agaralabs / transformer-drg-style-transfer Goto Github PK

Hi, thanks for your great work! I was using delete and generative from the old paper, and want to have a test with your model. Unfortunately the link you provide for model trained on yelp seems to be expired. Do you have plans to maintain that link, or release the trained model somewhere else?

Hi, thanks a lot for sharing the code. I noticed that there is bleu package in the "eval_yelp.ipynb" file, however, I do not know where to find or install it. Could you tell us where it is from? Thanks again!

Thanks for sharing the code. Wondering where the bert model should be downloaded. I see there are two models in the google drive, but neither seems to be the correct format for the bert model (which requires .gzip file format)

Hello
I have a couple of questions regarding the paper.
① Firstly, the paper mentions that the fasttext model was used to calculate the accuracy. Can you please provide the parameter settings used for training fasttext on the five datasets (yelp, amazon, imagecaption, gender, and political)?
② Secondly, I noticed that the paper states that the size of the gender test is 534k, but the two xlsx files in the results/gender folder only add up to 1k data. I have the same issue with the political dataset. Could you please explain why this is so?
Thank you.

In the OpenAI_GPT_Pred what is the reference file reference_0.txt and reference_1.txt. Wondering if you could add instructions for obtaining them. Thanks so so much :)

Hi, I have already done the first 3 steps. When I in step 4, I should export DG_TRAIN_DATA and DG_EVAL_DATA. I can't understand it is sentiment_train_0.txt or sentiment_train_1.txt. Because in previous steps, it seems that the total train or eval file was not generated.
I'm not sure what I'm missing, I follow the steps. Please help me.

Hi，thank for sharing your code.
Can you tell me where is "train.csv, dev.csv" ，which for bert classifier.
Is it the data what I will use? Just put it together and name this?

Hey!
I was trying to run the OpenAI_GPT_Pred and I get this error at one of the steps:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-18-2cd4da19970d> in <module>
----> 1 op=preditction_with_beam_search("<POS> <CON_START> it is not terrible , but it is very good . <START>",4)
      2 op

<ipython-input-16-115b14581f7e> in preditction_with_beam_search(ref_text, beam_width, vocab_length)
     66             temp_lt=[0 for i in range(beam_width)]
     67             for i,x in enumerate(best_scoes_inx):
---> 68                 temp_lt[i] = beam_indexes[x] + [correct_inx[i]]
     69             # Update the Beam indexes
     70             beam_indexes = temp_lt

TypeError: list indices must be integers or slices, not float

I tried printing and x seems to be a float and not an int. I don't get why this is happening any clues?

Hi, Thank you for sharing your code.
I have difficulty running BERT_DATA_PREPARATION.

After running BERT_Classification_Training, I tried to run BERT_DATA_PREPARATION. However I don't have the bert_classifier_model which is in bert_classifier_model_dir so i got following error :

'Model name './bert_classifier/' was not found in model name list (bert-base-uncased, bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, bert-base-multilingual-cased, bert-base-chinese). We assumed './bert_classifier/' was a path or url but couldn't find any file associated to this path or url.'

I appreciate it If you could help me.
Thanks.

While running Delete_Retrieve_Generate_Data_Preparation I notice that there is no such file as sentiment_test.txt in the processed_files_with_bert_with_best_head folder and I'm not sure which step generates it as I tried searching and couldn't find anything. What could be going wrong?