Inspecting a Dataset for Bias using TensorFlow Data Validation and Facets Overview Reviews

Prakhar K. · Reviewed over 1 year ago

Julianna G. · Reviewed over 1 year ago

Chaitanya P. · Reviewed over 1 year ago

Augusto E. · Reviewed over 1 year ago

Augusto E. · Reviewed over 1 year ago

Khoa V. · Reviewed over 1 year ago

Marcin B. · Reviewed over 1 year ago

Thejavathu C. · Reviewed over 1 year ago

Mai M. · Reviewed over 1 year ago

Lavanya C. · Reviewed over 1 year ago

---

Geetha S. · Reviewed over 1 year ago

Jupyter notebooks pip install package too slow?

Huy M. · Reviewed over 1 year ago

not user friendly, needs more elaborate instructions on Task 4

Hemali P. · Reviewed over 1 year ago

Lucas B. · Reviewed over 1 year ago

Maria L. · Reviewed over 1 year ago

Pranusha K. · Reviewed over 1 year ago

MADHURANK H. · Reviewed over 1 year ago

Chaitali S. · Reviewed over 1 year ago

good one

Rakesh S. · Reviewed over 1 year ago

Jake H. · Reviewed over 1 year ago

Nelton D. · Reviewed over 1 year ago

Duong Q. · Reviewed over 1 year ago

Swathi D. · Reviewed over 1 year ago

Ravindra S. · Reviewed over 1 year ago

TL;DR 1. Unable to run module because dependancies failed to install correctly. 2. User is not provided with the URLs for the preprocessed training and validation datasets in exercise "TODO 1". 3. User is expected to be familiar with the tf.keras utility functions and syntax. DETAILS: 1. When installing the dependancies, after installing the Fairness indicators and restarting the kernel, I received the following error: --------------------------------------------------------------------------- ModuleNotFoundError Traceback (most recent call last) Cell In[1], line 13 11 import tensorflow_hub as hub 12 import tensorflow as tf ---> 13 import tensorflow_model_analysis as tfma 14 import tensorflow_data_validation as tfdv 15 from tensorflow_model_analysis.addons.fairness.post_export_metrics import fairness_indicators ModuleNotFoundError: No module named 'tensorflow_model_analysis' #END OF ERROR MESSAGE This effectively prevented me from continuing this lab. 2. For the "TODO 1" exercise the location of the preprocessed training and validation datasets was not called out anywhere in the Jupyter Notebook or elsewhere in the documentation. I was forced to inspect the "bias_tfdv_facets.ipynb" file to find the file location URLs. Due to the aforementioned "ModuleNotFoundError", I was unable to complete the "TODO 2" exercise and received the following error: --------------------------------------------------------------------------- NameError Traceback (most recent call last) Cell In[3], line 4 1 # TODO 2 2 3 # The computation of statistics using TFDV. The returned value is a DatasetFeatureStatisticsList protocol buffer. ----> 4 stats = tfdv.generate_statistics_from_tfrecord(data_location=train_tf_file) 6 # TODO 2a 7 8 # A visualization of the statistics using Facets Overview. 9 tfdv.visualize_statistics(stats) NameError: name 'tfdv' is not defined 3. In order to complete any of the "TODO" exercises without consulting the answer cheat sheet, the user is expected to be familiar with the tf.keras utility functions and syntax, which is an entire language in and of itself.

William V. · Reviewed over 1 year ago