Enhance TensorBoard test verification in test notebooks

[coderabbitai]

Problem

Currently, the test_tensorboard() method in various test notebooks (e.g., jupyter/rocm/tensorflow/ubi9-python-3.12/test/test_notebook.ipynb) trains a model with TensorBoard callbacks but does not verify that the logging functionality actually worked or that log files were created.

Background

This issue was identified during review of PR #1259 (#1259 (comment)) where the TensorBoard test lacks proper verification of its core functionality.

Acceptance Criteria

• Update test_tensorboard() methods across all test notebooks to verify log file creation
• Use temporary directories for log storage during testing to avoid filesystem pollution
• Add assertions to check that log files exist and contain expected content
• Ensure tests clean up temporary files after completion
• Verify the solution works across all affected notebook images (TensorFlow, PyTorch, etc.)

Implementation Guidance

1. Search for all affected files:

   find . -name "test_notebook.ipynb" -exec grep -l "test_tensorboard" {} \;

2. For each affected file, update the test method to:

   • Import os and tempfile modules
   • Use tempfile.TemporaryDirectory() for log storage
   • Add verification that log files are created
   • Include proper error messages in assertions

3. Example implementation pattern:

   def test_tensorboard(self):
       import os
       import tempfile
       
       # Create model and prepare data...
       
       with tempfile.TemporaryDirectory() as log_dir:
           tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=log_dir)
           # Train model...
           
           # Verify log files were created
           log_files = os.listdir(log_dir)
           self.assertTrue(len(log_files) > 0, "TensorBoard should create log files")

Related

• Original discussion: RHOAIENG-27434: Create Rocm Tensorflow Python 3.12 Image #1259 (comment)
• Requested by: @jiridanek