Issue: Difficulty in Extracting mTet V1.0 and V2.0 Datasets

[TranNhiem]

Hi VietAI team,

I want to express my gratitude for publishing the 2nd version of the mTet dataset, V1.0 and V2.0. I followed the provided instructions for downloading the dataset from the shared file. However, I encountered an issue when using the source code provided on GitHub to extract the dataset into the tfrecord format.

Steps Taken:

• Original command used for generating tfrecords from raw text files:

python t2t_datagen.py
--data_dir=$path_to_folder_contains_vocab_file
--tmp_dir=$path_to_folder_that_contains_training_data
--problem=$problem

Environment Setting:

• Tensorflow version: 1.15
• Tensor2tensor version: 1.13.3

Issue Details:
The output of the provided code only downloads the EnviIwslt32k dataset and converts it to a readable format and tfrecord files. However, the two data files from the mTet dataset (highlighted in the attached image) are not being extracted.

Authors have any suggestions or updates on how to successfully extract these datasets?. Thank you so much for your assistance and your valuable contributions.