trax

trax copied to clipboard

Inconsistency in function's doc-string

1

### Description Inconsistent function description at https://github.com/google/trax/blob/master/trax/layers/attention.py#L330C1-L342C23 The function states that it "Returns attention-computed per-head activations and unchanged mask." but returns only the activations.

jaindhairyahere

Replace references to deprecated device_buffer attributes `jax.Array.device_buffer` and `jax.Array.device_buffers` will be deprecated as of jax version 0.4.22; see https://github.com/google/jax/pull/18844.

copybara-service[bot]

Refactoring to 1.5.1

1

The code, originally in version 1.4.1, has been upgraded to version 1.5.1. The main changes include: - Compatibility with Tensorflow 2.13.0 - Compatibility with jax==0.4.20 - Update of dependencies -...

mmarcinmichal

Error loading loop from a checkpoint

1

### Description I'm getting an error while trying to load a training loop from a checkpoint ``` File ~/srrf/srrf.py:104, in make_trainer(model, train_data, test_data, loss, metrics, name, optimizer, schedule, steps_per_cp, eval_batch_size)...

nwbvt

AttributeError: 'function' object has no attribute 'n_steps_per_checkpoint' for NLP Machine translation model

1

### Description Facing error when trying to run training.Loop ### Environment information ``` OS: Ubuntu 22.04 $ pip freeze | grep trax trax==1.4.1 $ pip freeze | grep tensor #...

Sumit1673

And it is when I load data from custom simple txt file

mihalt

Just want to let you know about [this issue](https://github.com/tensorflow/tensorflow/issues/61781#issuecomment-1712041209) in tensorflow

mihalt

Update core.py: fix docstring "y = Wx + b" -> "y = xW + b"

1

Matrix multiplication is not commutative so the documentation might be misleading.

arvyzukai

Can I do simple tokenization like [this ](https://github.com/https-deeplearning-ai/tensorflow-1-public/blob/main/C3/W1/ungraded_labs/C3_W1_Lab_3_sarcasm.ipynb) with trax?

mihalt

0o001