-
I looked at the benchmarks, obviously for RNNs.
Currently the speed for the default transformer on 1 GPU GTX 1080ti (close to Titan X, right ?)
is at 5500 words per second. My only change was -w 9…
-
I download the code, go to tutorials/rnn/translate, run the command:
```
python translate.py --self_test
```
but something wrong happened:
```
Self-test for neural translation model.
Traceback …
-
We tried running Marian decoder for bi-deep models trained with Nematus with following command:
```
./../../../marian-dev/build/marian-decoder \
--type nematus \
--models $preprocess/mode…
-
Published: https://doi.org/10.1186/s13059-017-1189-z
Preprint: https://dx.doi.org/10.1101/055715
-
Hi guys,
I'm just trying to running example from here:
https://tensorflow.github.io/tensor2tensor/cloud_mlengine.html
```
# Note that both the data dir and output dir have to be on GCS
DATA_D…
-
Hello,
I am pretty new to using Keras and it has been quite nice to use.
I tried to find a solution to my issue here but couldn't.
Some issues were about handling different length of timesteps in t…
-
Hello,
I’m trying to build a seq2seq Model as in [this paper.](https://arxiv.org/pdf/1406.1078.pdf) I want to use a bidirectional GRU as Encoder and an unidirectional GRU as Decoder. The decoder re…
-
While developing a Hierarchical Attention Network, we have discovered that changing the batch size of the input effects the output of dynamic RNNs (while keeping everything else constant). In other wo…
-
The image_ocr code contains the typical mistake of trying to construct a bidirectional RNN network manually and failing to notice that the go_backwards sequence output is unfortunately reversed. Simp…
pasky updated
7 years ago
-
Here I open a new issue, I'm sorry for my misoperation. @emjotde.
Is the method to use 'marian' different between the Repo 'marian' and 'marian-dev' ? I got the error as follow:
```
boost::filesys…