88bca8223d
set dataset_location as the default location. ( #1399 )
2019-05-31 19:58:19 -07:00
d626d06ec2
Do not download results to local ( #1062 )
2019-03-29 23:51:08 -07:00
3b949b37aa
Fix a missing precheck ( #1028 )
2019-03-23 00:41:05 -07:00
5ccbbb9a0c
Fix hard-coded model export dir name ( #1015 )
...
* FIx hard-coded model export dir name
* Prefix is not the full gcs path.
2019-03-21 14:11:38 -07:00
e09a8ffb0e
Improve dataflow job to use GCS to keep temp job states. ( #985 )
...
* Improve dataflow job to use GCS to keep temp job states.
* make default location code reusable.
2019-03-19 18:08:48 -07:00
54925c7c00
Add job_dir.txt as output files of CMLE trainning job. ( #986 )
2019-03-19 18:08:31 -07:00
82c7c17ad8
Ignore not found error in delete cluster ( #991 )
2019-03-19 16:44:22 -07:00
70ee7482b1
Add documentation and sample notebook for dataproc components. ( #971 )
2019-03-14 17:44:58 -07:00
7aa581eba9
Fix cancellation bug for dataproc job. ( #967 )
2019-03-14 15:23:31 -07:00
f7faf3eea9
Add missing main_class arg to spark job and set default values. ( #966 )
...
* Add missing main_class arg to spark job and set default values.
* Adding missing main_class parameter passing.
2019-03-13 18:36:24 -07:00
2fc51bd3a8
Set default values for payload objects in case of empty. ( #962 )
2019-03-13 10:29:38 -07:00
6e24762ed2
DataProc commands implementation ( #862 )
...
* dataproc implementation
* Add documentation to dataproc commands.
* Add documentation for dataproc commands.
2019-03-08 16:06:54 -08:00
b94c731603
Fix bugs in bigquery component ( #921 )
...
* Update bigquery component yaml with latest interface.
* fix a typo in bigquery component.
* Add logging to bigquery component
2019-03-05 19:56:03 -08:00
2b07bb1832
Add bigquery docstring and dump output path. ( #885 )
...
* Add bigquery docstring and dump output path.
* Auto create dataset if it's not exist and dump results in local files
* make dataset location configurable
* Add todo to make kfp output path configurable.
* Fix comment
2019-03-05 16:00:20 -08:00
4e936f3ecd
Add online prediction instructions and print friendly msg in console. ( #914 )
2019-03-05 15:17:41 -08:00
639dec2402
Support customized export TF model folder names. ( #903 )
...
* Support customized export TF model folder names.
* Fix pr comments.
2019-03-05 14:18:01 -08:00
806e123bb5
Change viewer to markdown instead of html according to FE changes. ( #895 )
2019-03-04 18:36:13 -08:00
b00081d01f
Dump job id and change output to /tmp/kfp/output ( #878 )
...
* Dump job id and change output to /tmp/kfp/output
* fix test error
* FIx bug in validate_only mode
2019-03-01 18:40:30 -08:00
17bb778ef9
Refactor MLEngine code and add deploy and set_default commands ( #864 )
...
* Refactor MLEngine code and add deploy and set_default commands
* Rename to short name to avoid confusion.
* Add comment to version names
* Rename short name to id
2019-02-28 16:46:06 -08:00
a791197baa
Fix bugs in dataflow pipeline ( #870 )
2019-02-27 21:32:49 -08:00
ea26a574bf
Dataflow SDK to support launch beam python code or template ( #833 )
...
* initial files for dataflow commands
* dataflow commands
* add launch python command
* Support launch_python command
* Use display API and default outputs to /tmp/output folder
2019-02-26 16:18:43 -08:00
c4baa8a0c6
bigquery command ( #836 )
2019-02-22 15:43:48 -08:00
e282bd38b2
add container image uris support in train API ( #837 )
2019-02-20 17:32:11 -08:00
bd1e770a5b
MLEngine Commands Implementation ( #773 )
...
* MLEngine API commands
* Add job_id_prefix to create_job command
* add create training job command
* add display API
* Refactor CMLE APIs
* Fix PR comments.
* Add warnings before overwrite an existing file.
2019-02-16 12:47:44 -08:00
4e8b8a324f
Add Launcher module in component SDK ( #769 )
...
* Add launcher module
* Add argparse to wrap file launcher to avoid dup output prints.
* Fix PR comments
* fix test issue
2019-02-11 12:38:17 -08:00
cea1632479
Refactor component SDK with context manager instead of base class. ( #762 )
2019-02-05 11:10:07 -08:00
752256c24d
Create kfp componet sdk ( #729 )
...
* Create kfp componet sdk
* Add doc comments and rename pod name envir
* Add copyright header and use module init to expose public interface
* move @patch to class to avoid redundant code
* add k8s client
* remote stage states and tmp location.
* return value from execute to make test easier.
2019-01-28 22:06:40 -08:00
hongye-sun