Matt Watson comments

Results 341 comments of


                                            Matt Watson

Porting Gemma 2 transformers checkpoint

Thanks!

Convert TF and Numpy ops in whisper_audio_convert.py to Keras Ops

@abheesht17 if you have time can you review this? You probably have the most insight here.

In colab, keras-nlp doesn't detect GPU anymore

Thanks for filing! I would guess the issue is that colab currently ships `tensorflow==2.14` and `pip install keras-nlp` will by default pull in the latest stable `tensorflow-text` and `tensorflow` (2.15),...

Add CLIP tokenizer to Keras NLP

@divyashreepathihalli can you detail more the exact differences that you needed to cover in KerasCV? I'm not sure this needs to be a whole new class, maybe just some new...

BertPreprocessor raises AttributeError with keras.Input

Do you have a colab repro available? That would help us dig in.

Would Keras support unique function?

I believe @hertschuh is working on a unique op. Though he is out for two weeks, so we can expect more action on this when he is back!

Fix BytePair special tokens tokenization

Thanks very much @abuelnasr0! Finally freeing up from our Gemma release. I'll try to review #1447, #1445 and #1397 as a set, but just a heads up I'll probably post...

Add qwen2 support

This is open for contributions if anyone would like to take this up!

Fix incorrect reference to model in Transformer class

Can you updated the generated files? markdown and ipynb?

Allow per-variable optimizer, add DispatchOptimizer.

> It's the layer that knows that certain variables that it owns need special handling. In this case, the layer knows that these large embedding tables can't use a traditional...