Updated model and checkpoints to run in tf v2.

[brunomcebola]

Changes:

1. Folder Structure:

   • Moved the existing codebase to a new folder named tf_v1.
   • Created a new folder named tf_v2 to accommodate the updated version.

2. Migration:

   • Migrated the model and associated checkpoints from TensorFlow v1 to TensorFlow v2 within the tf_v2 folder.

3. Environment Setup:

   • Added requirements.txt and .python-version files to both the tf_v1 and tf_v2 folders to ensure proper environment setup and compatibility with respective TensorFlow versions.

[brunomcebola]

@diegolascasas

[joaoluiscarreira]

Thanks! Do we need to have the data folder copied into both tf1 and tf2 folders or would it be better to keep it shared ? Did you test both versions ?

[brunomcebola]

Regarding the data folder, the actual data (.txt, .npy and .gif) could be shared accross both versions. The checkpoints, however, are version specific and should be left inside the respective folder for organization sake.

Nonetheless, since the access to this data is performed by setting the constants at the beginning of the scripts (making it possible for this folder to anywhere on the user machine), I choose to keep it separate just be more user friendly for anyone wanting to use the code.

Regarding the test of both versions, yes, I tested them and both worked in my machine with GPU support. An important note however are the subversions of tf, which need to be chosen according to the cuda versions, but this does not afect the final code of each version at all.

Hope this has been clear!

[joaoluiscarreira]

thanks for the changes!