WARNING: Don't convert course from TRAX to TensorFlow versions

I updated the course to the TensorFlow version after completing weeks 1-3 in the TRAX version and achieving 100% on the respective assignments. I did finish week 4 with the TensorFlow version, again with 100% on the weeks assignment.

However, all my completed assignments from weeks 1-3 disappeared. But worse, the course only has 3 weeks in it not 4.

Sadly, Coursera chat couldn’t do anything but suggested that I complete the new versions of weeks previous weeks if I wanted.

So, be forewarned: you may loose a week (weeks?) of the course and any completed work will not be counted as such.

This is likely my last post–take care…

Perhaps the three weeks of the TensorFlow course provides the same (or more?) benefits as the four-week version with Trax.

1 Like

Perhaps, but I think the compression from 4 weeks to 3 weeks was a mangled and unintentional conversion of the course from TRAX to Tensorflow as the programming tool.

The weeks of the original course are named:

  1. Neural Networks for Sentiment Analysis
  2. Recurrent Neural Networks for Language Modeling
  3. LSTMs and Named Entity Recognition
  4. Siamese Networks

The likely unintended version of the course after converting to TensorFlow is:

  1. Recurrent Neural Networks for Language Modeling
  2. LSTMs and Named Entity Recognition
  3. Siamese Networks

The omitted 1st week content in the "new’ course, Neural Networks for Sentiment Analysis (review), helped bridge the ideas completed in early NLP specialization courses. I’m lucky to have had other experiences with TensorFlow (TF) because the new course version had no course lecture/background on TensorFlow even though a number of the new assignments contained TF features that were challenging–with the lectures still referring to TRAX and the assignment using TF, this course was the least coherent, understandable of all the deeplearning.ai courses I’ve complete here which include: all the Deep Learning Specialization courses, all the Generative Adversarial Learning (PyTorch), and all the TensorFlow Developer Specialization courses (I’ve taken random other courses here and there).

I’m guessing that a week will be added back into the course to fix the mess that was created.

Good luck!

1 Like