François Chollet
François Chollet
Thanks for the PR. Unit tests are failing and the value-add of most of these changes is highly questionable (most of it is unreachable code which will only serve to...
With Keras==3.0.0 you do not need to have TensorFlow installed (unless for specific features like tf.data support).
So you've verified the `StochasticDepth` layer works fine in JAX otherwise? Typically the "tracer leak" type of issue happens when you take a tensor from an intermediate computation and store...
Hi, thanks for the suggestion. Yes, I think this can be a useful layer. Where you would we put it? in `layers/preprocessing`?
> ModuleNotFoundError: No module named 'torch._dynamo'` This module is only present in the latest PyTorch version, I believe. @kiukchung @haifeng-jin could we make sure to avoid importing `_dynamo` if the...
> We'd still want to decide up to which torch version we want to support since it'll be good to run keras unittests with the major versions of pytorch we...
I assume it can -- maybe you can try it and let us know how it goes.
Indeed. This is the consequence of many people working on the translation at once. Unification of terminology would be great to have!
Thanks for the suggestion. We should add it in Keras 3. Rather than a new standalone layer, an alternative option is to add a `fill_mode` argument in `keras.ops.image.PadImages`. We use...
> Up. I'd be happy to push a PR and fix some code if someone would direct me towards the right direction. I would recommend simply adding a `convert_to_tensor` call...