 'latent_nat' , is glat input error to flat?

baoy-nlp commented 3 years ago

Please share your fully training command here without the detailed path? I can run it in my local environment for more details.

ustcdane commented 3 years ago

Please share your fully training command here without the detailed path? I can run it in my local environment for more details.

hi, now I can run without error when modify this line:https://github.com/baoy-nlp/CNAT/blob/7f270cff443db5dced048f7233f6a88b75075f0b/latent_nat/__init__.py#L4

from latent_nat import glat

baoy-nlp commented 3 years ago

Thanks for pointing out this. I have fixed this typo.