Files
b17e0d7380 Initial inference graph API implementation (#1910)
* Add inference graph API

Signed-off-by: Dan Sun <[email protected]>

* Implement inference graph API

Signed-off-by: Dan Sun <[email protected]>

* add validation and splitter

Signed-off-by: Dan Sun <[email protected]>

* Fix inference graph status

Signed-off-by: Dan Sun <[email protected]>

* Refactor knative reconciler

Signed-off-by: Dan Sun <[email protected]>

* Resolve service urls

Signed-off-by: Dan Sun <[email protected]>

* Improve logging

Signed-off-by: Dan Sun <[email protected]>

* add switch and root node configuration.

Signed-off-by: Dan Sun <[email protected]>

* Update docs/samples/graph/README.md

Signed-off-by: Dan Sun <[email protected]>

* Add inference router dockerfile

Signed-off-by: Dan Sun <[email protected]>

* Fix comments and error handling

Signed-off-by: Dan Sun <[email protected]>

* Updating single routerType to sequence for more flexisble composition

Signed-off-by: Dan Sun <[email protected]>

* Some InferenceGraph code streamlining

- Avoid unnecessary goroutine use
- Avoid most unnecessary unmarshal+remarshals
- Deduplicate logic between router types
- Handle async errors in ensemble case
- Don't require stepName to be set
- Return http error response when applicable
- Simplify pickupRoute func
- Fix incorrect log parameter syntax
- Add webhook validation for unique step names and inference targets

Signed-off-by: Nick Hill <[email protected]>
Signed-off-by: Dan Sun <[email protected]>

* Fix inference graph spec

Signed-off-by: Dan Sun <[email protected]>

* Add condition tests

Signed-off-by: Dan Sun <[email protected]>

* Update graph doc

Signed-off-by: Dan Sun <[email protected]>

* rename service to serviceName

Signed-off-by: Dan Sun <[email protected]>

Co-authored-by: iamlovingit <[email protected]>
Co-authored-by: Nick Hill <[email protected]>
2022-06-03 16:42:30 -07:00
..