François Chollet comments

Results 342 comments of


                                            François Chollet

Tensorflow is still imported when using jax or torch backend

I agree that we need to remove the TF required dependency. There are several reasons why this is not possible right now: - We use `tf.nest` for processing nested Python...

Add new strings to process docstrings

I'd rather we standardize the docstrings on either `Examples:` or `Example:` or `Usage:`.

Add new strings to process docstrings

I would recommend converting all the titles containing "usage" to just "Example usage:"

Correct LoRA weights merging

@mattdangerw Can you take a look?

Update Models API for Keras CV

I'm seeing the following error on CI: > 6.609 AttributeError: module 'keras_cv.models' has no attribute 'SegFormerB0' What is the issue?

Changes related to TF backend compatibility check for Vision Transformer on small datasets

Can you add the md and ipynb files to the PR?

Changes related to TF backend compatibility check for Vision Transformer on small datasets

> kernal stopped after 24 hours in my macbook automatically at 46th epoch out of 50 epochs You're going to need a GPU VM (like a V100), running on CPU...

Migrating supervised contrastive learning example to Keras 3

I think you can just omit `y_true.set_shape([None, None])`.

Non accesable example TFrecords on Keras IO Example tfrecord.ipynb

@mjang2000 Amy, could you please take a look at this, if you have the bandwidth? It seems we should take the same approach as for the pneumonia dataset files.

Value error fixed in keras-io/guides/keras_nlp /transformer_pretraining.py documentation

I cannot reproduce the issue with the latest version of keras-nlp (0.7.0). What version are you using?