Qinlong Wang
Qinlong Wang
I execute ``./configure.sh` with "Y, Y, n" and then fail to run `bazel build build_pip_pkg` after ` using the image `tensorflow/tensorflow:custom-op-ubuntu16`. ``` ERROR: /temp/custom-op/tensorflow_zero_out/BUILD:10:1: //tensorflow_zero_out:python/ops/_zero_out_ops.so depends on @local_config_tf//:libtensorflow_framework in repository...
We only need those proto files to build elasticdl.proto, so we can only download those files by `wget`. Because TensorFlow repo is huge which makes the image huge. https://github.com/sql-machine-learning/elasticdl/blob/5fac16d29854001f8f68d4a0d514f4f57b95b8eb/elasticdl/docker/Dockerfile#L57-L60
If we want to show a demo using SQLFLow to launch ElasticDL jobs, it is better to use tables in MySQL. But, ElasticDL can only read records from MaxCompute tables...
The job in the "tests" stage in `travis.yml` contains many sub-jobs. Those jobs can run in parallel to speed up the stage. For example, the job to execute `unittest.sh` is...
It is boring for users to develop feature transformation using Keras layers or feature columns when the features count is very large. We can provide a configuration class for uses...
The column schemas in a MaxCompute table may be different, such as int, float, and string. The output tensor dtype of dataset should be the same as the column schemas....
If there is an error in `record_generator_with_retry`, the process to read records cannot stop and always recall `record_generator_with_retry`. I think the reason is that the exception in the sub-process to...
Now, we have developed `SparseEmbedding` layer to support sparse inputs. It is better to implement an embedding layer which can be fed with dense or sparse inputs. And it also...
In TF 2.1.0, `tf.keras.layers.Embedding` has supported inputs with `RaggedTensor`. https://github.com/tensorflow/tensorflow/blob/e5bf8de410005de06a7ff5393fafdf832ef1d4ad/tensorflow/python/ops/embedding_ops.py#L294 And using embedding with `RaggedTensor`, we don't need `SparseEmbedding` in ElasticDL. ```python import tensorflow as tf rag_tensor = tf.ragged.constant([[1,2],[5]]) ragged_embedding...