Milan Straka

TensorBoard callback does not collect data per batch (accidentally removed)

21

**System information**. - Have I written custom code (as opposed to using a stock example script provided in Keras): **No** - OS Platform and Distribution (e.g., Linux Ubuntu 16.04): **Linux...

[Feature Request] Add cuDNN-accelerated LSTM and GRU to PyTorch

6

Hi, are there any plans to add cuDNN-accelerated versions of LSTM and GRU to the PyTorch backend? Without cuDNN acceleration, the LSTM and GRU are considerably (several times) slower, even...

[Bug] With TensorFlow backend, using PyTorch DataLoader with different per-batch size does not work.

2

Hi all, consider a model obtaining on input sequences of dynamic sizes (sentences padded to the longest one). Here is a simple example: ```python inputs = keras.Input([None, 2]) hidden =...

[Feature request] When using PyTorch backend, use torch.data.tensorboard in keras.callbacks.TensorBoard

2

Hi all, even with PyTorch backend, `keras.callbacks.TensorBoard` uses `tensorflow.summary` to generate the Tensorboard logs, which can be inconvenient when only PyTorch is installed. However, PyTorch has a module `torch.data.tensorboard` capable...

Create browser plugins for Korektor service.

2

Create browser plugins for Korektor service. Current plan is to have nearly same plugins for Firefox and Chrome. The plugin add context menu items which appear when clicking on `input`,...

Change internal representation to UTF-8

2

Currently, we are using UCS-2 as internal encoding, which disallows us to use Unicode characters outside of BMP. We should change the internal representation, the current plans is to use...

If we sort out #10, that will leave us with the following complexity of language model lookups - 38.3/(100-47.6) = 73.1% ot time spend in language model lookup in `diacritics_h2mor.conf`...

Currently the similar word searching takes quite a lot of time. Considering `korektor-czech-130202` with `viterbi_beam_size=15` and `viterbi_stage_pruning=2.5`, similar words searching takes: - 47.6% with `diacritics_h2mor.conf` (for comparison, LM lookup takes...

TensorBoard callback does not collect data per batch (accidentally removed)

22

**System information**. - Have I written custom code (as opposed to using a stock example script provided in Keras): **No** - OS Platform and Distribution (e.g., Linux Ubuntu 16.04): **Linux...

tf.keras.losses on RaggedTensors crash during gradient computation on a GPU

8

**System information**. - Have I written custom code (as opposed to using a stock example script provided in Keras): **yes** - OS Platform and Distribution (e.g., Linux Ubuntu 16.04): **Linux...